CEO Power and Firm Strategy: Evidence from the Netherlands

(1)

CEO Power and Firm Strategy:

Evidence from the Netherlands

August 2010

Tiberiu Rahau Supervisor:

S1797123 Dr. Padma Rao-Sahib

(2)

Abstract

(3)

1. Introduction

Ever since the popular work of Berle and Means from 1932, the activity of CEOs – as part of Corporate Governance – has become a so-called mainstream subject in economic literature. It remained so until recently and, during the past three decades, CEO activity has truly moved into the spotlight of public interest. After many influential papers such as Murphy (1985), Jensen (1986), Lambert and Larcker (1987), the economic literature experienced a boom in terms of the number of studies and articles related to this topic. Soon enough, at the end of the 1990s, the literature on CEO activity and especially CEO compensation has gone through another period of extra attention, fuelled by the time’s large economic scandals from all around the world, such as Enron or Parmalat, both entailing huge implications for corporate ownership, CEO compensation, and generally, for corporate governance as a whole.

Obviously enough, CEOs are greatly responsible for the good performance of their companies, and as such, they should (and do) wield significant power over their company’s future actions; but for a good measure of exactly how powerful the CEOs are, and exactly how tight the relationship is between their power and their companies’ strategies, one must look deeper into the structures of corporate governance, and relate them to the other decision-making factors at the top level. Despite the large amount of literature that has been written previously on the subject of CEO power, only few studies have actually measured CEO power relative to the rest of the executives, which is surprising, given that this would give a realistic indication about the CEOs’ influence. A further improvement in analysing the CEO’s relative power, in an even more accurate fashion, would be to also include in the analysis the influences of all the other stakeholders, but, to our knowledge, this is something that has not been done yet.

(5)

A very capable and revealing measure of CEO power – although not the most widespread in common use – is, up to this date, the CEO centrality index of Bebchuk, Cremers and Peyer, from their 2007 paper “CEO Centrality”. To the present day (2010), this is still the only paper employing this concept and an appropriate methodology to evaluate it, which is rather surprising, given its usefulness in terms of depth and relevance for any analyst using it. However, this is still a relatively new paper, so it is likely that the concept shall start to be referred to, and used accordingly in the coming years. In their study, Bebchuk, Cremers and Peyer use the term “CEO centrality” to refer to the relative importance of the CEO in the team of top executives (top five, to be more precise), in terms of power, contribution or ability. As a proxy for CEO centrality, they use the CEO’s pay slice (CPS), defined as the percentage of the total top-five executives’ compensation which is awarded to the CEO. This was also our choice of an indicator, but due to data availability we restricted the calculations only to the top four executives instead of five. The main advantage of the centrality index over other, alternative indicators of CEO power is that the power of the CEO is expressed as a ratio, and is relative to the power of the rest of the top management team of the same company, thus providing a time-accurate evolution of the power ratio within the firm, and excluding any outside influences (such as differences in firms size, or differences in the compensation of different executives) that would bias the comparison among centrality indexes of CEOs of different companies.

The study of Bebchuk et al. (2007) focuses on a panel data of companies from the United States (US), and so do other works concerning CEO power, such as Morck, Shleifer and Vishny (1989) or Chhaochharia and Grinstein (2007 and 2009) as well as other papers concerning, more generally, merger and acquisition activity observed on samples of companies from the United States, United Kingdom, or Japan. Since most of these studies focus only on large and heavily-investigated markets, there remains a gap, which our study will try to fill: apart from the emerging markets, there are many developed, mature markets which, due to various circumstances, such as less territorial expansion or less influence in the current economic setting, have not been among the favourites of the researchers this far. Such is the case with the Dutch market, a developed economy with an advanced corporate governance system, but with smaller economic impact on the world economy, when compared to the American, British and Japanese economies. Also, the relatively restricted number of Dutch multinational companies, compared to the number of companies from the other countries might have been a reason that contributed to its omission in the studies this far.

(6)

or lesser extent, internationally. Our choice of measuring CEO power by means of CEO centrality levels was based on the data that was available to us, and on the fact that this is an indicator scarcely used in the economic literature. Despite calculating the level of CEO power based entirely on their pay, CEO centrality is an accurate enough method of estimating power levels, as it is normal for CEOs to be remunerated accordingly, proportional to their responsibility and influence – and thus, according to their power. Moreover, studying the Dutch market from this perspective by using the indicator of CEO centrality constitutes a novelty, since there is no previous research done with the same methods on the same market.

There is a high variation in levels of CEO power (calculated by CEO centrality or CPS) among the companies in our sample, both among individual companies as well as between industrial sectors. There could be several reasons why CEO power levels differ so widely across our sample, but most likely this is a result of the former low transparency of executive pay arrangements in the Dutch economy, a problem that has been solved only recently, with the introduction of new legislation demanding disclosure of compensation (Van der Laan, 2009). In some cases, the CEO retains a great deal of power over the other members of the board. This is a very delicate issue, and it should be regarded in relation to the size of the company and the board. It is worth mentioning here that for any given company there is a certain level of the CEO’s pay slice (as defined by Bebchuk et al., 2007) which would be optimal for that company’s size. This is why, when observing CPS, studies such as the one cited above have split this indicator in two parts: normal, optimal CPS and excessive CPS. The “optimal” CPS could be defined as the appropriate level of relative power that a CEO has, given the size of the company, the pool of candidates for the CEO position, and the Supervisory Board’s supervision power over the CEO. Simply put, the optimal level of CPS exists when there is a fair balance between CEOs’ effort and their remuneration, so as to ensure the shareholders’ long-term goals, to satisfy the CEOs’ goals without incentivising them to make value-decreasing acquisitions, and generally to maximise the joint utility of all stakeholders. However, as the goal of our investigation is not related to executive incentivising, we do not make use of this measure in our study.

Since higher executive pay (salary/stocks/options) tends to show a relatively higher importance of the CEO in the team of top executives, it is safe to use CPS as a proxy for centrality. Furthermore, calculating such a centrality index based on compensation figures from executives which are all employed with the same company, controls for any cross-company specific characteristics that would influence the average level of compensation.

(7)

research questions refer to transaction-specific data. First, we investigate whether the levels of CEO power are in some way related with the likelihood of the company to engage in mergers and or acquisitions (M&A activity), based on data over five years. Moreover, we also investigate the amount of the companies’ M&A activity and the diversity of their M&A activity, relating both of them to CEO power. In this case, we define “diversity” as an M&A deal done with a company from an industry other than the one in which the other company is present in. The difference in industries has been chosen based on the Standard Industrial Classification codes (SIC codes) at the two-digit level. This shall be discussed further on. Thus, the first three research questions are, respectively:

Q1: CEO power affects the likelihood of a company to grow through M&A activity: companies with high CEO power levels are more likely to have M&A activity;

Q2: CEO power affects the amount of M&A activity done by a company: companies with high levels of CEO power tend to have a more intense M&A activity, expressed in the number of M&A deals per year;

Q3: CEO power affects M&A diversity: companies with high levels of CEO power tend to have a more diversified portfolio of M&A activity;

We find a strong negative association between CEO power and M&A activity. When analysing the yearly data, it appears that companies with low levels of CEO power are the ones which tend to have mergers and/or acquisitions. For the second research question we discover that companies with high levels of CEO power are rather associated with a low intensity of M&A activity, expressed in number of finalised M&A deals per year. For the third research question, we find out that companies with high levels of CEO power also are associated with a low diversity of M&A activity. We further try to provide some explanations, since some of these results are unexpected. However, no identical previous research has been done, therefore our expectations were based largely on extrapolations from existing theoretical studies and economic theory and, as a consequence, it should not be surprising if the results do not confirm such expectations. Furthermore, we find generally positive associations between CEO power levels and firm value, firm performance and firm size.

(8)

the difference in the layout and sorting of the data rather than from the theoretical construct behind them, as it will be discussed in the methodology section of this paper. We first analyse whether companies with high levels of CEO power also have a larger fraction of international M&A deals in their portfolio. Finally, we investigate whether companies with powerful CEOs attempt M&A deals with higher value. As the causality relationship in this instance might not be clear on first sight, precautions have been taken in order to control for endogeneity, and an instrumental variable was used. Thus, the rest of the research questions that we propose this far are:

Q4: CEO power affects the international orientation regarding M&A activity: companies with high levels of CEO power attempt a higher share of international M&A deals; and Q5: CEO power affects the size of the contracted M&A deals: companies with high levels of CEO power attempt larger M&A deals.

Our results show that companies with powerful CEOs do tend to attempt and finalise a higher share of international deals. Finally, our results have also found a positive association with the value of the M&A deals, showing that companies with powerful CEOs do tend to engage in larger deals.

After analysing the relationships above with CEO power expressed as overall compensation, we set off to re-do all the analyses but with CEO power expressed through disaggregated measures of pay: fixed pay, stocks or options awarded, and bonuses for executives. The results, although not always significant, confirm the relationships identified above with the shaping of the company’s growth strategy. We also attempted to replicate the previous study of Bebchuk et al. (2007), but on our own sample of companies this was possible only with an adapted methodology, given the unavailability of some of the data employed in this study. The results are mentioned in a sub-section which is part of the discussion sub-section of this paper.

(9)

(10)

2. Literature Review

A large body of literature has been written in the past three decades on the subject of the compensation of chief executive officers (CEOs). Empirical studies, such as Murphy (1985), Jensen (1986), Lambert and Larcker (1987), Kostiuk (1990), Murphy (1999), Core, Guay and Larcker (2003) are only a few of the many influential studies published in the past three decades. The increasing number of large acquisitions has provided enough research material and reason for this ever-growing interest in the activity of CEOs and the reasoning behind their decisions. Among these studies, a favourite subject among researchers has been the relation between CEO compensation and various measures of firm size. The outcomes were mixed, ranging from strong positive connections (Hardford and Li, 2007) to partial correlations (Bliss and Rosen, 2001, found a correlation for companies in the banking sector; DeYoung, Evanoff and Molyneux, 2009), and some found no correlation whatsoever (Avery, Chevalier and Schaefer, 1998).

While the majority of these studies have focused on examining more or less the same phenomenon but applied on different samples, there are some authors who have developed their own measures and terms in which to observe the subjects in question. Those measures or concepts were designed to give a more detailed view on the subject at hand, with a prevalent focus on the desired objective. Such is the case of the “CEO centrality”, as it will be discussed later on in the current part of this paper.

2.1. Literature concerning CEO pay or CEO power

Recent works have investigated CEO compensation while focusing on various aspects. Frydman (2005) and Frydman and Saks (2007) have observed that the percentage of the top-five compensation awarded to the CEO has risen with time. The same trend is shown in Murphy and Zabojnik (2004). The findings of Core, Holthausen and Larcker (1999) also add that those CEOs who are also the chairperson of their company’s board are also awarded a higher compensation as well as have a stronger influence over selecting the members of the board.

(11)

Chhaochharia and Grinstein (2007, 2009) provide an interesting study (although only partly relevant for this research) on CEOs and board oversight, as well as on the dynamics behind this power ratio after the major financial scandals in 2001 and 2002.

While there is a wide range of articles on CEO power, the concept of CEO centrality has been developed mainly by Lucian A. Bebchuk, in order to define the “managerial power gap” between the CEO of a company and the company’s respective members of the executive team (the top management team). It uses the “CEO pay slice” (CPS) as a compound index, and focuses on the difference in levels of power expressed as difference in compensation between the CEO and the rest of the top managers. In Bebchuk’s studies involving CPS, this is defined as the percentage of the total compensation received by the firm’s top five executives, which is awarded to the CEO (Bebchuk, Cremers and Peyer, 2007).

Results of Bebchuk et al. (2007) include, among others, a negative relation between CEO pay slice and Tobin’s Q (industry-adjusted), and associations of CPS (as a measure of CEO centrality) with several dimensions of company performance or of company governance. The study of Bebchuk (2007) has found that firms with high CPS have comparatively lower rates of return on assets (ROA), as well as lower industry-adjusted rates of operating income to assets. Similarly to Adams et al. (2005), firm-specific variability of stock returns is observed to be negatively related, but this time with CPS instead of just with the CEO pay. More powerful CEOs who yield more power over the boards often are rewarded with significantly higher bonuses, as argued by Grinstein and Hribar (2003).

2.2. Literature concerning company growth and performance

The subject of mergers and acquisitions has also been given particular attention recently, and especially in this past decade. Although very many papers have been written about this subject alone, as well as about the relationship of mergers and acquisitions with other business effects or company events, there are only a few works which have studied the relationship between mergers and acquisitions and the particularities of CEOs. The relation between firm value (as measured by Tobin’s Q) and governance arrangements (CEOs, boards, etc.) has been investigated by Bebchuk and Cohen (2005), finding out that Tobin’s Q is influenced negatively by staggered boards. Yermack (1996) shows that the presence of a large board of directors also tends to create a negative relationship with Tobin’s Q.

(12)

address this topic on an extensive basis. On a slightly different note, with regard to merger and acquisition diversification, studies have shown that CEO compensation has a tendency to be higher for diversified companies (Rose and Shepard, 1997, Finklestein and Hambrick, 1989). The present paper tries to tackle this subject as well, observing this, as well as the relationship between merger and acquisition diversity and CEO centrality.

2.3. Literature concerning CEO pay/power and M&A activity

The relation between company size and CEO pay has been present in studies dating back as far as Baumol (1959), but in recent studies as well, such as Jensen (1986), Lambert and Larcker (1987), Finklestein and Hambrick (1989), Morck, Shleifer and Vishny (1990), Murphy (1999), Core et al. (1999), Cyert, Kang and Kumar (2002), all confirming the general conclusion that managers might have high incentives to increase firm size beyond the optimal level, by making acquisitions that would decrease the firm’s value. A series of studies come to support this affirmation, among which Morck et al. (1990), Lang, Stulz and Walkling (1991), Qui (2004), Moeller, Schlingemann and Stulz (2004; 2005), Bargeron, Schlingemann, Stulz and Zutter (2007) and Masulis, Wang and Xie (2007).

The relationship between CEO relative power and merger and acquisition deal size has been studied also by Grinstein and Hribar (2003) and the results indicate that the higher the managerial power of the CEOs, the higher the likely size ratio of target to acquirer. This finding is also in line with the previous findings of Jensen (1986). Furthermore the same study finds that there is a positive correlation between managers with high levels of managerial power within the board and higher bonuses, as well as with engagement in larger M&A deals.

(13)

board, or when they are the only insider in the board, or when they have a “founder” status within the board. Some CEOs might choose to expand their firms by making a series of smaller acquisitions, but the most certain way of increasing company size generally seems to be by making large acquisitions, and so, also the compensation of the CEO seems to be largely dependent on these large acquisitions, as argued by Hardford and Li (2007). This might be a “chain reaction”, as a larger company size might entail the necessity to provide the CEO with larger and more powerful incentives (Baker and Hall, 2004; Anderson, Becher and Campbell, 2004; Rosen, 2004). Interestingly enough, Avery et al. (1998) discover no correlation between large acquisitions and subsequent CEO pay.

Concerning firm performance, Bebchuk et al. (2007) in their study on CEO centrality and firm value, argue that also CPS is negatively associated with Tobin’s Q. More generally, the influence of how the type and style of the CEO is affecting the firm’s outcomes has been studied in papers such as Malmendier and Tate (2005) and Bertrand and Schoar (2003). There is one study that examines the relationship between CEO pay and company performance in the Netherlands (Duffhues and Kabir, 2008), which finds a negative relationship between performance and cash compensation. Grinstein and Hribar (2003) did not find any relationship between bonus compensation and deal performance. This is somewhat in line with the findings of Jensen and Ruback (1983) and Moeller, Schlingemann and Stulz (2003) which confirm that there are no positive relationships between announcement returns and acquisitions or merger deals.

(14)

The available literature which focuses on any form of differentiation between the power ratios of CEOs to the team of top executives is scarce (mostly notable here being Bebchuk et al. 2007). Therefore, another contribution of this paper is that it adds to the literature on CEO centrality, by analysing a pool of data from Dutch companies’ CEO pay. As in the fore-mentioned studies, panel data is used along with cross-sectional data, containing information between the years 2002 and 2006.

There are several methods to evaluate the concentration of CEO power; some of them are related, while others are non-related. CEO centrality (as defined by CPS) is clearly a pay-related method, but it is not the only one. As another pay-pay-related measure of CEO power, we employ in this study the CEO pay gap, chosen as such in order to determine whether the results are consistent with both methodologies. The methods used to calculate it are multiple: sometimes it is merely the difference between two compensation levels and sometimes it is the difference between median values of two or more different compensation groups (Ensley, Pearson and Sardeshmukh, 2007; Lee, Lev and Yeo, 2005). But regardless of the methods’ slight variation, the results are generally in line with other studies examining CEO power.

Another method that has been used in the literature – albeit scarcely – is the pay dispersion and it expresses the coefficient of variation among salaries of the CEO and the salaries of the rest of the executive team. Alternatively, it can be expressed in terms of standard deviation, but, at least in theory, the results should be similar since the method itself is rather similar with the coefficient of variation method. Since the pay dispersion is showing only the variation between the CEO’s pay and the average pay of the top management team, we deem this measure to be not as precise as the pay gap or CEO centrality and therefore not as relevant to our study. For this reason, we do not use it in our analysis.

(15)

3. Hypotheses / Research Questions

We are basing our research on two major theories: the Managerial Power theory, which claims that CEOs can and do have the power to influence their own pay in spite of the supervisory board (or any other supervisory entity), and the Agency theory, which claims that managers (CEOs) have to be properly incentivised in order not to misdirect the actions of the company towards their own interest, disregarding the best interest of the shareholders and of other owners. These are the reasons why we expect our research questions to have some specific outcomes, as discussed in this section.

As mentioned before, the main research focus of our study is the dynamics of CEO power and the influence of its level on company growth, and additionally its influence on various indicators of company performance. More specifically, the relationship that we are investigating is the connection between how powerful CEOs are, and the extent to which their companies are expanding. This is a complex relationship, which results from the influence of several factors which may come from within as well as from outside the company. As external factors (such as global economic settings) are unlikely to be quantified appropriately, we turn our attention to factors closer to the companies, and to the results of interactions of these companies with other companies (targets). We are considering factors such as the number of M&A deals, diversity of M&A deals, percentage of international deals in total M&A deals, and deal size as items that would help us shape the relationship between company growth and CEO power. In order to explore this relationship on several directions, we have devised five research questions to aid us in our research, which range from general (the likelihood of having M&A activity) to very specific topics (degree of diversity or deal-level data).

(16)

companies with high levels of CEO power who are involved in M&A activity rather than companies with low levels of CEO power. Thus, our first research question is the following:

Q1: CEO power affects the likelihood of a company to grow through M&A activity: companies with high CEO power levels are more likely to have M&A activity;

Building upon the previous question, we try to observe not only the likelihood of having M&A activity, but also the quantitative difference among M&A deals and the relation with CEO power. Therefore we also hypothesise that since powerful CEOs are more incentivised to increase the size of their company, then those companies with high levels of CEO power would also be involved in a larger number of mergers and/or acquisitions, as this would be a certain strategy of company growth. The relationship that we are studying is the same as for the first research question, but here a quantitative dimension is added to our analysis: the number of M&A deals of a company in a year. The added value of this research comes from the fact that the analysis is based literally on the amount of deals performed by each company in each year. We present this as our second research question, although it could be seen as an alternative to our question Q1. Therefore, our second research question is:

Q2: CEO power affects the amount of M&A activity done by a company: companies with high levels of CEO power tend to have a more intense M&A activity, expressed in the number of M&A deals per year;

(17)

compensated for – in the long run – by a more stable position on the market or by other long-term advantages. As a simple example, a manufacturing company might receive only a reduced profit soon after acquiring a transport company, given the operational and organisation costs that may or may not be estimated correctly prior to acquisition, but in the long term there are clear advantages for the newly-diversified manufacturing company when it comes to shipping its own products by using its own transport services. As such, we investigate whether companies with powerful CEOs do engage in such diversified activities. Our third research question is as follows:

Q3: CEO power affects M&A diversity: companies with high levels of CEO power tend to have a more diversified portfolio of M&A activity;

Following the same logic as in the previous research question, it can be argued that another strategy to diversify and reduce risks is to merge with or acquire companies from other countries. While this can be true, a more important factor in choosing to have international M&A activity is the added benefit of operating in several markets. As economic theory suggests, once a company becomes a multinational enterprise, it can enjoy several opportunities offered by its presence in multiple markets, in the form of differences in prices, higher demand from certain markets, advantages for the supply chain, and so on. While this rationale is generally accepted to be true, we investigate whether this factor has an influence with powerful CEOs. Thus, we investigate the relationship between CEO power levels and the fraction of international M&A deals attempted (and completed) by companies, and our fourth research question is:

Q4: CEO power affects the international orientation regarding M&A activity: companies with high levels of CEO power attempt a higher share of international M&A deals;

(18)

therefore, we shall expect that companies with high levels of CEO power would also attempt M&A deals of higher value. Our fifth and final research question is:

(19)

4. Methodology

The methodology employed in this study is based for the most important part on the one used by Bebchuk (2007), namely using CEO centrality calculated as CEO pay slice (CPS) as a proxy for CEO power. CPS in this case is measured as the percentage of CEO compensation in total top executive team compensation, per company per year. The top executive team, in this situation, is considered to consist out of four persons: the CEO and the first three board members, other than the CEO, ordered by the value of their respective compensation. In a few cases in the examined sample, there are only three members, namely the CEO and two other top executives. The few companies that appear to be managed only by one manager and therefore have a CEO centrality index of 100% have been excluded from the analysis in order not to skew the results. There were eleven such companies in our sample, for which there was no data available regarding executives, other than the CEO.

In order to investigate the direction and effects of the relationship between CEO centrality and firm performance, we have developed a series of research questions relating to present and past data from the companies. However, the only study that this research can be related to in the following, is Bebchuk (2007) since it is the only one containing results on CEO centrality. Thus, our first investigation refers to whether companies with high levels of CEO centrality are more likely to make mergers and/or acquisitions.

In order to analyse whether this happens or not, the panel data is sorted by company and year, and a logit model is used since the dependent variable is nominal (a dummy variable) and there are several independent variables. The dependent variable is whether the company has had any M&A activity, and the main independent variable is CEO centrality. Industry dummies are added to control for those industries in which a large number of acquisitions and/or mergers have taken place in a certain year.

The outcome of this research question is expected to result in a positive relationship, based on the most part of previous research, as previous studies show that more powerful managers tend to engage more into M&As and possibly even empire-building. Therefore, we expect a positive correlation between the index of CEO centrality and the number of acquisitions.

(20)

companies with high levels of CEO power tend to have a more intense merger and acquisition activity, expressed in the number of M&A deals per year.

In the light of previous studies concerning the frequency of mergers and acquisitions, we expect the answer to show a positive relationship, and therefore that high levels of CEO power entail a high number of M&A deals per year. Once more, this idea is also supported by the fact that it has been previously shown in several instances that powerful CEOs are likely to engage in activities such as empire-building.

A count data model is used in this instance, with an associated negative binomial regression, since the values of the dependent variable are in the format of positive integers. The dependent variable is the number of mergers or acquisitions per year (ma_count), and the main independent variable being CEO centrality. Other variables used in the regression and chosen for their revealing of the firm’s size are the assets of the company, the stocks and the number of employees.

Our third research topic adds another dimension to the research, namely the diversification. Why would a company want to become active in several industries and thus become diversified? Because investing in diversified portfolios greatly reduces the company’s risks, as well as provide additional market segments or potential developments to its existing products or services.

In this context, the companies’ business scope is considered. The international standard industrial classification code (SIC code) was obtained for each of the companies, only considering their main scope of business, so that each company can appear in only one industry. The SIC code is composed of four numeric characters, the first one being the identifier of one of nine main areas of business. The following three characters are narrowing down the business area in which the respective company operates. Diversification, in this context, refers to the variability of the main business scope of the company; based on the SIC classification, we consider that a company which makes a merger or acquisition has a diversified M&A activity if the first two digits of their respective SIC codes are different. Obviously, diversification occurs even if only the first digit of two SIC codes is different, but the second digit is also encompassing enough to also make a difference regarding the diversification between companies sharing the first digit of the SIC code. Thus, the third area of investigation is focused on whether companies with high levels of CEO centrality tend to engage in more diverse M&A activity.

(21)

relation in this case as well. Since we start from the assumption that companies with central, more powerful CEOs are more prone to attempt a large number of mergers and/or acquisitions, it might be expected that many of the deals attempted by these companies are not limited to the same industry in which the CEO’s company activates.

The diversity index in this case is measured as the percentage of M&A deals made with relation to companies from other industries, in the total M&As that that company has done in a year. The dependent variable is the M&A diversity index (divma), and the main independent variable is the yearly CEO centrality (ceo_c_y).

Our fourth topic of research is relating to whether the M&A deals attempted and completed by companies are international or not. More specifically, whether there is a connection between the levels of CEO power and the fraction of international M&A deals attributable to the respective companies. Based on previous research, this research question is expected to show a positive relationship. Again, although we do not have the results of any research which would be identical to this, by extrapolating we can expect that companies with high levels of CEO power would also attempt a higher share of international M&A deals, since international mergers or acquisitions are not only more complex, but also more costly – as well as more rewarding.

The data for this analysis is sorted as a panel, since the focus lies in analysing the dynamics of the CEO pay over the 2002-2006 time periods, while considering the properties of the M&A deals. The data is sorted by company and year, and a new variable has been calculated, the proportion of international deals (fraction_of_int_deals), as a percentage of how many of a company’s deals in one year involved a company from a different country than the Netherlands. This variable is also the dependent variable, while the main independent variable is CEO centrality (ceo_c_y). The controls used here are the same as in the previous research questions, acquirer assets (assets), the acquiring company’s return on assets (roa), and the number of employees to control for firm size (emp).

(22)

(23)

5. Data and Summary Statistics

The data used in this study was collected from secondary sources, and the CEO compensation data was collected from the annual reports of the investigated companies. The rest of the data is available on online databases such as Zephyr. Four of the five research questions are tested across the same panel data set, but involving different variables. The only exception from the panel data format is the fourth research question, which uses yearly firm-level data but in combination with individual M&A deal data. For that reason, the regression made for the fifth research question uses a cross-section type data set. A description of each variable follows.

(24)

Table 1. CPS by industry (CEO Pay Slice index given for each of the six industries – wholesale trade and retail trade have been divided, given the large number of companies in each industry).

Overall Construction Manufacturing Transport

Trade wholesale

Trade

retail Financial Services Median 44.774 46.173 43.674 36.376 39.311 46.159 45.16 46.173 Mean 45.107 47.269 43.891 43.779 45.827 53.577 42.997 46.107 Std. Dev. 15.556 18.603 14.963 18.37 21.912 16.594 13.386 12.672

Minimum 8.621 8.621 8.621 9.652 15.875 27.003 17.439 14.463

Maximum 91.244 91.244 87.369 76.666 91.244 77.002 75.194 76.715

Merger and acquisition activity (ma) is selected as a dummy variable, indicating simply whether the company has made any mergers or acquisitions in the respective year. Its value is “1” if the company has merged with or acquired at least one other company, and its value is “0” if there has been no M&A activity in that year for that company. This variable is the main dependent variable in the first hypothesis, but it will also be used in the dataset for the second and third research questions.

To address the number of M&A deals that a company has made in a year, we have devised the variable, ma_count. It is linked to the variable ma in the sense that if one company did not have any M&A activity in one year, then that company will automatically have the value “0” assigned to its ma_count variable for that year. The introduction of this variable is the main determinant of the usage of a negative binomial regression for the second research question. Conversely, the variable takes the value “1” for one merger or acquisition during that year, value “2” for two finalised deals in a year, and so on. The summary statistics show that the M&A activity for the analysed Dutch companies ranges between zero and fifteen M&A deals per year, with the majority of companies signing between one and two M&A deals per year.

(25)

Based on the values of the dummy variable indicating the existence of M&A activity, and on the variable expressing the number of M&A deals completed by a company, we devise another discrete variable, fraction_of_int_deals, which indicates the proportion of international deals in total deals per year. This is expressed as a percentage, and is calculated as the number of deals that a company completed in a year with another company from a different country (than the Netherlands), divided by the total number of deals of that company in that year. Thus, if a company has had any “international” M&A deals in a year, then that company will get a corresponding percentage for this variable; otherwise, the fraction of international deals will be considered zero. The summary statistics show that 80.37% of all companies have had some form of international M&A activity anywhere between 2002 and 2006. While for any given M&A deal there is a unique correspondent target company, the acquiring company is sometimes shared by several deals (and target companies): there are 1478 target companies and 107 acquiring companies.

The main variable used for the last regression is dealvalue, showing the actual value of a deal. Deal value (variable dealvalue) data was not available for all deals, and as a result, the sample available for the associated regression is considerably smaller than the usual, with only 135 observations after adjustments. Deal value is expressed in hundreds of thousands of Euro.

Seven industry dummies are used in order to control for industry-wide effects which, if otherwise left uncontrolled for, would bias the results. The industry dummies follow the classification of the SIC codes, and have the following meanings: ind_constr for companies that deal mainly with construction activities; ind_manuf for companies whose main business scope is manufacture;

ind_transp for companies whose main business scope is transport; ind_trade_wh for companies

(26)

Table 2. Industry-level merger and acquisition statistics. Industry Dummy variable % in all industries Number of companies Number of M&As Relative index of M&A activity** Construction ind_constr 5.607 6 46 7.667 Manufacturing ind_manuf 42.056 45 276 6.133 Transport ind_transp 5.607 6 61 10.167 Trade ind_trade* 14.019 15 65 4.333 Financial ind_fin 13.084 14 82 5.857 Services ind_serv 19.626 21 186 8.857 100 107 716 43.014

* there are actually two "trade" dummies, one for wholesale and one for retail; in this table the aggregate trade industry volume is shown

** indicates which sector has experienced the highest proportion of M&As compared to the number of its companies

However, if we measure the number of mergers or acquisitions done by companies from a given industry, relative to the total number of companies from that industry, then the overall picture changes significantly: the transport industry seems to have the highest number of mergers or acquisitions relative to its sample size (more than ten times as many M&A deals as the number of companies that are active). The next industry in this classification is the services industry, with almost nine times more M&A deals than the number of companies; the manufacturing industry only ranks fourth, with little more than six times more mergers and acquisitions compared to the number of companies. The last in this classification is the trade industry, with an only four-times higher index.

(27)

As an indicator for the firm’s performance, return on assets (ROA; variable: roa) is defined as a ratio between the company’s net income over total assets. It can take values in the [-1; 1] range, and in the present study, its values lie between -0.8 and +0.5, with a concentration around the 0.1 and 0.2 levels, corresponding to a median ROA of a bit less than ten percent (Table 10). This data shows that many of the companies’ assets are much under-valued on the market.

Table 3: Examples of large players included in the study, separated by industry

Construction Manufacturing Transport Trade

wholesale Trade retail Financial Services

BAM Groep AKZO Nobel TNT Ahold Beter Bed

Holding

ABN

AMRO VNU

Heijmans Philips Vopak Sligro Food

Group Stern Groep Aegon Ordina Boskalis

Westminster Unilever HAS Beheer Buhrman Laurus Fortis Stork Ballast

Nedam Mittal Steel

Smit

Internationale Getronics Schuitema ING Groep KPN

(28)

Opposing the panel data format used previously for research questions Q1 – Q3, in the final research question, Q5 where we investigate the companies’ M&A activity more into detail, focusing on the M&A deals themselves, there are some particularities that should be mentioned. Because of the deal-level data being the main focus, the structure of the variables is not a panel data anymore, but a cross-sectional data. Therefore, for each deal the deal size is mentioned (in hundreds of thousands Euro) while the CEO compensation data is still a yearly value corresponding to the acquiring company in that year, and it is expressed in percentages. The same applies to other yearly variables, such as target-specific indicators (assets, turnover, EBITDA, etc.) and acquirer-specific indicators (assets, turnover, EBITDA, employees, stock price, etc.). Other variables which have suffered only minor or no alterations and therefore remain similar to the ones in the previous data set include ma, the industry dummies (ind_constr, ind_manuf,

ind_transp, ind_trade_wh, ind_trade_re, ind_fin, ind_serv), and stock1/0 (the dummy variable

(29)

6. Results and Analysis

6.1. CEO centrality and M&A activity

The dependent variable used for determining whether there is a connection between powerful CEOs and the M&A activity is the dummy variable indicating whether the company has had any M&A activity per year, while the main independent variable is the yearly CEO centrality index. The controls used for the regression are the company’s assets to control for firm value (assets), the return on assets, to control for firm performance (roa) and the number of employees (emp) to control for company size.

The results, shown in Table 11, indicate that high levels of CEO centrality are negatively correlated with the happening of M&A activity. The negative association is a rather unexpected result, but the significant coefficient although very low, proves that the relationship is correct. The coefficient is significant at the 5% level, and its value is -0.0079. It may also be possible that there are other, external conditions affecting the M&A strategy but whose effects cannot be quantified. All the other controls used have significant coefficients with a positive effect on the M&A activity: this is positively associated with the size of the company’s assets, the company’s return on assets and the company’s size, expressed in the number of employees.

6.2 CEO centrality and the number of M&As per year

Due to the specific format of the data (specifically, the dependent variable), in this case a count data model is used. Once more, the data is in a panel structure, sorted by firm and year. The dependent variable is the number of mergers or acquisitions that a company has had in any given year (ma_count) and the main independent variable is the yearly CEO centrality index over the investigated time frame (ceo_c_y). Controls used include the size of the company’s assets (assets), stocks value (stocks), number of employees (emp) to control for firm size and performance. Our previous expectations were contradicted again, since it appears that more powerful CEOs do not necessarily have a more intense M&A activity.

(30)

the obligation of the board to give all due consideration to all stakeholders’ interests in regard to the mergers or takeovers strategy. The enforced set of best practices is thus requiring a slightly more demanding outcome with regard to the results of any decision taken at the top level of the company. Thus, if the “joint utility” must be maximised for all the involved parties, then this would be a factor that could restrict the number of merger or acquisition deals a company performs, even with a highly-powerful CEO.

In our analysis, all coefficients are significant at the 5% level. All other controls used appear to have a significant positive correlation with the company’s M&A strategy; the results show that highly-valued firms do tend to have more mergers and/or acquisition in a year, as well as firms which perform better, as indicated by their stock prices. As expected, company size is also linked in a positive way to the number of mergers and/or acquisitions.

6.3 CEO centrality and M&A diversity

The M&A diversity, expressed in the number of M&A deals done involving companies from other business sectors, shows again a negative relation with CEO centrality levels. The main independent variable, CEO centrality (ceo_c_y), has a significant coefficient at the 5% level and its value is -0.029, indicating that CEO power is linked to a rather un-diversified portfolio of M&A activity. This could be so because of a(n undocumented) tendency of decision-makers to prefer investing in the same business sector in which they already have experience. Not only that, but it may be possible that even if choosing a diversified M&A deal, the returns might be lower than in the case of an un-diversified acquisition, possibly due to the lack of economies of scale or, more generally, reduced efficiency. There could be other reasons that escape our logic, but this is a topic that may well be a subject for a future research paper.

Regarding our analysis, it is worth noting that with the exception of one other control, relationships with all the other variables have been found to be insignificant. The results are available in Table 11. The control variable that seems to have an impact is firm size, in our case expressed in the number of employees. Thus, it would appear that larger firms tend to have more diverse M&A activity, the coefficient of this variable being small but positive.

6.4 CEO centrality and the fraction of international M&A deals

(31)

such as the number of employees (emp), acquirer’s assets (assets), and acquirer’s return on assets (roa).

The results, shown in Table 11, indicate that there is a positive association between high levels of international M&A activity and high levels of CEO centrality. The results are confirming the expectations. Although the model itself is significant, one of the control variables, emp (number of employees) does not have a sufficiently high significance for our dependent variable. All other coefficients are significant at the 5% level. While it seems that companies with high levels of CEO centrality and therefore more powerful CEOs tend to engage in a higher share of international deals, this seems to also happen more often in higher valued companies. Furthermore, there seems to be a positive trend among companies whose assets are over-valued to also have more international deals than those companies with under-valued assets.

6.5 CEO centrality and M&A deal size

As mentioned before, in order to address the issue of comparison between companies who have had an international M&A activity and to assess their association with CEO power (and CEO centrality), the data had to be re-organised, as the focus has moved on specific deal-level data, therefore losing the panel data format, and switching to a cross-section. Since the focus in this analysis is on the deals’ size, each deal was analysed in connection with its corresponding target company and its acquiring company. While for any given M&A deal there is a unique correspondent target company, the acquiring company is sometimes shared by several deals (and target companies): there are 1478 target companies and 107 acquiring companies. The data is sorted as a cross-section dataset, but this time the focus is on the value of the deals. The dependent variable in this case is the actual deal size expressed in hundreds of thousands of Euro (dealvalue), and the main independent variable is CEO centrality measured again as a yearly percentage value (ceo_c_y). The main controls used are the value of the two involved companies, acquirer and target (acquirorassets and targetassets) and acquiring company size, expressed in number of employees (emp).

(32)

6.6 Results for compensation expressed in disaggregated measures of pay

Since the structure of CEO compensation varies greatly from company to company, analysing CEO power in terms of some disaggregated compensation indicators might reveal additional and relevant details. For this reason, we set off to analyse the six relationships described above once more, but instead of overall CEO compensation, the indicators that we will use to determine CEO power will be CEO centrality in terms of CEO fixed pay, stocks and/or options awarded to the CEO, and CEO bonus pay – the components of overall CEO compensation.

Table 4. Compensation expressed in other measures

Fixed

pay Stocks/options

Bonus pay Number of CEOs who did not receive the specific

type of compensation, at least once 10 86 55

Total instances over the 5 years when the CEOs did

not receive the specific type of compensation 15 367 111

Total number of CEOs: 107

Total observations: 535

6.6.1 CEO centrality expressed as fixed pay

(33)

coefficients have been found to be non-significant (the coefficient for CEO centrality and target company’s assets value), most likely because of the greatly reduced sample available for analysis (73 observations for which all data, including fixed compensation was available).

The similarity of these results is not surprising given the fact that the ranking of most of the CEO compensation packages is more or less the same as the ranking of the fixed CEO compensation (i.e. CEOs that receive the highest compensation also receive the highest fixed compensation). Furthermore, the number of CEOs in the sample that do not receive any sort of fixed compensation are very few, therefore this variable is highly observable in the current sample. However, what is surprising is the opposite result obtained in the case of the relation between completion time and CEO centrality. This shall be discussed in the following section of this paper.

6.6.2 CEO centrality expressed as awarded stocks and/or options

CEO centrality in terms of stocks and options is calculated based on the yearly value of stocks or options awarded to the CEO as compensation. There are many companies that do not offer any such securities as part of their CEO compensation package, and therefore the sample available for analysis is smaller than the sample available previously (for overall compensation, and for compensation expressed in fixed pay).

Thus, we discover that, by referring to the stocks and options awarded to CEOs, companies with more powerful CEOs tend to engage in more M&A activity and they tend to be less diversified than the companies with lower levels of CEO power (Table 13). Surprisingly, the results for our first research question were inconclusive, although the results for the second question show a positive association between the two indicators. While the rest of these results are in line with the results of the previous classification, once more, a relationship between CEO power and deal size could not be established, given most likely by the reduced sample available for this analysis (78 observations).

6.6.3 CEO centrality expressed as bonuses

(34)

Thus, we find that most of the results are similar to the results for overall compensation; a high level of CEO centrality in terms of bonuses is negatively linked to the existence of M&A activity. However, in the long run, when examining the entire five year period, a positive association has been found between the company’s M&A strategy and CEO centrality expressed in bonuses. This seems to be in line with the studies that have pointed out the positive relationship between company growth and subsequent increases in CEO bonus compensation. We find as well that companies with high levels of CEO power are also likely to have a low level of diversity in their M&A activity. In terms of bonuses, CEO centrality is again associated with a high fraction of international deals and a long completion time.

In the studied sample it is easily noticeable that the three components of the CEO compensation package – fixed pay, stocks and/or options and bonuses – are usually complementary. For instance, if a CEO is given a large fixed pay, then it is possible that this CEO is receiving neither stocks nor options, and is probably receiving a small bonus compensation. Of course, the three components have many variations, but as an aggregated result they produce similar results to the overall compensation analysis, because they complement each other for the most part.

Table 5. Summary statistics for various types of compensation

Type of

compensation Median Std. Dev. Minimum Maximum Mean

Overall CPS 44.774 15.556 8.621 91.244 45.107

Fixed CPS 44.403 28.095 0 100 53.288

Stocks/options

CPS 0 26.884 0 100 15.743

Bonus CPS 44.693 34.935 0 100 48.036

Levels in the USA (Bebchuk): 1.8 34.4

6.7 Alternative measure of CEO power: Pay gap

(35)

the CEO centrality index with the pay gap indexes. A short review of the index will be given in the following sub-section, along with the data description and the results of the analysis.

As previously mentioned, the methods used to calculate a pay gap are multiple: sometimes it is merely the difference between two compensation levels and sometimes it is the difference between median values of two or more different compensation groups (Ensley, Pearson and Sardeshmukh, 2007; Lee, Lev and Yeo, 2005). But regardless of the methods’ slight variation, the results are generally in line with other studies examining CEO power.

Studies considering pay gap are more abundant compared to studies involving CEO centrality indexes. Pay gaps can be calculated for any pair of working groups, and so, a simple search on “pay gap” will most commonly reveal studies which observe wage differences between different groups, such as between female CEOs and male CEOs, or between top management and base workers salary. In line with this diversity, there are also multiple methodologies which calculate the pay gap index. The one that we decided to employ in our study is the one used by Henderson and Fredrickson (2001). Thus, we have defined the CEO pay gap per year as the natural logarithm of the amount equal to the difference between the CEO’s compensation and the average pay of the other, non-CEO members of the top management team, in the respective year. The use of natural logarithms was intended in order to reduce heteroskedasticity in the results of the regressions. The resulted index is applied to each company for each of the five years. Due to data availability (or lack thereof), in the top management team we have included only the three top managers instead of the usual four, and in some cases the average is calculated only on the basis of the first two top managers.

(36)

6.8 Replications of previous studies

Moreover, to observe whether CEO centrality can be used interchangeably in other studies, we have also tried to replicate the results of the first study dealing with this measure of CEO power, namely the CEO centrality study of Bebchuk et al. (2007).

Since the sample which was analysed in this study is a very specific sample, consisting of only Dutch companies, and since the number of companies is not as extensive as in other studies concerning American or British companies, we have tried to reproduce the study based on our sample by using a similar methodology as employed in Bebchuk et al. (2007), and approximating for or replacing some variables with others that were available in our dataset. Many variables had also to be created or calculated based on existing ones. It has to be noted that, because of the less-than-complete availability of certain data (such as board insider ratio, or depreciation), the original methodology employed in the study mentioned above could not be perfectly reproduced. Therefore the results, although sometimes favourable, have to be viewed as providing a less-than-complete accuracy.

The study of Bebchuk et al. (2007) focused on the relationship between CEO centrality and the value of companies. Its results show high CEO centrality to be negatively associated with, most importantly, high company value, lower performance (lower accounting profitability) and lower stock returns. In order to be able to apply the method on our sample, several variables had to be created. As a measure of firm value, Tobin’s Q (tobinsq) was introduced, measured as the market value of equity plus the book value of assets minus the book value of equity, all divided by the book value of assets. Log book value (logbva) represents the log of the book value of assets, and Capex/Assets (capexassets) is the ratio of capital expenditures to assets. As said, due to the unavailability of data several variables are missing, such as leverage, R&D activity, the presence of the CEO in the role of the chairman of the board, insider ownership, and entrenchment index, and therefore the model is not as significant to describe the current sample. The coefficients we obtained for the variables in use are similar (see Table 16), in the sense that they show the same kind of relationship, but unfortunately the main independent variable does not seem to be significant enough for describing the relationship with CEO centrality.

6.9 Results of robustness testing

(37)

Dutch companies (both targets and acquirers), and therefore might influence the companies in our sample as well. Furthermore, the introduction of the Euro currency in 2002 could have had a slowing-down impact on the economy as a whole. Also, the new corporate governance code introduced in late 2002 might have had an impact in making CEO earnings more transparent; therefore, for 2002 and 2003 the CEO compensation data could be less accurate than in the remaining years. In order to check the robustness of our results independently from the potential effects of such events, we test our results for robustness by excluding from our analysis of the panel data the years 2002 and 2003. At the other end of the analysed period, 2006 has brought a growth trend for the economy of the Netherlands, but since this occurred only towards the end of the year (according to CPB) we do not consider this as a significant impact on the 2006 year data for our panel dataset. Also, for the cross-section data, we decided to use three sub-samples of companies to test for robustness, based on company size. We chose this method only for the cross-section and not for the panel data as well, because doing so in the panel dataset would result in missing variables for several years, and thus we would have to choose between either dropping the panel format or removing so many companies from the sample to the point where the remaining companies would not be a significant sample anymore.

Table 6. Results of robustness testing for research questions Q1 and Q2; there are two pairs of columns, the first one in the pair showing the results for the full sample, and the second one showing the results for the reduced sample (without 2002 and 2003); where there is a important difference between the coefficients, the corresponding coefficients in both columns are coloured.

(1.1) (1.2) (2.1) (2.2)

VARIABLES M&A M&A M&A Count M&A Count

ceo_c_y -0.0079*** -0.0027*** -0.0048*** 0.0001** (0.0017) (0.0022) (0.0018) (0.0019) assets 0.0003** 0.0003** 0.0001 0.0001* (0.0001) (0.0003) (0.0001) (0.0001) roa 1.5409** 1.5853** (0.7766) (1.0563) stock 0.0141*** 0.0068** (0.0047) (0.0044) emp 0.0112*** 0.0195*** 0.0076*** 0.0073*** (0.0029) (0.0065) (0.0017) (0.0016) Observations 535 321 535 321 Number of firm_id 107 107 107 107

(38)

variation is not high and the direction of the relationship is the same. All other coefficients remain the same, at the same levels of significance, despite the reduced sample. With regard to the second research question linking CEO power levels with the amount of M&A activity of companies, there seems to be a discrepancy in the sense that in the reduced sample the coefficient for CEO centrality is no longer negative. Despite the very low value of this coefficient (0.0000125), it is positive nonetheless, thus meaning that the research question stating that companies with high power levels have a higher amount of M&A activity is true for the reduced sample (2004-2006). The coefficient for company stocks is also different, showing a smaller but still positive association with the number of M&As.

Table 7. Results of robustness testing for research questions Q3 and Q4; there are two pairs of columns, the first one in the pair showing the results for the full sample, and the second one showing the results for the reduced sample (without 2002 and 2003); where there is a important difference between the coefficients, the corresponding coefficients in both columns are coloured.

(3.1) (3.2) (4.1) (4.2)

VARIABLES Diversity Diversity International International

ceo_c_y -0.0296*** -0.0239*** 0.0022*** 0.2823** (0.0027) (0.0031) (0.0002) (0.0318) assets 0.0001 0.0001 0.0783*** 0.0882** (0.0001) (0.0001) (0.0279) (0.0001) roa -1.9108** -1.5847* 0.1521** 0.1234** (0.7911) (1.2401) (0.0643) (0.1220) emp 0.0067*** 0.0069** 0.0810* -0.0001* (0.0025) (0.0037) (0.0588) (0.0002) Observations 535 321 430 258 Number of firm_id 107 107 86 86

(39)

employees is also different and of opposing direction, but other than this, the results are consistent

with the full sample.

Table 8. Results of robustness testing for research question Q5; there are three additional columns, the first one in the pair showing the results for the full sample, and the second through fourth one showing the results for the reduced sub-samples (split accordingly by the 1 million Euro and 100 million Euro milestones); where there is a important difference between the coefficients, the corresponding coefficients in both columns are coloured.

(5.1) (5.2) (5.3) (5.4)

VARIABLES Deal size Deal size

<1 mil Deal size [1;100] mil Deal size >100 mil ceo_c_y 742.2996* 82.4137** 579.239*** 2805.420* (2236.718) (15.4609) (128.2837) (5278.963) emp 327.5181*** 6.9489** 11.2957 542.026*** (157.0344) (3.8291) (8.6628) (279.0164) targetassets -0.0012 0.0003** -0.0002 0.0738*** (0.0058) (0.0003) (0.0002) (0.0219) acquirorassets 0.0012*** -0.0002* 0.0001* -0.0011 (0.0005) (0.0002) (0.0001) (0.0011) Observations 135 37 49 49 Number of firm_id 88

(40)

7. Discussion and Limitations

7.1 Discussion

This study investigated a sample of Dutch companies regarding CEO power expressed as CEO centrality, and company growth strategy, and afterwards analysed the relationship between these two subjects. The lack of previous empirical work on this matter makes it difficult to compare the results of the present study, but by extrapolating from related theoretical studies we could define a set of expectations for what our results could look like. While some of our expectations have been met, there is also a part of the results which are new and rather unexpected.

(41)

(42)

Third, our results also shed light on the relationship between the CEO power levels and the diversity of mergers and/or acquisitions. Only the study of Rose and Shepard (1997) has tackled on the subject of M&A diversity and its relationship with CEO compensation. According to this study, the relationship is positive, but there is no identified direction of causality. Nevertheless, a causality relationship might as well be irrelevant for the current study, since either of the two indicators (a high level of CEO power expressed through CEO centrality and a high level of diversification) can influence the other and, in the end, the connection between the two is subjected to chain effects that grow exponentially, in which each of them contributes to the modification of the other. It may well be possible that companies with high levels of CEO power could incentivise the CEO to engage in non-risky M&A deals, and therefore to pursue their M&A strategy mostly within the same business sector in which they belong themselves. It could be, as well, that once a company engages in an M&A deal with a company from another business sector, the efficiency of closing the deal is lower, given the reduced expertise of the acquirer with the particularities of the other business sector, and as a result, the compensation awarded to the CEO post-acquisition might be less, compared to a potentially similar M&A deal with a company from the same business sector. Our expectations were that the CEO power would have a positive relation with the diversity of M&As, but our contradicting results could make sense as well in the context presented above. Furthermore, we have found a positive relation with company size, but this time we could find no relevant relationship between M&A diversity and company value or performance.

CEO Power and Firm Strategy: Evidence from the Netherlands