• No results found

Show me the money! “A comparative study on the influence of board structure on the relation between board control and CEO compensation in the Netherlands, UK and USA”

N/A
N/A
Protected

Academic year: 2021

Share "Show me the money! “A comparative study on the influence of board structure on the relation between board control and CEO compensation in the Netherlands, UK and USA”"

Copied!
53
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Show me the money!

“A comparative study on the influence of board structure on the relation between board

control and CEO compensation in the Netherlands, UK and USA”

University of Groningen

Faculty of Economics & Business

Msc International Business & Management

First supervisor: M.A.G. van Offenbeek

Second supervisor: B.J.W. Pennink

Master Thesis International Business & Management

by

(2)

ABSTRACT

This paper analyzes the potential influence board structure can have on the relation between

board control and CEO compensation. Results point out that an increase in board control goes

along with a decrease in CEO compensation. Furthermore, a relation between CEO

compensation and board structure is not existing. Also a relation between board control and

board structure is not recognized. However, board structure has an influence on the relation

between board control and CEO compensation, when viewing the variables of board control

separately.

(3)

TABLE OF CONTENTS

1. INTRODUCTION ... 3

2. LITERATURE REVIEW ... 5

2.1

Corporate Governance ... 5

2.2

Agency Theory ... 7

2.3

Institutional Context ... 11

2.4

Board structure ... 13

2.5

Hypotheses ... 16

3. METHODOLOGY ... 17

3.1

Sample and Data Gathering ... 17

3.2

Dependent variable ... 19

3.3

Independent variables ... 19

3.4

Control Variables ... 22

3.5

Data analysis ... 22

4. RESULTS ... 24

4.1

Inconsistencies data collection ... 24

4.2.1

Hypothesis 1 ... 25

4.2.2

Hypothesis 2 ... 27

4.2.3

Hypothesis 3 ... 29

4.2.4

Summary of the results ... 32

4.3

Moderator effect ... 34

5. DISCUSSION ... 38

5.1

Analysis of the results ... 38

5.2

Limitations ... 42

5.3

Future Implications ... 44

5.3.1

Academic Implications ... 44

5.3.2

Implications for management settings ... 45

6. CONCLUSION ... 46

(4)

1. INTRODUCTION

Executive compensation is a multifaceted and notorious subject. Academics, policymakers, economists and the media raise alerts to the high pay levels for executives since many years. Heated debates are rising about the quality and fairness of pay setting policies in different organizations and the compensation arrangements that it creates. It was a wave of corporate scandals that began in late 2001 and shook confidence in the performance of public organization boards and drew attention to potential flaws in their executive compensation practices. One can think of the financial scandal around Enron in 2001, Parmalat in 2003 with its accounting scandal and mutual fund fraud or Fannie Mae in 2004 with its widespread accounting errors. In the Netherlands, it were the events at Ahold in 2003 that caused an enormous shock.

As a result, executive compensation has emerged as a key corporate governance issue for which the regulations highlight how the board of directors should play a significant role in CEO compensation practices (Bebchuk and Fried, 2006; Albuquerque and Miao, 2006; Boyd, 1994; Matsumura and Shin, 2005; Magman et.al., 1999; Sanders and Carpenter, 1998; Stiles and Taylor, 2001: 4; Lippert and Moore, 1995).

Obviously, the expansion around executive compensation issues hasn’t been neglected by the academic world. Although a long history of research attempt concerning CEO compensation has been made, the event of several corporate scandals caused a tremendous input on executive compensation research to be performed. The controlling role of the board of directors became a clear focus of discussion. With agency theory serving as the most used foundation for explaining the present occurrence of excessive pay levels, research attempt has been made within different extremes in this context. (Conyon, 2006; Bebchuk and Fried, 2006; Hendry, 2005; Bebchuk et.al., 2002; Boyd, 1994; Stiles and Taylor, 2001: 12; Roberts et.al., 2005; Dalton and Dalton, 2005). On the one extreme, researchers such as Bebchuck and Fried (2006) write about the growing influence and power of the CEO to entrench themselves with excessive pay arrangements and board of directors having a minimal influence in this process. On the other side researchers such as Sanders and Carpenter (1998) claim a “traditional” optimal contracting approach in which the CEO bargains on arms length with the board of directors to be still prevalent and argue causes of high executive pay levels to be caused by other factors, such as e.g. the increasing globalizing world and the along coming task complexity of the CEO that needs to be compensated. For sure, researchers have a lot of different thoughts concerning the trend of growing executive pay levels and the role the board can play. Hence, although research attempt has been made for the role between the CEO and the board of directors concerning executive pay, inclusive evidence is found about how this exact role can be defined.

(5)

informs about the way the board of directors is structured. First of all, they serve as the bridge between the CEO and the stakeholders. Secondly they are the ones having a certain degree of control over CEO compensation. Therefore, it is interesting to study whether the way they are structured influences the assumptions made by researchers.

The influence of board structure fills an existing research gap by being a new variable in this research field. Along comes another appropriate gap. Since board structure is often country bounded, and limited research attempt concerning board control and CEO compensation has been made outside the context of the USA – which is characterized by a solely one-tier structure - the case of other countries and situations in general has been neglected. By including countries with different board structures, the context can be internationally expanded and the gap can be filled.

Next to the possible contribution to the academic knowledge, a contribution can be found within practice. The development of executive compensation as a key corporate governance issue also attracts the attention of several stakeholders, however they are mainly characterized as the shareholders. It is in their interest to know how the role of the board can be characterized and in which way CEO compensation arrangements are made in the best interest of the CEO or the shareholder. Gaining knowledge into this issue might provide the shareholder with useful insights and might eventually even lead the shareholder to take action when not pleased with the forthcoming results.

A last contribution can be found within corporate governance itself. By including board structure as a possible influencing variable of the relation between board control and the CEO concerning executive compensation, corporate governance practices become challenged. When differences between board structures are found concerning board control and CEO compensation, organizations itself may reconsider the choices they made in developing corporate governance, with a focus on board structure.

The central objective of this paper is to answer the following main research question: “How does board control affect CEO compensation and what is the influence of board structure on CEO compensation practices?”

(6)

2. LITERATURE

REVIEW

Since several years, a new trend evolved in examining the CEO compensation landscape. It was a wave of corporate scandals that began in late 2001 that caused the confidence in the management of public boards to be undermined and drew attention to dubious practices concerning CEO pay settlements. Consequently, several researchers have studied different explanations and determinants for CEO compensation practices, with the role of the board of directors and the supervisory board in these executive pay processes to be the focus of discussion. Their role has been highlighted from different point of views but all share the thought that corporate governance should form the basis of the potential position the board of directors and the supervisory board can play (Bebchuk and Fried, 2006; Albuquerque and Miao, 2006; Boyd, 1994; Matsumura and Shin, 2005; Magman et.al., 1999; Sanders and Carpenter, 1998; Stiles and Taylor, 2001: 4; Lippert and Moore, 1995; Aguilera, 2005).

2.1 Corporate Governance

Governance is known in his Latin roots as ‘gubernance’, which means to steer and a quotation in this context is ‘he that governs sits quietly at the stern and scarce is seen tot stir’. According to Cadbury (2002: 1) this indicates that governance need not and should not be heavy-handed. Nowadays, corporate governance became known as the solution and prevention method for corporate scandals. It concerns the structure of responsibilities and rights between the different groups with a stake in the company (Aguilera and Jackson, 2003). It is necessary to balance the shifting interests between the stakeholders and the active asymmetries in information accordingly (Stadler et.al., 2006; Davies, 2006: 8). Moreover, corporate governance describes the way in which ownership is organised and how company’s are directed and controlled. It is the board of directors or the supervisory board that decide on the point the organization takes within this construction. They are the mediator between the providers of capital and the executives and have to find a balance between the different stakeholders of the company and try to accomplish their different demands. Corporate governance therefore focuses on their role (Cadbury, 2000; Stiles and Taylor, 2001: 4).

(7)

board with the CEO as their director (Davies, 2006; Cadbury, 1990; 2000; Berch et.al., 2003; Bakker et.al., 2005).

Due to the corporate scandals, investors became more involved in the organizations activities concerning governance. Investors became more suspicious about the activities of the board of directors and whether they are still acting on behalf of the interest of the shareholder. To conclude, the event of corporate scandals indicates that an excessive CEO pay goes along with a weaker monitoring and controlling of the board of directors.

The market forces had to be strengthened by encouraging publication of the guidelines of the way the organization was operating its activities. Corporate governance aims to strengthen the position of the shareholders and to raise the influence of the board over the organization they direct (Cadbury, 2000). Therefore, the two principles on which every code is created are the need for sufficient disclosure and the need for decent checks and balances in the governance structure, regulation and control (Berch et.al., 2003; Sheridan et.al., 2006; Cai et.al., 2006; Cadbury, 2000; Beller, 2004).

Disclosure is essential for a successful functioning in the stock market. If CEOs are able to benefit at the expense of shareholders by inside information or if insufficient information is made available by companies, then investors might extract funds or demand a major premium for the extra risk involved. So, a high level of disclosure will lead to a decrease in the level of risk as well the level of return required by investors, leading to increased overall levels of investment (Sheridan et.al., 2006). Therefore, disclosure gives CEOs the opportunity to reduce the information asymmetry between themselves and the owners. By disclosing they can convince the shareholders that they are acting on behalf of maximizing shareholders wealth and therefore keeping the owners satisfied (Cai et.al., 2006). By monitoring the CEO one can control and prevent for the abuse of power of the CEO, simply because the board of directors lack the information necessary to effectively control them (Rutherford et.al., 2007; Berch et.al., 2003; Cadbury, 2000; Mendonca, 2006). To conclude, good control of the CEO by the board of directors is necessary to prevent excessive pay arrangements for the CEO.

(8)

To sum, corporate governance describes the way in which ownership is organised and how company’s are directed and controlled. Due to the event of several corporate scandals the focus on good corporate governance became increasingly important. The two main principles of corporate governance are disclosure and regulation. When both principles are performed well within the organization, the organization is transparent for all the stakeholders and integrity is satisfying. This leaves no space for the CEO to decide on self enrichment practices. Other said, when the board of directors perform their control function well, excessive pay can be prevented.

2.2 Agency Theory

The way a board can contribute in the organization, and for this paper more specifically in the relation with the CEO, depends crucially on which theoretical perspective is adopted (Stiles and Taylor, 2001: 10). The theoretical perspective called “stewardship theory” rises out of the organization theory and defines a role of the board of ensuring the stewardship of firm assets. Limitation of the theory is that it is largely untested. The “resource dependence theory” is originated out of the sociology field. This theory claims that boards serve as a provider of information to reduce environmental uncertainty and to extract resources for company operations. Boards can exercise their control by being the source of key information. Although it focuses on resource attainment no focus on resource use is included in this theory. In line with this theoretical perspective the “class hegemony theory” comes into play. This theory adopts the view that organizations are the agents of a social class and that the board will emphasize on selecting the “right” directors in terms of social status and influence. A power elite can control the extent of management involvement (Roberts et.al., 2005; Stiles and Taylor, 2001: 18). A last theoretical perspective applicable to the board of directors, according to Stiles and Taylor (2001: 12) is called the “agency theory”, and it is this last theory that will be used as the basis for the theoretical framework of this paper. Although it is limited by its ignorance of the complexity of organizations, its basic validity remains taken for granted and a majority of research had been conducted under the theoretical umbrella of agency theory (Conyon, 2006; Bebchuk and Fried, 2006; Hendry, 2005; Bebchuk et.al., 2002; Boyd, 1994; Stiles and Taylor, 2001: 12; Roberts et.al., 2005; Dalton and Dalton, 2005). In addition, “agency theory” has its theoretical origin within the economic field, the same field the paper objective addresses.

(9)

growth to returns where empire building might go along with status and higher pay. This continuous tension between the principal and agent has to be moderated by the board of directors. So, according to this normative theory, moral hazard refers to the potential for agents to neglect their duties and act opportunistically (Rutherford et.al., 2007; Stiles and Taylor, 2001: 15). To conclude, when the board of directors fail to sufficiently control the CEO, the CEO has the possibility to entrench him or herself with higher pay.

Therefore, different tools are required to align the different interests of the principal and the agent and hence to overcome this so called “agency problem” (Aguilera and Jackson, 2003; Stiles and Taylor, 2001: 14; Bebchuk and Fried, 2006; Boyd, 1994; Rutherford et.al., 2007).

Agency problems arise when shareholders face difficulties when monitoring the management of the company. One problem occurs when the shareholder does not posses complete information to make justified decisions since it is hard to verify what the agent is actually doing (Aguilera and Jackson, 2003; Eisenhardt, 1989; Brunello et.al., 1996). A second potential problem that could evolve comes from Eisenhardt (1989) and comes along when a conflict between the goals and desires of the principal and the agent occurs. The last problem arises when there is a conflict in risk sharing when the principal and the agent have different attitudes toward risk bearing what causes them to prefer different actions to overcome their risk issues (Eisenhardt, 1989; Fama and Jensen, 1983).

Ultimately the agency theory aims to align this dispersion of interests and consequently minimize agency costs. According to Stiles and Taylor (2001: 60) and Aguilera (2005) several control mechanisms have been proposed to ensure alignment of managerial discretion with shareholder interests but the greatest deal of attention has focused on the chief internal mechanism for control: the board of directors. CEOs departure from strategies favoured by the shareholder may involve inefficient behaviour which reduces the size of the “corporate pie”. The reductions in aggregate organization value caused by such deviations are called “agency costs” (Bebchuk and Fried, 2006: 16). Therefore an organization seeks to design the most proficient compensation arrangements by bargaining at arm’s length with CEOs over their pay, solely with the best interests of the organization and the shareholders in mind (Bebchuk and Fried, 2006; Conyon, 2006). This approach is called the optimal contracting approach (Bebchuk and Fried, 2006; Conyon, 2006; Bebchuk et.al., 2002; Bebchuk and Fried, 2003).

Optimal contracting approach. The optimal contracting approach was often addressed as a framework for researchers in understanding CEO pay arrangements (Bebchuk and Fried, 2003). Its assumption is that the board of directors take an independent position vis-à-vis the CEO in bargaining the compensation arrangements, with the ultimate goal to arrange sufficient incentives for the CEO to maximize shareholders value (Bebchuk and Fried, 2006).

(10)

increasing shareholders value. In that way the board of directors can make up for the fact that they are not able to fully monitor the CEO (Bebchuk and Fried, 2006). Another way to slot in manners of the CEO that are consistent with shareholders interest is to gain access to richer information to verify the behaviour of the CEO. In that way CEO pay will be less likely based on firm performance (Eisenhardt, 1989). According to Fama and Jensen (1983), the board of directors has always the supremacy to compensate the CEO and to approve and monitor vital decisions for shareholders interest. This setting of rewards by the non-executive also has the result of avoiding the invidious position of executives determining their own pay rises, with the obvious threat of excessive pay increases (Stiles and Taylor, 2001: 75). In other words, in the optimal contracting approach the board has a certain degree of power to control the CEO compensation arrangements, preventing CEOs from gaining too much control over their own CEO pay arrangements. To conclude, when the CEO is controlled well by the board of directors, the CEO won’t be able to entrench him or herself with excessive pay arrangements.

Managerial power approach. Like the optimal contracting approach, the managerial power approach does recognize the agency problem inbuilt in the relationship between the shareholder and the manager. However, it does not identify CEO pay as a solution to the agency problem. Rather, the pay setting arrangements itself are seen as part of the problem (Bebchuk and Fried, 2006). Under the managerial power approach the board of directors pleases CEOs by going along with excessive compensation arrangement that are in favour of the CEO (Bebchuk and Grinstein, 2005). To conclude, by pleasing the CEO the board of directors is not performing its controlling tasks well and causes the pay of the CEO to raise.

This excess CEO pay constitutes a rent which is an extra amount what they obtain beyond what they would get under the optimal contracting approach. When setting CEO pay the board of directors and the CEO face potential reputation loss and embarrassment when caught extracting rents. Therefore, to avoid outrage, the compensation designers try to camouflage the exact composition and amount of CEO pay (Bebchuk and Fried, 2006; Conyon, 2006).

This recognized power of the CEO in setting CEO pay arrangements is becoming increasingly more present and accepted (Bebchuk and Fried, 2003; 2006; Bebchuk et.al., 2002). According to Shen (2005) anecdotal and empirical evidence has consistently suggested that many non-executives are too passive, and sometimes inattentive, to management decisions. Hence, research from a social control perspective has shown that there exist strong social pressures for the NEDs not to pursue governance reforms or to challenge management decisions. The same research also points out the existence of disincentives for the NED when challenging the CEO leading to limited managerial autonomy (Westphal and Khanna, 2003).

(11)

able to gain excessive pay since they have close ties with their board of directors (Bebchuk and Fried, 2003; Murphy and Zabojnik, 2004). Moreover, the CEO plays an important role in re-nominating directors of the board, providing directors with an incentive to please the CEO by feeling gratitude and obligation toward the CEO (Matsumura and Shin, 2005; Bebchuk and Fried, 2003; Core et.al., 1998; Becht et.al., 2003; Ghosh and Sirmans, 2005, Shen, 2005). The traditional way of selecting NEDs for British boards was by this “buddy system”. The Chairman would come up with some suitable names for nomination in the board of directors (Cadbury, 2002: 59, Shen, 2005; Stiles and Taylor, 2001: 6). Also Singh and Harianto (1989) noticed that the CEO of a company may offer the board of directors an attractive contract and consulting agreements. When personal relationships constitute the foundation for the board of directors´ affiliation, a common sense of loyalty to the CEO may be even stronger (Daily et.al., 1996). This boardroom atmosphere has been recognized for decades and provides a valuable way for CEOs to influence their own pay arrangements (Kozlowski and Grasso, 2003; Knowledge at Wharton, 2005). As an example one might think about Cedric Brown at the privatized British Gas, whose 75 % pay rise in 1994 (to $ 885.000) caused wide spread criticism, although this was little compared with the 170 % pay rise (to $ 3.17 milion) for Carlton’s Michael Green a year later (Stiles and Taylor, 2001: 5). In addition, the Higgs Review notes that almost half of the NEDs surveyed were recruited through personal contacts or friendships (Aguilera, 2005). Old customs die hard and one of the findings in a recent study pointed out that these practises are still around (Cadbury, 2002: 60). To conclude, by pleasing the CEO, the board of directors is not performing its control tasks well, causing an opportunity for the CEO pay to raise.

Interpersonal influence methods. Westphal (1997) also pulls forward the imbalanced power of the CEO in accordance with the fat cat theory. Westphal (1997) argues that due to the increased call for a stricter monitoring and controlling on management decisions by making the board of directors more independent, the opposite effect will be gained. More specifically, Westphal (1997) prompts that greater board independence may move the CEO to use interpersonal influence methods – which are divided into persuasion and ingratiation - that significantly offset the effect of board independence. Moreover, Eisenhardt (1889) and Boyd (1994) also acknowledge that the CEO has ´no choice´ but to circumvent the control mechanisms of the board of directors by making use of interpersonal influence methods. The consequences of the CEOs interpersonal influence behaviour lead to higher salary increases (Kipnis and Schmidt, 1988; Westphal, 1997). Core et.al. (1998) and Bebchuk and Fried (2006) also acknowledge the power of the CEO. Along with the managerial power approach, they support the assumption that CEO compensation is higher when the board is relatively weak of ineffectual, giving the CEO the opportunity to influence its own wages. In other words, the board of directors should be independent in order for the CEO compensation to be dogged by the board of directors.

(12)

focus on satisfying the shareholders. The interest of the principal is present in the optimal contracting approach. Its assumption is that the board of directors take an independent position vis-à-vis the CEO in bargaining the compensation arrangements, with the ultimate goal to arrange sufficient incentives for the CEO to maximize shareholders value. The interest of the agent have a tendency toward the managerial power approach and the fat cat theory. In this approach the board of directors pleases CEOs by going along with excessive compensation arrangement that are in favour of the CEO. Although for both approaches a stable foundation exists, the existing interest in the high pay levels of the CEO of the last years in combination with the corporate scandals seem to be more consistent with the managerial power approach. Hence, the weak confidence of the public in the actions of the board of directors leaves space for the managerial power approach to be appropriate in this era. To conclude, a weaker board control will go along with a higher CEO compensation.

2.3 Institutional Context

To put the literature about corporate governance and agency theory more in the perspective of this paper’s research domain, I’ll now look into the institutional context. In this paper different countries are included to overcome the limited framework in which earlier research attempts have been made. Hence, to make board structure a relevant and valuable variable, the two board structures need to be represented by different countries. By developing an institutional context more insights can be gained about specific country behaviour. These insights can be used to prevent bias to be caused in the results due to specific country behaviour.

Anglo-Saxon versus Euro-Continental. First of all, a distinction can be made between two different institutional systems in which a country can behave. The USA and the UK are placed within the Anglo-Saxon system whereas the Euro Continental system has it roots in Northern and Western Europe. Aguilera and Jackson (2003) describe the former in terms of financing through equity, dispersed ownership, active markets for corporate control, and flexible labour markets. The latter is characterized by long term debt finance, weak markets for corporate control, ownership by large block holders, and rigid labour markets. Moreover, Vloerberghs (2004) makes a contribution by claiming that the labour market in the Anglo-Saxon countries is external and people are being hired and fired more easily in the USA and the UK. In the Euro-Continental system, the labour market is internal and people stay longer within the organization. Especially the regulated labour market with its high job security in the Euro-Continental countries is remarkable (Vloerberghs, 2004; Graafland, 1992).

Graafland (1992) assumes a certain incumbent power that is prevalent in the Euro-Continental countries. This leaves an opportunity for CEOs in the Netherlands to set their own wages higher above the Anglo-Saxon countries with the flexible labour markets (Lindbeck and Snower; 1988, 2001). To conclude, CEO compensation is higher in a two-tier board structure than in a one-tier structure.

(13)

culture1. In his study, Hofstede points out a high score on individualism for the USA. Bakker et.al.

(2005) puts this in a broader perspective by adding that the Anglo-Saxon model is based on individual success. Trompenaars study on cultural differences created dimensions that differentiate cultures and how these cultural dimensions will have impact on the way societies conceive organizations (Kreitner et.al., 2002: 101; Lasserre, 2003: 290). Trompenaars shows by a preference for pay-for-performance that the UK and the Netherlands are individualistic. To conclude, whatever the board structure of the organization may be, the CEOworks for money rewards and their own benefits, being characterized as an individualist (Kreitner et.al., 2002: 102).

Also Crane and Matten (2007) assume executives in the Anglo-Saxon system to be rewarded with regard to the firm’s performance on the stock market. However, they claim executive pay in the Euro-Continental system to be less performance related. Rather, according to this normative theory, it has been viewed as an arrangement between the CEO and the board. To conclude, in a one-tier board structure the board of directors is better able to manage the interest of the CEO toward maximising shareholders value than in a two-tier board structure, indicating a higher board control in a one-tier structure than in a two tier structure.

Another dimension of Hofstede´s study is the long-term versus the short-term orientation. The focus of the Netherlands is more on a long-term orientation whereas the Anglo-Saxon countries are more focused on short-term considerations (Bakker et.al., 2005). Therefore, the shareholders model described later on, suits the Anglo-Saxon countries well. However, concerning CEO compensation the USA is more focused on long term compensation whereas the Netherlands have significant less stock options and other long term incentive plans making part of their CEO pay arrangements2 3.

Concerning masculinity the USA and the UK score high but the Netherlands score exceptionally low4. The Netherlands are more characterized by nurturing feminine traits instead of

masculine traits such as success and performance.

Another dimension of Trompenaars concerns specific versus diffuse cultures. In specific countries home and business are strictly separated and in diffuse cultures both worlds are interrelated (Kreitner et.al., 2002:102). The USA, the UK and the Netherlands are all characterized by a highly specific culture (Lasserre, 2003: 292).

A last dimension by Trompenaars is achievement versus ascription. In achievement cultures emphasis is put on what you have accomplished, in ascription cultures your personality counts (Kreitner et.al., 2002: 103). Anglo-Saxon countries can be characterized as extremely achievement oriented (Kreitner et.al., 2002:103; Lasserre, 2003: 292), whereas the Netherlands is more moderate.

Stakeholder versus Shareholder. A last relevant point of the institutional context will be discussed within the presence of different power allocation within the organization. The most widely

1 Retrieved from http://www.geert-hofstede.com, Cultural Dimensions 2 Retrieved from http://www.Forbes.com.

(14)

accepted, stylized dichotomy of power allocation is between the so called shareholder-oriented and stakeholder-oriented models.

The Anglo-Saxon model is styled by the shareholder model, were the shareholder favours a dominant position (Aguilera, 2005). His way of thinking and acting have far reaching consequences for the way management is fulfilling its duties. In UK law the rights of property have traditionally had precedence over most other rights, so that ownership of a company gives its shareholders pole position in the ranking to stakeholders (Davies, 2006: 6). In the Euro-Continental model the needs of all stakeholders are taken into account (Crane and Matten, 2007; Bakker et.al., 2005; Stadler et.al., 2006; Lasserre, 2003: 297; Aguilera, 2005, Aguilera and Jackson, 2003; Berch et.al., 2003). Hence, sometimes the role of the stakeholders is equally or even more important than that of the shareholder. For example, in the Netherlands all nominations to the supervisory board have to be acceptable both to the employees and to the shareholders (Cadbury, 1990: 64). Contrary, in the USA an employee has nothing to say about the appointment of the board of directors (Crane and Matten, 2007). In their article, Stadler et.al. (2006) point out an empirical research of Turner and Trompenaars (1993). They show that the US is indeed more shareholder based since 40% of the US CEOs are convinced that the solely goal of a company is making profit whereas in Germany 24% of the managers opted for a second goal, namely responsibility for the well-being of the stakeholders, making Germany indeed a stakeholder oriented country. It seems to be that the degree of conflict of interest between the shareholders and the management is less within the shareholder approach in comparison with the stakeholder approach (Bajuk, 2005). Sanders and Carpenter (1998) show e.g. that due to the preference for the USA to focus on long-term CEO compensation the interests of the CEO with those of the shareholders converges. To conclude, due to the difference between the shareholder model and the stakeholder model, board control will be higher in a one-tier structure than in a two-tier structure.

To sum, both the UK and the USA can be styled by an Anglo-Saxon system and a presence of the shareholder model. The Netherlands can be styled by the Euro-Continental system and is characterized by a stakeholders model. Also with regard to the cultural characteristics, the distinction between the two systems is remarkable. It can be stated that the UK and the USA will experience high pressure for achieving results, for the principal this will mean maximizing shareholders´ value. The Netherlands, in contrary, are relatively more focused on uncertainty avoidance and feminine traits, and the value of other stakeholders within the organization. For the UK and the USA this means that a strict monitoring policy will be applied on the CEO, since maximizing shareholders value is extremely important. To conclude, board control will be better in a one-tier structure than in a two-tier structure.

2.4 Board structure

(15)

Whereas the UK and the USA are characterized by a one-tier structure, the Netherlands is styled by a two-tier structure. While the structures are different there are still similarities in actual board practices in that both types of systems recognize a supervisory position and a managerial position. In general, both the board of directors (in the one-tier structure) and the supervisory board (in the two-tier structure) are elected by shareholders and are responsible for appointing the executives. In addition, both the board of directors and the supervisory board have the responsibility of realizing compliance with laws and regulations. (Krivogorsky, 2006).

Two-tier board structure. Each structure has its own characteristics. The two-tier board has a clear separation between task of monitoring and that of management. The supervisory board oversees the direction of the business and the management board is responsible for the running of the company by obtaining the goals, aims, policy and strategy (Cadbury, 2002: 70). It has to render account to the supervisory board and the shareholders5. The structure rigorously separates the control function from

the management function, and members of the one board cannot be members of the other. The two-tier structure has as its main advantage its clarity of design and the separation of duties since the functions are kept absolutely distinct (Cadbury, 2002: 73) This separation is enshrined in law and the legal responsibilities of the two sets of board members are different (Cadbury, 2002: 70).

One-tier board structure. In a one-tier structure the executive board with the CEO and the board of directors are caught together in one body. In the one-tier board, and mostly prevalent in the UK, the board of directors is held responsible for the company’s fall from grace (Cadbury, 1990: 29). In practice one-tier boards are frequently close to management (Berch et.al., 2003; Vutt, 2006). In a one-tier structure the board of directors are said to have not any knowledge disadvantage over the executive board (Bakker et.al., 2005). Hence, it is even said that the potential competitive advantage of the one-tier structure lies in its combination of the depth of knowledge of the business of the executive directors and the breath of experience of the non-executive directors (Cadbury, 2002: p.50). This knowledge advantage causes the board of directors to be always perfectly informed about the operational business of the CEO, leading to perfect monitoring and control being an advantage for the one-tier structure. To conclude, board control will be higher in a one-tier structure than in a two-tier structure.

Despite sharing a common legal system, the UK and the USA have chosen to address corporate innovation in very different ways, as illustrated by comparisons between the situation before and after the Sarbanes-Oxley Act (Aguilera, 2005). Differences between USA and UK board structures are that USA boards are mainly made up of non-executive directors, less executive directors and that the top positions of CEO and Chairman are often combined by one person (Cadbury, 1990: 27; Dalton and Dalton, 2005; Davies, 2006: 7). A split of these two roles is generally perceived as a transitional arrangement or a sign of weakness. The result is that the power is usually in the hands of

(16)

the CEO (Cadbury, 2002: 69, Aguilera, 2005). This over centralization of power is evident in the gap between the CEOs salary and that of other executives. This gap is much larger in the USA than in Euro-Continental countries, were contingent pay never fully developed (Aguilera, 2005).

One-tier versus Two-tier. When comparing the board structures, some differences are quite prevalent. First of all, in a two-tier structure, the supervisory board members are barred from performing management functions (Berch et.al., 2003). This leads to different kind of people on a two-tier board in comparison with the people on a one-tier board. This distinction arises because the members as a whole of an operating board will require a different set of attributes from those that will be looked for from the members of a strictly monitoring body (Cadbury, 2002: 71). Consequently, this makes the task of the management board in a two-tier structure more complex and time-consuming in comparison with the one-tier board, where the structure leaves space for the board of directors to perform more tasks. To conclude, the more complex and time-consuming tasks of the CEO in the two-tier board structure go along with a higher CEO compensation in the two-two-tier structure than in the one-tier structure.

Second of all, one might claim that in a two-tier structure - due to the strict separation - the supervisory board has less interaction with the management board in comparison to the one-tier structure. In the one-tier structure there remains full control of the whole board one every aspect of the organizations activities. It initiates action and sees that the action which it has initiated is carried out whereas the supervisory board of a two-tier structure has to approve action, not to initiate it (Cadbury, 1990: 63). Moreover, the strength of the one-tier structure may result in a closer relationship and a better information flow between the board of directors and the executive board through its more frequently meetings (Krivogorsky, 2006; Cadbury, 1990: 64). Tosi and Gomez-Mejia (1989) also investigated the role of the board of directors in the CEO pay setting process. They found that monitoring of the board by frequent interaction of the board of directors with the CEO reduced the influence of the CEO in its pay setting arrangements. Bajuk (2005) argues that the integration of the two bodies in a one-tier structure brings about a more efficient and faster adoption of important decisions due to the frequent interactions within one body. To conclude, the existence of one body provides the structure with a knowledge advantage offering the board of directors a better opportunity to sufficiently control. Therefore, the board control will be higher in a one-tier board structure than in a two-tier board structure.

(17)

board, it will earn significantly more in comparison with the CEO of a tier structure. In the one-tier structure there the total board of directors is responsible for the well being of the organization, leaving space for the supervisory board to perform more tasks. Along with this strict separation in the two-tier structure, the separated bodies have less interaction in comparison with the one-tier structure causing a high possibility that the two-tier structure fails to sufficiently monitor. Since the one-tier structure is characterized by an in depth knowledge and interaction, the possibility to control the CEO compensation practices is higher within this structure.

2.5 Hypotheses

The different theories and approaches according corporate governance and CEO compensation policies have been discussed. They give different kind of thought for the relation a board can play vis-à-vis the CEO and what role board structure can perform. Hence, when trying to interpret the several approaches more in depth, remarkable directions in which the relations can evolve became apparent.

First of all, the corporate scandals over the last years showed that weak corporate governance could lead executives to entrench themselves with large pay increases. Therefore, the two principles on which every code is created are the need for sufficient disclosure and regulation and control. When both principles are performed well within the organization, the organization is transparent for all the stakeholders and integrity is satisfying. This leaves no space for the CEO to act on behalf of self interest. Furthermore, agency theory pointed out that there is a continuous tension between the principal and the agent and that it is the control task of the board of directors to align these interests. This leads to the first hypothesis:

“There is a negative correlation between board control and CEO compensation.”

Secondly, in a two tier structure the separation between the monitoring and management positions are significantly more strict than in a one-tier structure. Along comes a larger responsibility for the CEO of a two-tier structure to manage the organization and keep it from falling to grace. In a one-tier structure the separation becomes more blurred and therefore leaves significantly more space for the CEO to share its responsibilities. Furthermore, out of the institutional context it became apparent that the Euro-Continental countries have a certain incumbent power, leaving an opportunity for CEOs to set their own wages in the Netherlands higher above the Anglo-Saxon countries with their flexible labour markets.This leads to the second hypothesis:

(18)

Third, the one-tier board structure is characterized by a shareholders model. This means that the organization aims at maximizing shareholders value. This is in accordance with the optimal contracting approach, in which the CEO and the board of directors settle pay arrangements in order for the shareholder to maximise its value. A strict monitoring policy is required to make sure the interests between the agent and principal will be kept save. Furthermore, the one-tier board structure has a main advantage over the two-tier structure in its in depth knowledge of the business of the board of directors and the frequent interaction between them. Due to this advantage the monitoring and controlling functions become more easily to apply. This leads to the third hypothesis:

H3: “The degree of board control is higher in a one-tier structure than in a two-tier structure.”

In the upcoming section the methodology will be provided. I will first present the sample design and the data, followed by an overview of the variables used. After that the methodological approach will be presented along with the additional requirements.

3. METHODOLOGY

This outline presents the research design and describes in which way the research is conducted. First, the characteristics of the dataset and accompanying inconsistencies are given. Second, an overview of the dependent, independent and control variables are given followed by the actual methodological approach. In the end some additional methodological requirements are presented.

3.1 Sample and Data Gathering

(19)

However, for the two-tier structure there is only the Netherlands to represent the board structure. Therefore, specific country behaviour can not be eliminated.

For each country the database shows information about the 1st till the 25th highest paid CEOs

and the 75th till the 100th highest paid CEOs of the specific countries. The decision for these

companies comes out of the compiled lists used to create this database and the necessity to create variance. Variance needs to be obtained to check for a correlation within the different CEO compensation levels. For the Netherlands, the list is based on De Volkskrant and for the USA the list is obtained from Forbes. For the UK the data is retrieved from The Guardian. However, a side draw needs to be made. The list for the UK is based on the FTSE 100 largest organizations, measured in annual sales. Therefore there is no guarantee that there are no other CEOs in the UK who are better paid than the 100 used for the research. However, the group objective for the research has a significant possibility of being present within these 100 largest organizations. Even more, research seems to claim that CEO compensation is related to organization size (Finkelstein and Hambrick, 1996; Sanders and Carpenter, 1998; Zeithaml, 1992; Daily et.al., 1996). For all the ranked lists concerning CEO compensation, the data is based on the year 2006.

Furthermore, this database contains information about the compensation of the CEOs within these listed organizations and whether the CEO operates within a one-tier structure or a two-tier structure. Also different variables to measure board control will be given. Moreover, the database consists of data of the organizations to make part of the control variables, which will be explained more in depth later on.

Also, the sample of my research is taken from three different countries. Since all the countries make use of a different currency it would be hard to compare the different data concerning CEO compensation, NED Pay and Company Size. Therefore I decided to randomly choose one currency for all the data, namely the US Dollar ($). The data for the overall research is based on the year 2006. As a result, the average exchange rate for the year 2006 is used to convert the European Euro and the British Pound to US Dollars. In the table below you can find the average exchange rates for the year 2006.

$1 = €0,79632 €1 = $1,2569 $1 = £0,54299 £1 = $1,8443

Source: www.gocurrency.com

(20)

way I can compare the CEO compensation between the countries by these ratios, even though differences might prevail for whatever cause.

3.2 Dependent variable

The dependent variable used in hypothesis 1 and 2 is CEO compensation. As mentioned in the literature review, there are four basic components to executive pay, namely base salary, an annual bonus plan, stock options and additional compensation (Conyon, 2006). However, according to Sanders and Carpenter (1998) the form of CEO compensation is now being recognized as critical to the governance of companies. Compensation can take the form of cash which consists of salary and bonus or long-term pay, e.g. stock options and other long-term incentive plans (Sanders and Carpenter, 1998). The measurement used in most studies on CEO pay are taking the total cash compensation as their dimension. Total cash compensation is the sum of salary and bonus (Boyd, 1994; Core et.al.., 1998; Rajagopalan and Prescott, 1990). However, in this research all the components of CEO compensation as mentioned by Canon (2006) are required, therefore taking also the long-term pay into account. This was necessary since different remuneration policies regarding the CEO compensation were found within the different countries. In the USA e.g. exercised stock options account for the main component of pay, which is 48%6. In the Netherlands exercised stock options

account for far less in setting CEO compensation arrangements7. Measuring the total compensation

separately by dividing it in long term pay (LTP) and total cash pay (TCP) for CEOs in this form leaves out biases caused by these different pay policies.

The data was collected out of the three compiled country-sample lists and consists in total out of 150 organizations with CEOs. To make the three-country sample better comparable, TCP and LTP have been converted on a ratio level. In that way, variance was saved. The level of measurement is ratio since the distances between the numbers on scale are of known size and there is a known true zero point as its origin (Siegel, 1956: 12).

3.3 Independent variables

Board structure is an independent variable which can be measured as a one-tier or a two-tier structure. Although some exceptions might exist, in general the Netherlands is characterized by a two-tier board whereas the USA and the UK are styled by a one-two-tier structure. According to Siegel (1956: 12) this can be measured by a nominal scale.

The second variable of subject is board control. For this variable it is necessary to define variables on which to estimate the selected organizations with their board of directors and CEOs. Board control needs to be made operational before being used. There are several fundamentals for a

6 Retrieved from http://www.Forbes.com.

(21)

board to be able to fulfil its monitoring and controlling tasks effectively (Boyd, 1994). Therefore, Eisenhardt (1989) recommends to use more ways to make board control operational.

I will decide on five different variables to define board control in this paper, as will be described in the upcoming text. They will be combined and their effect will be tested on the dependent variable. This variable is called overall board control and will be created on an ordinal level. The object in one category is different with the objects in another category but they also stand in a kind of relation to each other (Siegel, 1956: 12). The way overall board control is made operational can be found back in appendix 1 together with the schedule used to measure overall board control.

As said by Eisenhardt (1989) it is recommended to use more ways to make board control operational, indicating that more than the five variables used in this paper possibly constitute board control.In addition, perhaps not all variables are of equal relevance to CEO compensation. Therefore, it is interesting and necessary to test the five variables also separately with the dependent variable. Although it is in contrast with the conditions defined by scale analysis to test all variables separately while they are also combined in one variable together, in this paper the variables will also be tested separately. Overall board control will be the main independent variable but the five variables defining overall board control will be used as a ´check` for the results in this paper.

A first variable to measure board control is the level of non-executive directors compensation (Boyd, 1994). According to Fama and Jensen (1983) is the true aim of a directorship to develop a status as an expert in decision control, accompanied with increasing the human capital. To own this status direct payments to the non-executive directors (NED) should be small. When these incentives would be to high, the independence and credibility of the board NEDs would be doubtful. Therefore, board compensation of the NEDs will have a negative correlation with board control (Boyd, 1994). NED compensation is unique in that the amount of compensation can vary between the individual NEDs of the board (Boyd, 1994). NED compensation is composed of three elements; an annual retainer, per-meeting fees, and committee pay and will be measured as the total sum of these elements. For comparison with the CEO compensation to be possible, the average of a NED pay will be used.

(22)

According to Wesphal (1997), Daily et.al. (1996) and Magman et.al.(1999) board composition can be a method to measure the strength of the relationship between the CEO and the board. In addition, the effectiveness of the board depends on its composition (Cadbury, 2002: 50). Magman et.al.(1999) suggests that the right proportion of the board of directors ensures CEOs are properly paid. Several researchers claim a board of directors to be weak when it consists of a large amount of NEDs (Boyd, 1994; Core et.al., 1998; Bebchuk and Fried, 2006; Cai et.al., 2006; Magman et.al., 1999; Cadbury, 2002: 51). A large board of directors will be less cohesive since the individual feels less responsible and therefore may focus less on management pay (Bebchuk and Fried, 2006; Core et.al., 1998). Another theory with its roots in the sociology and psychology is called the `bystander effect´. This theory is the work of Darley and Latané (1968) and finds that people are more likely to receive help when a single bystander is present, than when a group of bystanders are present (Levine, 1999; Solomon et.al., 1982; Kulis, 2004). Ones own responsibility simply slips away into the crowd of onlookers (Kulis, 2004). In addition, the relation is claimed to be strongly negative (Solomon et.al., 1982), therefore the more non-executives the bigger the ´bystander effect´. Additionally, boards with a majority of non-executive directors believe they have much less scope for improvement in their performance than those where executives are in the majority. This in one of the key findings in a survey concerning board practices in the UK, published by Henley Management College in association with Towers Perrin (1998). Therefore there is a positive relation between board composition ratio and board control. The number of executives will be divided by the total number of board members.

Wesphal (1997) and Ghosh and Sirmans (2005) also give non-executive director’s side duties as a measurement of board control. They claim that directors of the board to busy serving on multiple boards diminish the board’s power. Hence, they are likely to be distracted and therefore not in the position to provide strong board control (Core et.al., 1998). In contrary, an organization might also be interested in appointing a non-executive director who is also a main board director elsewhere: they are likely to be similar minded on the debatable subject of directors pay (Fraser, 1998). However, they may have great experience in their industry, often they are very busy men with little spare time and have hardly time to mingle with the managers who are not on the board (Keenan, 2004). I will measure directors side duties by taking the board of directors average serving on other boards and committees and claim a negative relation between NED side duties and board control.

(23)

and average NEDs tenure is measured as the number of years that they had held his or her current positions (Sanders and Carpenter, 1998).

3.4 Control Variables

Based on guidance from previous research, several control variables are included. I will use a control variable to "explain" as much as possible about the performance of both the independent and the dependent variable. By adding controlling variables one corrects for inconsistencies in the measurement of board control that might be caused by these variables.

One of the control variables is company size, this one is related to the level of executive compensation (Finkelstein and Hambrick, 1996; Sanders and Carpenter, 1998; Zeithaml, 1992; Daily et.al., 1996). Sanders and Carpenter (1998) explain that the bigger the company, the higher the complexity, what will lead to a higher compensation for the CEO. Firm size has been represented by various measures, the measurement that will be used here is yearly sales (Rajagopalan and Prescott, 1990).

The second control variable I will use is CEO position tenure. Although I use CEO tenure in my independent variables, the relevance here will be different since CEO tenure will be taken on its own to check whether it affects the level and structure of CEO compensation. Earlier studies have found positive relationships between CEO compensation and its experience as a CEO (Rajagopalan and Prescott, 1990). Other studies claim for a minimum influence of CEO tenure on CEO compensation (Deckop, 1988). CEO tenure is measured as the number of years that the CEO had held his or her current position (Sanders and Carpenter, 1998; Rajagopalan and Prescott, 1990).

The last control variable will be CEO age. Studies based in the behavioural perspective have found positive relationships between CEO compensation and age (Rajagopalan and Prescott, 1990). Contrary, Deckop (1988) found evidence that CEO age has a scarce influence on CEO compensation. Age is measured in terms of number of years (Rajagopalan and Prescott, 1990).

3.5 Data analysis

(24)

For the first hypothesis the dependent variable is CEO compensation and board control is the independent variable. Total cash pay (TCP) and long term pay (LTP) will be conducted in the test separately. With a linear regression analysis I can check for a regression between CEO compensation and overall board control.

Furthermore, the hypothesis will be tested by a multiple regression analysis. With a multiple regression analysis the five variables of board control (= NEDs pay, meeting frequency, board composition, NEDs side duties and experience by tenure) can be checked for individually. In that way the test can show whether the variables correlate significantly with CEO compensation, when they are tested for separately. The multiple regression analysis will be conducted step-wise. By conducting a step-wise multiple linear regression, the independent variables are added to the model separately, all one at the time. The accumulation sequence depends on the relative influence of the dependent variable. The independent variable with the highest F value (lowest significance level) will be added first.

Also, with a multiple regression analysis multi collinearity can erase when the independent variables are strongly inter-correlated: the effect on the dependent variable becomes arbitrary (Keller and Warrack, 2003: 675) . Hence, all the independent variables ´explain´ the same variation in the dependent variable. There is no absolute rule when multi collinearity becomes ´grave´. The consequence of collinearity is two folded. First, because the variability of the coefficients is large, the sample coefficient may be far from the actual population parameter, including the possibility that the statistics and parameter give opposite signs. Second, when the coefficients are tested, the t statistics will be small, leading to the inference that there is no linear relation between the independent variables and the dependent variable, when in fact this inference might be wrong (Keller and Warrack, 2003: 675). Since I want to check for this collinearity I will take collinearity diagnostics into account.

For the second hypothesis CEO compensation is the dependent variable and board structure the independent variable. I will conduct a multiple regression analysis. This test is often used in cases in which a linear regression analysis creates a to simplistic view of realty since more factors can trigger the dependent variable. By making use of a multiple regression analysis the results get more value and meanwhile corrections are made since other influencing factors on the dependent variable are included (Keller and Warrack, 2003). The variables that will be added to control for the results are company size, CEO position tenure and CEO age. The multiple regression analysis will be conducted step wise. Also here, a check for collinearity will be performed.

(25)

For the third hypothesis my dependent variable is board control and my independent variable is board structure. It is possible to conduct a linear regression analysis when viewing overall board control with board structure. However, when viewing the variables of board control separately the test requirements change. The dependent variables exists out of five variables with one independent variable. Therefore, I will conduct, besides the linear regression analysis, a MANOVA analysis. MANOVA is an extension of analysis of variance (ANOVA) and is developed to cover for situations in which there is more than one dependent variable. In addition, MANOVA also seeks to identify the possible effects of interaction between the dependent variables.

Last, to be able to answer the main research question, I will check for a the moderating effect of board structure on the relation board control and CEO compensation. Preliminary, the graphical interaction effects will be presented to check for any interaction before conducting the moderating analysis.

4. RESULTS

This section will provide an overview of the research that has been conducted and the results evolving from the tests. First of all, the inconsistencies concerning the data collection will be presented. After that, the results are given per assumed hypothesis. At the end, a table will be presenting the summary of the results.

4.1 Inconsistencies data collection

For my research I aimed at creating a four country sample, two countries standing for one kind of board structure each. The USA and the UK would be representing the one-tier structure and the Netherlands and Germany would be representing the two-tier board structure. However, it was impossible to find any data about the CEO compensation in Germany. This might be caused by the fact that transparency in this kind of information was not necessary in Germany until recently. Consequently, I have used solely the Netherlands to represent the two-tier board structure.

(26)

4.2 Results

In this part the results of the hypotheses will be presented. Thereafter, I will conclude with a summary of the results in which the results are explained and a schematic overview is presented. 4.2.1 Hypothesis 1

For the first hypothesis: “There is a negative correlation between board control and CEO compensation” , the here after described results were found.

For board control the five variables discussed in the methodology part are taken into account. To resume, these are NED pay (NEDpay), meeting frequency (MF), board composition (BC), NEDs side duties (DSD) and experience by tenure (EXP). In order to test this assumption I took the three sample countries separately to explore whether there are any country influences. Also, the three-country samples had to be merged, since the assumption has a general basis. To recall, the countries are the Netherlands, the USA and the UK.

Pearson Correlation (one-tailed). At first sight, based on the three-country sample, board composition has a significant negative correlation at α = 0,05 with CEO compensation, indicating that a higher CEO compensation (TCP) goes along with a lower ratio of executives. Furthermore, CEO compensation (LTP) has a significant positive relation with NEDs side duties at α = 0,05. This means that the more a CEO earns in LTP the more side duties an NED on average has. When the separate measures are taken together in an ordinal scale, a significant negative correlation between overall board control and CEO compensation, both in terms of TCP and LTP, is found (Appendix 9).

When viewing the countries separately on their correlation between board control and CEO compensation, the following can be noticed. The Netherlands has a significant positive correlation between TCP and NED pay at α = 0,01, meaning that an increase in TCP goes along with an increase in NED pay. Furthermore, a significant negative correlation is found with board composition and TCP at α = 0,01, meaning that an increase in TCP goes along with a lower board composition ratio. For LTP, the Netherlands has a positive significant correlation with non-executive directors side duties at α = 0,05, indicating that an increase in LTP goes along with an increase in the average amount of side duties a NED performs. For overall board control the Netherlands has a significant negative correlation with TCP (Appendix 9).

(27)

correlation is flagged with TCP and LTP at α = 0,01 and α = 0,05 respectively. For the USA there is no correlation to be found with not one variable (Appendix 9).

Multiple regression analysis. In order to test the hypothesis, a stepwise multiple regression analysis stepwise was conducted. Since a distinction was made between CEO compensation in terms of TCP and in terms of LTP, two regression analyses had to be conducted. The model includes interaction variables to see whether the independent variables are influenced by each other. In addition, also a check for collinearity will be performed. The following first order linear model with five predictor variables and interaction was used:

Y = β0 + β1χ1 + β2χ2 + β3χ3 + β4χ4 + β5χ5+ β1,2χ1χ2 + β1,3χ1χ3 + β1,4χ1χ4 + β1,5χ1χ5 + β2,3χ2χ3 + β2,4χ2χ4 + β2,5χ2χ5 + β3,4χ3χ4 + β3,5χ3χ5 + β4,5χ4χ5 + ε

Where Y = CEO compensation (TCP ratio and LTP ratio separately) χ1 = NED pay

χ2 = board composition χ3 = meeting frequency χ4 = NEDs side duties χ5 = experience by tenure ε = error term

Regarding TCP (ratio in %) a model summary is given in table 2. The R Square of 0,118 is the percentage variation in the dependent variable that can be explained by the model after four steps, 11,8 % in this case (Voeten and van den Bercken, 2004). For the LTP the R Square shows that 6,3 % of the variation in the dependent variable can be explained by the model (table 8). Thus the contribution of board control to CEO compensation is small but systematic.

INSERT TABLE 2 AND TABLE 8 HERE

(28)

For the Netherlands, a negative significant relation can be found at α = 0,05 for board composition and TCP and a positive significant relation can be found at α = 0,01 for NED pay and TCP.

Collinearity. For collinearity, a low tolerance goes along with high VIF (variance inflation factor) and points out instability of the regression coefficient. For all the variables, the VIF is higher than the tolerance (table 4). When looking at table 6, the lowest row present the highest ´condition index´. This dimension should point out two or more high variance proportions (values near 1) collinearity to exist. For TCP only meeting frequency gains a high variance proportion. The same results can be noticed for LTP and for the case of the Netherlands with TCP and board composition and TCP with NED pay. This is not surprising since it concerns the same independent variables. To conclude, no collinearity is present

INSERT TABLE 4, TABLE 5 AND TABLE 10 HERE INSERT TABLE 6 AND TABLE 12 HERE

Linear regression analysis. Overall board control can also be tested with a simple linear regression analysis (table 14 – table 20). For TCP, a linear regression analysis shows a significant negative relationship with regard to overall board control at α = 0,01. For the LTP there also seems to be a significant negative relationship at α = 0,05. Both tests indicate that a higher CEO compensation goes along with lower board control. For both the UK and the Netherlands a significant negative relation can be found with overall board control and TCP at α = 0,01.

To sum, overall board control is negatively significant with CEO compensation. However, when the variables of board control are measured within the MRA hardly any variables are flagged as significant. Therefore, hypothesis 1 is partially accepted.

INSERT TABLE 14, TABLE 15,TABLE 16, TABLE 18, TABLE 19 AND TABLE 20 HERE

4.2.2 Hypothesis 2

The second hypothesis was formulated as follows: “CEO compensation is higher in a two-tier structure than in a one-tier structure”. The here after described results were found.

(29)

TCP for the one-tier structure of $ 3.937.693 and a $ 1.693.923 for the two-tier structure. For the LTP the average compensation for the CEO in a one-tier structure is $ 51.505 and for a two-tier $ 255.491. Pearson Correlation (one-tailed). At first sight, the correlation matrix of the three-country sample shows no significant correlation between CEO compensation and board structure. The correlation matrix does show a positive correlation between TCP ratio and company size at α = 0,01, one of the control variables (appendix 9).

The inclusion of the Netherlands is interesting given the fact that the Netherlands is characterized by organizations with both one-tier and two-tier structures, whereas the UK and the USA only have one-tier organizations. For the Netherlands there is a significant negative correlation between CEO compensation (TCP) and board structure, at α = 0,01. This indicates that in a one-tier structure in the Netherlands the CEO earns significantly more than in a two-tier structure. This is in contrast with the expectations. The Netherlands, taken separately, shows a significant positive correlation between TCP and company size at α = 0,01, indicating that an increase in TCP goes along with an increase in the yearly revenues of the organization. The last significant positive correlation is also found with one of the control variables, between TCP and CEO age at α = 0,05. This indicates that an increase in TCP goes along with an increase in the age of the CEO.

Multiple regression analysis. Furthermore, a multiple regression analysis was conducted. Since a check for collinearity will be performed the model includes interaction variables to see whether the independent variables are influenced by each other.The following first-order linear model with four predictor variables and interaction is used:

Y = β0 + β1χ1 + β2χ2 + β3χ3 + β4χ4 + β1,2χ1χ2 + β1,3χ1χ3 + β1,4χ1χ4 + β2,3χ2χ3 + β2,4χ2χ4 + β3,4χ3χ4 + ε

Where Y= CEO compensation (TCP (ratio) and LTP (ratio) separately) χ1= one-tier/two-tier

χ2= company size χ3= CEO age χ4= CEO tenure ε = error term

Concerning the multiple regression analyses conducted, the R Square is able to explain 23,1 % of the variation in the dependent variable for TCP, for LTP no significant results were found (table 22 and table 28).

(30)

The tables 24 and 30 show the significance for TCP and LTP, where no significance for the LTP can be remarked. However, for TCP a positive significance can be found with company size and CEO age, both at α = 0,05 level, a higher TCP goes along with a higher company size.

For the Netherlands the multiple regression analysis shows an R Square of 0,363 for TCP, telling that 36,3 % of the variation in the dependent variable can be explained by the model. For the LTP 5 %, can be found for the R Square. Furthermore, there is no significance in the relationships of the variables with the dependent LTP variable. However, for the dependent TCP variable, there is a positive significance with company size at a α = 0,05.

INSERT TABLE 24 AND TABLE 30 HERE INSERT TABLE 33 AND TABLE 39 HERE INSERT TABLE 35 AND TABLE 41 HERE

Collinearity. For the TCP there is an overall clear high VIF and a low tolerance to be noticed, indicating an instability of the regression coefficient. Hence, the company size and CEO age have a high variance proportion that is significant to conclude that there is a presence of high collinearity. For the LTP, the board structure and company size are characterized by a high variance proportion, also indicating an interaction effect between the both independent variables.

When checking for collinearity between the variables in the Netherlands for TCP, no significant collinearity can be noticed. For the LTP, however, the collinearity was significant, with an interaction effect between board structure and company size, the same collinearity as remarked for the three-country sample. To conclude, collinearity is present.

INSERT TABLE 26 AND TABLE 31 HERE INSERT TABLE 37 AND TABLE 42 HERE

In short, for LTP the results are insignificant and there is a collinearity problem between board structure and company size. The hypothesis is falsified. For TCP the hypothesis is rejected.

4.2.3 Hypothesis 3

For the third hypothesis “The degree of board control is higher in a one-tier structure than in a two-tier structure” the following results were found.

Referenties

GERELATEERDE DOCUMENTEN

I draw on the complementary view to study how stock options are interactively influenced by board characteristics such as the gender diversity of board members, the

Since it is possible that auditors recognize the increased inherent and control risks associated with CEO overconfidence (financial reporting risk effect) and

The research question was: Does having a non-executive financial expert in the board reduce earnings management and how does the social status of the CEO affect this relationship..

Furthermore, the results show that board reflexivity does not statistically influence the relationship between board tenure and both board monitoring as board

In order to communicate the information to the public, there must be a process involving the collection, verification, analysis, quality control and accurate presentation of

The results of Oxelheim and Randøy (2005) study, indicates that for their sample, including the variable Anglo-American board membership lead to a 6.3% increase

It does not find support that higher salaries of CEOs and supervisory board chairmen or higher variable shares of salary of CEOs enhance the earnings management of

Van Grieken already serves for the second time on a supervisory board in which the CEO was honored as the female entrepreneur of the year in the Netherlands (Zakenvrouw van