The influence of board diversity on excessive CEO compensation

(1)

Thesis:

The influence of board diversity on excessive CEO

compensation

Final Draft

Naam: Ivo Spil Studentnr: 5603803 Datum: 3 oktober 2018

Studie: Executive Master of Finance & Control Vak: Thesis EMFC

(2)

2 Verklaring eigen werk

Hierbij verklaar ik, Ivo Spil, dat ik deze scriptie zelf geschreven heb en dat ik de volledige verantwoordelijkheid op me neem voor de inhoud ervan.

Ik bevestig dat de tekst en het werk dat in deze scriptie gepresenteerd wordt origineel is en dat ik geen gebruik heb gemaakt van andere bronnen dan die welke in de tekst en in de referenties worden genoemd.

De Faculteit Economie en Bedrijfskunde is alleen verantwoordelijk voor de begeleiding tot het inleveren van de scriptie, niet voor de inhoud

(3)

3

This paper’s aim is to provide more insights into the effect of board diversity on the level of CEO pay and the influence of board diversity on excessive CEO pay. First in the literature review we describe the Board of Directors as an internal governance mechanism with aim on good governance policy. Next, we describe how increased board diversity could have an influence on board effectiveness and on the board’s task performance. Improved task performance in the form of improved monitoring, could add to good governance and

therefore prevent excessive CEO compensation. We measure for U.S. companies in the years 2014-2016 six diversity characteristics, 1.) independence ratio, 2.) female ratio, 3.) financial expertise, 4.)ethnic background diversity, 5. age spread and 6.) board size and how those influence CEO pay level. Furthermore, we measure for the same sample the influence of aggregated diversity on excessive pay for larger companies compared to smaller companies and for companies with higher board diversity, compared to companies with lower board diversity. Last, we measure how board diversity could decrease the odds on the presence of excessive compensation in a firm. We do not find any significant presence of excessive compensation, nor do we find a different effect of board diversity for smaller firms compared to larger firms. Neither do we find significant evidence on a negative effect of increased board diversity on the decreased odds for a firm to have excessive compensation.

(6)

6

Chapter 1: Introduction

1.1 Background

Executive compensation remains a hot topic in news bulletins and the public debate. For example, during March 2018 the new executive compensation plan of Ralph Hamers, CEO of ING (one of the largest banks of The Netherlands) is openly discussed in Dutch newspapers. The Board of Directors of ING had decided that Hamer’s salary would be increased with 50 percent. This proposed increase of 50 percent was not a performance based add-on. According to labor unions this salary increase was excessive and conceived as a way to bypass the Dutch legislation about legal increases of performance based executive compensation e.g. bonus plans. The fact that this increase of salary was proposed by the Board of Directors made this a controversial topic. Questions were raised about why only the CEO’s salary was increased, instead of the salary of all the employees. In the next days the proposal from the Board of ING was discussed in late night programs and labor unions kept criticizing the proposed salary via the public media. On top of that, the Dutch Minister of Finance Hoekstra argued openly against the proposed salary increase. Eventually the pressure on ING about Hamer’s salary became too big. The board felt it was forced to withdraw the proposal, showing that executive compensation is a public matter (Heilbron, 2018).

But how did the board decide to come up with such a proposal? Isn’t it the board’s responsibility to keep executive compensation at level? Are boards making decisions according to the interests of the company’s stakeholders and are they checking the CEO’s actions? Are there some board characteristics that are supporting ‘good corporate

governance’, e.g. which characteristics could help prevent excessive executive compensation? To increase quality of corporate governance and to stimulate companies to implement proper corporate governance some countries have formulated a national corporate governance code. The Netherlands Corporate Governance Code (Code Tabaksblad) had formulated within it a target of 30% male/female1_{ratio for the Board of Directors of ‘large’ corporations.}

However, this target was released by the government in 2016 and now again back for legal proposal to initiate in 2019 (MCCG, 2016). This target was included to implement ‘better’ corporate governance since a ‘diverse’ board is favoring this quality according to national

(7)

7 institutions like the Financial Reporting Council in the UK, or in corporate governance codes like that of The Netherlands and Denmark (FRC, 2017).

To improve corporate governance and to incentivize decisions in favor of the interests of the stakeholders, corporate governance codes also direct to have ‘independent’ board members in the Board of Directors2_{. The used definitions of ‘independent board members’ do}

differ amongst several Corporate Governance codes, but main element is that these board members have limited affiliations with the company in which they serve as a board member.

This paper combines previous elements. We will test if diversity in the board indeed leads to better governance. We will measure this via excessive compensation, which we will consider as an indicator for good governance.

1.2 Research question

In this paper we connect the topics regarding executive compensation and board diversity. We would like to know if board diversity and its effect on the type of decisions the board makes influences the level of CEO compensation. The paper condenses this to the follow main research question:

Does board diversity influence excessive compensation?

The paper defines board diversity by distinguishing the following board diversity characteristics:

1.) Dependent vs. independent 2.) Gender

3.) Ethnical background 4.) Age

5.) Financial expertise (Yes or No) 6.) Board size

2_{Code Tabakblad as per 2016, rules regarding the independence of the board of directors and her members, rules}

(8)

8 1.3 Relevance of the research

This paper contributes to the literature regarding diversity within boards, the effect of diversity in boards and the behavior of these boards and on the literature on what type of decisions boards make from. We distinguish between several ‘diversity characteristics’, where many papers only focus on gender diversity (Adams & Ferreira, 2009, Campbell & Minguez-Vera, (2008), McCormick & Marcelinno, (2002) or on ‘independence’ (Adams & Ferreira, 2007), Laux, 2007, Rosenstein & Wyatt, 1990). This paper also adds the characteristics ‘age spread’, ‘ethnic background’ and ‘financial expertise’ to the equation. By adding more diversity characteristics as relevant variables that influence governance we aim to provide a broader view on the impact of ‘diversity’ on governance.

A significant amount of research is performed on the relation between board

characteristics and firm performance (Bhagat & Black, 2001, Gosh & Sirmans, 2003, Yuetang & Xiaoyan, 2006). These papers measure the effects on return of shareholders and board independence. Moreover, the relation between female ratio in the board is measured against firm performance (Carter et al, 2010, Campbell & Minguez-Vera, 2008, Triana & Miller, 2013). Instead of measuring the relation of board characteristics to firm performance these papers grasp, amongst others, the same board characteristics and relate those to effective governance of the firm. We measure effective governance by measuring excessive compensation and thus shift the research from the relation of board diversity to firm performance, to the relation of board diversity to governance.

Furthermore, this paper adds to the research done before regarding excessive compensation and firm size. Research like the one performed by Hill et al. (2016) test if ‘excessiveness’ of CEO pay increases with firm size. This paper follows the methodology of Hill et al (2016) and adds diversity into the tested model. By adding board diversity to the measured models, we acknowledge the influence of board on governance and excessive pay. And instead of only testing firm size and its effect on excessiveness of CEO pay, we measure the level of board diversity to excessiveness of pay. We would like to know if more diversity leads to better governance and thus decreases chances on excessive pay.

This paper aims to contribute to the public debate on excessive executive compensation and on required board diversity. Several governance codes as Code Tabaksblad direct on gender diversity and ethnical diversity, we will test if the perceived improvement on governance indeed shows up in our results.

(9)

9 For finance practitioners such as business controllers or financial controllers the results could be useful regarding a better understanding on remuneration levels. The paper does provide a view on which determinants do or do not play a role in decreasing the chance on excessive CEO compensation. It provides them a view on the literature about the dynamics between board diversity and (excessive) CEO pay. Controllers concerned with corporate governance topics, executive compensation and preventing value leakages from the firm with regards to board composition can gain more insight on the effect of some board characteristics and (excessive) CEO compensation. It provides them more insights on how board diversity does play a role and it could add to their critical view with regards to regulations set by governing bodies and regulations like Code Tabaksblad. These tend to create directives on board diversity with, amongst others, the aim to improve governance. But does this extra board diversity indeed has a positive effect on governance? This paper will test that. Controllers involved with board selection procedures could gain knowledge regarding which tools could be used on how to influence excessive compensation. These controllers will get a better understanding on how excessive compensation could be explained or could (not) be prevented.

1.4 Reading guide

After the introduction provided in Chapter 1, Chapter 2 will provide the theoretical

framework which will be de foundation of this paper. It will describe the concept of corporate governance, the role of boards and it will describe board characteristics which could influence board behavior. Next to this, it will describe the relations between boards of directors and CEO compensation and what the effects and on how adding diversity in the board could diminish the chance on excessive CEO compensation After providing the academic background we will formulate the hypothesis of the research conducted in this paper.

Chapter 3 will describe the research methodology. It explains the selection of the sample, it will elaborate on the measurement of relevant parameters and it will provide the model that will be used to test the hypothesis. The quantitative analysis is performed in Chapter 4. It will provide the descriptive statistics and the outcomes of the linear and logit regressions will be explained.

Finally, chapter 5 will present the conclusions of the research and chapter 6 provides a discussion on the limitations of the performed research in this paper and will provide some

(10)

10 proposals for further research on the topic of board diversity and excessive CEO

(11)

11

Chapter 2: Literature and hypothesis

2.1 Introduction

In the following chapter we will perform a literature review. We connect board characteristics with board effectiveness and we will link board diversity to positive influence on good

governance. The paper’s aim is to provide a view on board diversity and its influence on governance. We will do this by measuring excessive compensation, and by stating that

excessive compensation is an indicator for less effective governance. Lastly, we link effective monitoring performed by more diverse boards to the diminishing chances on excessive compensation.

2.2 Corporate governance and boards of directors

At the base of corporate governance lays the fundamental theorem of agency theory. In listed firms the ownership and control of the corporation and its assets are separated between the principal (shareholders) and the agent (manager). As ownership and control do not fully coincide, a potential of conflicts between the owners and the controllers could arise. Having ownership and control separately allocated between people could have many advantages. However, since it is impossible to write perfect contracts or to have perfect monitoring on the manager’s decisions, this dissolution between ownership and control reduces the value of the firm (Denis and McConnel, 2003). Realizing the flaws of this organizational design, the question arises on how to limit and minimize the loss of value of the corporation because of this corporate set-up. One of the many governance mechanisms to solve these kinds of organizational flaws is to implement a board of directors.

2.3 The role of the board of directors regarding governance

In most companies around the world a board of directors can be found at the top of the

organization. In the U.S. the main purpose of the board of directors is to represent the interests of the shareholders and to make sure actions of management are according to these same interests. Part of their main tasks are to hire, fire, monitor and to determine the compensation of the management of the firm (Denis and McConnel, 2003).

Mace (1971) elaborates on the monitoring role regarding the board. This monitoring is performed on behalf of the shareholders who could be quite dispersed in some situations, regarding their ownership of the firm. This dispersity makes it rather difficult for the

(12)

12 shareholders to monitor the firm’s management on a day-to-day basis. Since an individual shareholder only owns a relatively small part of the firm, he/she has little incentive to incur costs for monitoring the manager’s actions (John & Senbet, 1998). Furthermore, the free-rider problem reduces this incentive even more, reducing the effort of individual shareholders to coordinate their joint actions (Holderness, 2009). Therefore, a centralized board of directors would perform this monitoring role.

McNulty and Pettigrew (1999) elaborate on the ‘strategy role’ of the board of directors. In this role the board of directors facilitate in providing strategical context for the organization by support in strategy formulation, assist in shaping strategic content, support in making strategic decisions and therefore provide a context and strategical context for the organization. Mace (1971) adds to this that boards could act as a council for management. In this role as council, management could ask advice from the board of directors and the board of directors could function as a sounding board for top management.

Before we described two main roles3_:

1. The monitoring role 2. The strategy role

Later in this paper we will elaborate on what board effectiveness means. Regarding this effectiveness, this paper will focus on the monitoring role of the board. We state that better monitoring is a form of effective governance. Minichilli et al. (2009) provide a concept which describes the effectiveness of the board as a function of the boards influence on ‘task

performance’. This paper focuses on monitoring as one of the main tasks belonging to the board. We assume that better monitoring is a form of better ‘task performance’.

2.4 Governance and compensation

One topic which mostly lies within the scope of the board of directors is executive pay.

Especially, the remuneration of the CEO. However, views on how remuneration is determined do differ. One could ask questions like: is the board of directors actually in control? Do they determine CEO pay independently? Is this remuneration aligned with the interests of

shareholders and does the CEO have a say on his own remuneration too? Multiple views on

3_{This are two main roles of the board of directors we acknowledge in this paper. However, we are aware that in}

the literature multiple descriptions of the roles of directors exist. Therefore, we would like to emphasize that we just provide a way to look at board roles but we are aware that multiple views exist.

(13)

13 how CEO pay is determined exist. We would like to elaborate on The Managerial Power View.

2.4.1. The Managerial Power View

The Managerial Power View maintains that CEO compensation is the result of poor governance and the significant influence powerful managers have on their boards. Starting point of this is view is the self-interested CEO as presented by Berle and Means (1932) and Jensen and Meckling (1976). Supporters of the Managerial Power view are Bebchuk and Fried. They have written multiple papers (e.g. Bebchuk and Fried, 2004) on how entrenched executives ‘capture’ board members who will serve the CEO by approving lucrative

compensation plans. They describe that CEOs not only collect these rents via observable base salaries but also collect difficult to observe forms of compensation. Stealth compensation like deferred pay, pension plans, loans or other perquisites. The managerial influence of the executive on his own compensation packages is of such magnitude that pay arrangements get distorted which results in extra costs for the company. In other words, the executive increases its own compensation with such amount that it could be considered as a value leakage. This value leakage would not have been there if the managerial power was decreased by having an effective board which could set compensation level at a non-excessive level. The managerial power view states that executives are earning more than market efficiency and maximum shareholder value would dictate (Schneider, 2013). As a result, this can mean that

compensation packages are not mitigating the principal-agent problem, but they are actually a manifestation of this particular problem.

In the managerial power view, the dynamics are as follows. Powerful executives could distort optimal pay choices, by for example, having a large role in appointing the members of the board of directors. Independent directors could agree with pay arrangements that are in favor of the CEO since they want to keep their position as director. By disagreeing with a paying package, the director could risk its position during the next board appointment process (Murphy, 2012). Secondly, powerful executives sometimes have a say in the directors pay. By disagreeing with the executive’s remuneration package directors could risk the level of their own package and would therefore always agree (Schneider, 2013).

The managerial power view thus implicates that effective boards could offset the effect of the executive’s power. By optimizing their monitoring role, setting independent remuneration committees in place and being effect, board could enhance governance and thus reduce excessive compensation. Van Essen et al. (2012) perform a meta-analysis on studies

(14)

14 researching the Managerial Power View. They find that CEOs that tend to have more say in their own compensation setting process (via directors or even via themselves) tend to have relative higher compensation. This could be offset by the board having more power. This is measured by setting board size as a proxy for board power. They assume that the larger the board, how lower the level of CEO power. They measure a double role of the CEO by looking if he is also the Chairman of the board. When the CEO has both roles, this is considered to be higher CEO power. Measuring the relation of these two characteristics with the level of CEO compensation, they find that when boards are larger, and the Chairman role is separated from the CEO role, compensation of the CEO tends to be lower. Van Essen et al. (2012) are not able to confirm that the Managerial Power View could be taken as a fact, however, they aim to provide evidence on CEO power enhancing factors which are in line with the Managerial Power View and this Managerial Power could be reduced by having an effective board in place.

2.4.2 Views on excessive compensation

The Managerial Power View states that CEOs could increase their remuneration level to ‘excessive’ levels when no proper governance mechanism in the form of an effective board is in place. However, there is no consensus on what exactly ‘excessive’ means and what level of CEO remuneration is to be assessed as excessive. One example of an approach to measure excessive compensation is the approach of Conyon et al (2011). They compare CEO

remuneration in the U.S. and compare this with CEO remuneration in the U.K. They find that when corrected for risk stemmed from equity incentives pay, U.S. CEOs are not paid

excessively. This benchmarking approach assumes that market conditions in both countries are exactly the same and comparable.

Another approach is the one taken by Hill et al. (2016). They sample American listed firms and match those based on their size. They create clusters of firms within their sample based on industry type, year, size quartile and salary level of the CEO. To be able to measure the ‘excessive’ part of CEO pay, they rank the firms within the same cluster from large to small. After having the firms ranked, they match the largest firm with the next largest firm within the same cluster and measure the difference between the larger test firm and the

smaller, next largest, benchmarked firm. Then they subtract the level of remuneration paid out by the smaller benchmarked firm from the larger test firm. When a positive difference occurs, in other words, when the smaller benchmark firm pays out more than the larger test firm does, the part that is paid out more by the benchmark firm compared to the test firm, this part is

(15)

15 considered excessive. This is based on the assumption that managing a larger firm requires more ability from the CEO for which he/she probably is ‘fairly’ compensated. Hallock & Torok (2010) show that larger firms pay out more to their CEOs than smaller firms do.

Murphy (2012) takes a slightly different approach. He tends to see the part of remuneration as ‘excessive’ when paying out that part is against shareholders’ interests. In other words, the excessive part of the remuneration could be seen as value ‘leakage’ from the firm. Murphy acknowledges the fact that when excessive pay is interpreted in this way, focus is always on the level of pay, and on the level of firm performance what should be delivered against this. Naturally, this is quite difficult to assess. Is firm performance always to be measured in monetary quantities? This paper positions itself within the view of excessive pay as value leakage of the firm, and that good corporate governance measure, like having an effective board, could prevent or at least lessen this value leakage. Core et al (1998) do agree with this view. They state that effective boards do prevent value loss for shareholders by determining ‘right’ remuneration levels. We interpret this that effective boards are able to be prevent excessive compensation to some extent.

2.5 Board characteristics, board diversity and board effectiveness

John & Senbet (1998) conduct research on corporate governance and board effectiveness. They state that the effectiveness of a board is determined by its independence, size and its composition. This means that adding heterogeneity to a board could increase effectiveness of the board.

Van Ees et al. (2008) suggest that board characteristics could be seen as conditioning factors and that they play a big role in board behavior. Therefore, in their research they state that the specific characteristics of the board could be considered as an antecedent of effective work relations. Outside directors could for example provide previously unknown views to the organization. Influence from outside board members on other board members plays a large role in the effectiveness of boards on governance of the organization. In time, the different directors merge into an ‘in-group’ and this leads to an emphasis on consensus and cohesion over critical assessment. Figure 1. shows how Van Ees et al. (1998) view board characteristics as an endogenous element which influence board behavior and interactions within the board. The decisions made via these interactions within the board influences board effectiveness.

Minichilli et al. (2009) discuss the influence of board diversity on the effectiveness on board task performance. This board task performance could be related to the two main tasks discussed in paragraph 2.3. regarding their fulfillment of the strategy role and the monitoring

(16)

16 role. In their view effectiveness on board task performance is influenced by, amongst others, board diversity.

Figure 1.: View of Van Ees et al. (1998) on the influence of board characteristics on board behavior

Board diversity significantly influences the outcomes of board behavior, decisions and thus their task performance (Milliken and Martins, 1996). The enhancing effects of board diversity on its monitoring effectiveness is the ‘presence’ of knowledge and skills within the board that are would not have been there if this diversity was not added. Additionally, diversity could add to the critical debate existing within the board. This could add to, among other things, the critical view of the board on management’s behavior. This critical view enhances the

monitoring role and disciplining role of the board and could then add to effectiveness on governance. This good governance reduces the chance on excessive pay to the CEO.

Moreover, the presence of knowledge and skills could be considered as relevant predictors of the board’s ability and effectiveness to perform their board tasks.

Minichilli et al. (2009) interpret board diversity in two ways: diversity regarding professional background and diversity regarding ethnical background. These two kinds of diversity could add to the different perspective on issues a board handle. It adds to the

cognitive output of the board as a whole and could be in favor of decisions which are best for shareholders as whole. One of these interests is preventing value leakage in the form of excessive pay.

(17)

17 2.6 Which board (diversity) characteristics improve the governance role

This paragraph elaborates on the literature about board characteristics which could influence board behavior. These board characteristics are a limited selection of characteristics described in papers like the ones of Core et al (1998), Johnson et al. (2013), Berghe & Levrau (2004) and many more. We emphasize that this paper does not intend to capture all board

characteristics which could influence board behavior and its decisions. This paper has identified a selection of what we would like to call ‘diversity characteristics. Again, we emphasize that this is just a limited selection and that effective governance could be influenced by probably many more characteristics.

In the following subparagraphs we elaborate on the following selection of diversity characteristics of the board:

 Inside vs. outside directors  Gender

 Ethnic background  Financial expertise  Age

 Board size

2.6.1 Inside vs. Outside directors / dependent vs. independent directors

Hermalin and Weisbach (1998) show a relation between the ratio of outside directors in the board and board effectiveness. Core et al (1998) state that boards are influenced by the CEO. A higher ratio of insider board members could create a situation where the board loses its independence. This diminishes on monitoring and thus negatively affect governance performed by the board. Thus, increasing chances on excessive CEO remuneration. John et al (1998) add on the influence of number of outside directors on board

independence and effective corporate governance. The number of outside directors enhances decision making regarding increasing value for shareholders. These decisions could also play a role in keeping monetary value within the firm. Cash flow decision like limiting cash flow in the form of CEO pay could be such a value increasing decision. Increasing the number of outside directors could therefore result in better governance.

This paper aims to test the theory that more outside board members increase

(18)

18 the company, could disrupt the influence of the CEO has on the board via inside board

members (Jensen, 1993) and bring outside vision regarding the strategy of the company. We therefore hypothesize that higher outside ratio of board members strengthens the governance within the company

2.6.2 Gender

Having higher gender quality in boards starts with the rationale of having equality of treatment and equal chances for the women. In line with this thought, several countries implemented quota’s in their law or in legal directives (such as corporate governance codes) to encourage female involvement in top management (The World Bank, 2016). Per 2015, at least 15 countries had inserted such quotas in their corporate governance codes (Terjesen, 2015). The effects of increased female ratio’s in the board on board decisions and firm attributes is not conclusive.

Adams and Ferreira (2009) intend to show multiple effects of having women in the board of directors. Firstly, they show that female members of the board have higher

attendance rates than those of their male companions. Next, they show that female members of the board tend to sit on monitoring-related committees more often. They are more often assigned to audit committees, nominating boards or other corporate governance committees. They therefore assume that women have a positive effect on governance by improving monitoring behavior. Adding to the task performance monitoring role and a relative larger commitment, females add to participation and effectiveness which are relevant elements in impacting effectiveness of the board.

2.6.3 Ethnic background

The interest of ethnic and demographic diversity is increasing following cultural, social and political views on firms and on corporate directorships. This increase in interest is following two points of view. The first view argues that ethnic minorities with the corporate skills, network and competences deserve the opportunity to join top management and the corporate board. The second viewpoint argues that ethnic diversity in the board improves governance of the firm (Carter et al., 2010). According to Carter these ethnic minority directors show unique skills and competences which create additional value. This is based on the human capital theory of Becker (1964). Terjesen et al. (2009) indicate in their review that personal attributes like background, education, skills, and other experience could contribute to the value of the firm. Human capital is therefore not only influenced by gender, but also to ethnical

(19)

19 background of the directors. This complements to the theory resource dependence theory which states that the resources of the organization do influence the actions and decisions taken within the firm.

Peterson et al. (2007) claim that in US Fortune 500 companies, African-American board members fulfill significant different roles than their Caucasian counterparts and are more often fulfilling roles in an audit committee or other executive committee. They do correct for director traits, firm characteristics and resource-dependence roles.

In the research of Hillman et al (2002) they theorize that ‘status characteristics theory’ could explain that people with a different ethnic background are held to higher standards regarding moral standards and having competences. This could lead to different behavior, which could actually affect the way they fulfill their directorship within the board, leading to better governance.

Van der Walt and Ingley (2003) add to this discussion by stating that directors with a different ethnic background are in a way more distant and disconnected from incumbent management. In other words, they have less difficulty to step away from the general way of working that is incorporated in the company. Additionally, they have more ease with raising questions regarding management and would less likely echo the voice of management. By bringing other skills and experiences to the table, they do not only add value to the firm, but could also improve governance.

Stronger commitment of board members with a different ethnical background could add to the effectiveness on the board. Stronger monitoring could lead to a higher prevention rate on excessive CEO pay and could therefore significantly add to good governance by the board of directors.

2.6.4. Financial expertise

In the literature we find evidence that professional background, such as financial expertise, could be seen as a diversity component of the board (Van der Walt, 2003). However, the views on financial expertise and the effects on boards behavior do differ. Carcello et al (2006) measure these effects of financial expertise on board behavior. They find that financial

expertise in the board, and mainly adding independent board members with financial expertise in the audit committee could improve governance and could mitigation of earnings

management. Moreover, they find that this type of board member could be substituted by other corporate governance methods like stronger monitoring on financial decisions by management.

(20)

20 DeFond et al. (2005) find evidence of the market valuating board of director members with financial expertise. Share prices of companies who added a board member with financial expertise increased significantly shortly after the announcement about appointing these board members. Specifically, this was measured by the three day cumulated aggregated return of the stock. A fallacy of the research of DeFond et al. (2005) is that this positive market reaction is only significant when the new board member as accounting financial expertise and only in environments where already strong corporate governance is in place. They assume that this positive market reaction is the manifestation of the market valuating the probability of better governance.

Lastly, Güner et al. (2008) find no evidence of the influence of board members with financial expertise on compensation policy compared to the influence of board members without financial expertise. The last point we consider as an important one since this paper aims to measure the relations of diversity characteristics of the board of directors in this policy. However, this paper aims to add more weight on the view on financial expertise of board members and its enhancing effects on mitigating personal earnings management by the CEO. We acknowledge the enhancing effects on board effectiveness and good governance by adding financial expertise to the board and thus that increasing board members with financial expertise could lower chances on excessive CEO remuneration.

2.6.5 Age

Views regarding the age of director members differ in the literature. One of the assumptions we find is that board members after a certain age lose their effectiveness as board member (Core et al. 1998). The thought behind this is that when directors become older, they start to serve in more boards. Age then acts as a proxy for serving in a certain amount of boards by a board member. The higher this amount of boards grows, the less able the particular board member is to pay proper attention to his/her tasks as a board member. Therefore, decreasing governance effectiveness. In 1996 The National Association of Directors advocated to have a restriction of maximal age of 70 years old for board of director member.

Nevertheless, age acts as a double-edge sword. While Core et al. (1998) hypothesize that higher age coincides with lack of attention due to involvement in too many boards, Johnson et al (2013) argue that age could be a proxy for experience.

This argument is shared by Platt and Platt (2012), who measure the age of board members involved with firms that succumb to bankruptcy and find that the level of age is negatively

(21)

21 correlated with the firm going bankrupt. Platt and Platt (2012) do not answer the question if other factors besides ‘experience’ play a role.

When age is seen as a proxy for experience, bringing more aged board members to the board could add to governance since these directors could be more capable in monitoring the board. At the same time, governance effectiveness could decrease by age due to the increasing involvement in other boards. Thus, outcomes are ambiguous. This paper views that age adds to the effectiveness of the board on governance and improving task performance by adding experience. Increasing age diversity could add to responsiveness and consensus orientation, elements of good governance by the board (Patton, 2008). By having larger age diversity in the board, the board could include multiple perspectives to governance policy and hereby add to serving in the needs of all shareholders. By improving its task performance, we take note on the diminishing effect of age diversity within the board on excessive CEO compensation. 2.6.6 Board size

There is no consensus in the literature regarding the effect of board size on behavior of the board as both effects of the size of the board could be positive or negative. Amason and Sapienza (1997) theorize on board size being the equivalent of the cognitive capability of the board as a whole. Additionally, it could also be a proxy for the board’s expertise. Increasing the amount of board members could lead to more valuable experience on which the

organization could rely on. Smith et al. (1994) elaborates on the positive relationship between the size of the board and the amount of expertise in it as well. A larger board could consist of a higher amount of people with different backgrounds, different types of education and different skill sets. Therefore, boards with higher amount of expertise on it are able to provide better monitoring and advise of better quality than smaller boards do. Thus, a larger board could act as a better governance mechanism than a small board could (Eisenhardt and Schoonhoven, 1990). Lastly, Goodstein et al. (1994) notes that larger boards could provide better governance since they are better equipped to reduce CEO domination due of their size. It is more difficult for the CEO to build consensus in a larger group of people compared to a smaller group.

In contrary, Jensen (1993) describes that board size could have a turning point in which having an additional member to the board could actually have a downturn with regards to the efficiency of decision making within the board. Eisenberg et al (1998) and Hackman (1990) describe the consequences and the negative group dynamics that could have effect on too large boards.

(22)

22 Despite the different outcomes of research on board size and board effectiveness on

governance, this paper takes the view that increasing board size could have an enhancing effect on board task performance. A larger board could add to responsiveness, consensus orientation and effectiveness by having a better perspective on all the shareholders’ needs. Therefore, adding to the monitoring role and adding to good corporate governance. This paper will take the view that increasing board size could lower the chance on bad corporate

governance policies which add to excessive CEO compensation.

2.7 Could a board be better if she is diverse?

The literature above elaborated on different topics. First we elaborated on the role of the board of directors and its influence on corporate governance and thus its influence on CEO compensation policies. It elaborated on board characteristics which influence the effectiveness of board by improving on the monitoring function of the board. By improving monitoring, boards are enabled to provide better governance and therefore could have a positive influence on preventing excessive compensation. The definition of excessive compensation is not ultimately clear. However, we note that excessive pay could be seen as a value leakage from the company, something which is not in the interest of shareholders. By being more diverse, a board is enabled to improve on monitoring, improve on making decisions which are in the interests of shareholders and by this improve on corporate governance. This paper would therefore like to state that a diverse board could be more effective in preventing bad governance policies by better monitoring which decrease chances on excessive pay-out polices to the CEO.

2.8 Hypothesis

First we will test a general model. We conduct research on several diversity characteristics to evaluate the significance of the relationship they have regarding CEO pay. It is hard to connect this to terms like ‘better governance’, however it provides us a quantitative base regarding significant variables in play regarding excessive compensation. The first hypothesis we will test is:

H1.: More diversity in the board improves governance regarding the level of CEO pay.

(23)

23 Since this is a very broad statement, this paper adds sub-hypothesizes on the different board characteristics:

 Independent vs. dependent: A higher ratio of independent board members is expected to have a negative relation on the level of CEO compensation.

 Gender diversity: A higher ratio of women in the board is expected to have a negative relation on the level of CEO compensation.

 Ethnical background: A higher variance of ethnical background diversity is expected to have a negative relation on the level of CEO compensation.

 Age: A higher age spread is expected to have a negative relation on the level of CEO compensation.

 Financial expertise: A higher ratio of people with financial expertise in the board is expected to have a negative relation on the level of CEO compensation.

 Board size: A larger board is expected to have a negative relation on the level of CEO compensation.

While above hypotheses focus on the relationship of certain diversity characteristics on the level of CEO pay, we perform further research regarding the effects of board diversity on excessive CEO pay. First we test the effect of size of the firm on excessive pay. Larger firms tend to pay larger amounts to their CEOs. We elaborate more on this chapter 3. We therefore add the following hypothesis:

H2.: Excessive CEO pay is higher in larger companies versus smaller companies

In chapter 3 we explain on how we differentiate on small and large companies and on how we measure ‘excessive compensation’.

Next, we would like to know if board diversity influences the magnitude of excessive pay. We hypothesis that by having more diverse boards, they will become more effective regarding good governance. Therefore, this will reduce the magnitude of excessive pay. We formulate the following hypothesis:

(24)

24 Lastly, we will test the influence of diversity characteristics of the board on the presence of excessive pay. We conduct research to find out if by adding diversity to the board, chances on excessive compensation are reduced. Therefore, we test the third hypotheses:

H4.: A higher rate of board diversity decreases the chance on excessive CEO compensation

This paper will perform quantitative research on above hypotheses. In chapter 3 this paper will explain which research methodologies are used, how we collected the test sample data and which variables are included in our quantitative analysis.

(25)

25

Chapter 3: Research methodology

In the following chapter we will describe which databases are used for the data. Furthermore, we will elaborate about the sample, its size and for what items we filtered in this data to have a clean data set for our quantitative analysis.

3.1 Sample

For our analysis we need three types of data. First we need data about CEO compensation. Secondly, we would like to have information about the boards of the companies from which the observed CEOs are. We are looking for info about gender, ethnical background,

inside/outside, age, and financial expertise. Third, we need to have general info about financial performance of the companies of which the CEOs are from. Since all these three types of information are available for S&P 1500 companies we will take our sample out of this list of North American companies. We conduct the research on recent data, so we select for the year 2014 until 2016 which gives us a three-year window for our quantitative analysis.

For the retrieval of the first type of info, CEO compensation data, we collect our data from Compustat Execucomp. For the second type of data, board member information, we pick the Institutional Shareholder Services (ISS) database. This database keeps track of board member specifics for S&P 1500 companies. For the third type of information, company financials, we select the Compustat database for North America. Now we want to connect the data out of these three separate databases to make one useable dataset. The common factor of data out of these three databases is the CUSIP code for the companies. This is a unique serial number allocated to a specific company. We now select all matches for the years 2014 until 2016, so all CUSIP codes we find in all three databases for all three years.

Next, we filter for unavailable information per company. This means when board characteristics data is not available for the company, or when financial performance data or CEO remuneration data is completely missing, these companies are left out. After cleansing our sample consists of 3,041 company years of 1,095 unique companies.

3.2 Methodology & empirical models to be tested

We will conduct our quantitative analysis via the ordinary least squares regression method. We will perform robustness checks on this model and we will check for level of significance for all our variables. We measure the relationship of board diversity on CEO compensation. The model we will test:

(26)

26 Log CEO compensationnt = β0 + β1 Independence rationt + β2 Gender rationt + β3 Ethnical

background variancent + β4 Age spreadnt + β5 Financial expertise rationt + CONTROLSnt-1 +

εnt (1)

In this model we measure several variables of companies (prefix ‘n’) on time ‘t’. β0 is a

constant measuring an average level on which CEO compensation starts within our sample. The ‘Independence ratio’ is defined as the ratio of independent board members in the board. ‘Gender ratio’ is the ratio of women in the board. ‘Ethnical background variance’ is the variance of different ethnical backgrounds in the board. ‘Age spread’ is the variance of age of the board members within the board. Lastly, ‘Financial expertise ratio’ is the ratio of people with financial expertise within the board. We will elaborate on the ‘CONTROLS’ in the next paragraph. ε is an error term

In our empirical model we correct with control variables, ‘CONTROLSnt-1’ in model (1).

These control variables could better be seen as proxies. For example, we proxy for size since this could influence the size of the compensation for the CEO; larger firms are more likely to reward the CEO with relative higher compensation than smaller firms will. The same goes for the risk profile of companies. Companies with a higher risk profile are probably presenting CEOs with relative higher rewards than ‘low’ risk profile companies do since CEOs of riskier companies would like to be rewarded for running a riskier company. Next to the diversity variable explained in previous chapter, we will include the following control variables:

 Sales of the company in the previous period (t-1): The CEO will probably have a Short Term Incentive (STI) bonus plan based on sales. Higher sales in the previous period will probably lead to a higher compensation for the CEO in the current period.  Investment opportunities, which are defined by the average market to book ratio over

five years in the previous period. Rationale behind this control variable is that companies with higher investment opportunities perform better and so will reward their CEO with a relative higher compensation than companies with less investment opportunities do.

 Return On Assets (ROA), which acts as a performance measure. We will define this as EBITDA of the previous period divided by the book value of assets of the previous period. Better performing companies will probably pay-out relatively higher amounts than less performing companies do.

(27)

27  Stock return of the previous period, also a performance measure.

 Standard deviation of ROA over the last five years in the previous year. This will control for the risk profile of the company. Companies with a relative higher risk profile are probably more likely to compensate their CEOs for this higher risk compared to firms with a less higher risk profile.

 Standard deviation of stock return, this control variable is measured by the variance of stock return over the last five years in the previous period. This control measure is also meant to control for the risk profile of the firm.

Furthermore, we will measure the influence of board diversity on good governance by measuring the effect of board diversity on excessive CEO pay. We use the method described in the paper of Hill et al. 2016. To measure excessive pay we first replicate the framework laid out by Core & Guay (2010). They state that justified pay of a CEO is the sum of the CEO’s ability, plus the cost of his/her effort and plus an incentive which acts as a risk premium for the CEO. This model can be summarized as follows:

Justified Pay = CEO ability + Cost for Effort + Incentive Risk Premium (2)

Actual Pay = Justified Pay + Excessive Pay (3)

However, we extend this model with measures for good governance and state that good governance explains justified pay, equation (2) will then be as follows:

Justified Pay: CEO ability + Cost for Effort + Incentive Risk Premium + Good Governance (4) Given this model we now will benchmark CEOs out of the sample data against other CEOs out of the sample data active in the same year and same industry in which we use the 48 industry classification of Fama and French (1997). Then, in these groups of companies with the same year and industry we form clusters. All firms falling in the same cluster of industry-year combination are then considered a matched firm portfolio. A minimum of two companies is required to form a portfolio (Hill et al. 2016).

(28)

28 For each observed company we will benchmark for a similar company within the same

portfolio. We will match two firms by ordering the firms by market value4_{of the firm and}

then select a firm with the next highest market value. This benchmark represents our

empirical proxy for the test firm's next best CEO candidate. Our research design assumes that a reasonable candidate to replace the test firm CEO will be similar in terms of; 1) historical firm performance, and 2) industry, and 3) size (Hill et al. 2016).

Now having matched each firm we can design a model to measure excessive

compensation which measures excessive compensation that is not explained by factors related to economically justified pay. We do this by differencing the CEO compensation of each test firm with its benchmarked firm:

Difference Actual Pay = Test Actual Pay – Benchmarked Actual Pay (5)

In our setting this could be translated into:

Difference Actual Pay = Difference Justified Pay + Excessive Compensation (6)

Here the difference in the pay level is explained by CEO ability, CEO effort and his risk premium and good governance. Translated into a regression setting it could be noted down as:

Difference Compensation = β0 + ∑αAbility + ∑γEffort + ∑δRisk Premium +∑ηGood

Governance + ε (7)

Difference Compensation (D_Totalcompnt ) is calculated by subtracting the benchmarked

CEO compensation of that of the test CEO compensation. The compensation of the CEO is measured by the log of total compensation. All ‘Ability’, ‘Effort’ and ‘Good Governance’ variables are measured as the test firm value less the benchmark firm value, denoted with the prefix ‘D_’ (Hill et al. 2016). This presents us with the following model:

4_{We define market value as market value of equity plus book value of debt in year t. We will rank on revenue of}

(29)

29 D_Totalcompnt = β0 + β1 D_Revenuent-1 + β2 D_Stock returnnt-1 + β3 D_ROAnt-1 + β4 D_σStock

return over last 5 yearsnt-1 + β5 D_ σROA over last 5 yearsnt-1 + β6 D_divsumnt + ε

(8) In this model the first 5 parameters act as controls, or better said, proxies for firm performance which could translate to CEO success, proxy for firm risk and proxy for investment opportunities. The sixth variable ‘divsum’ is an aggregated ‘board diversity’ parameter in which the six diversity parameters 1.) board size, 2.) ratio of independent board members, 3.) female ratio, 4.) ratio of members with financial expertise, 5.) age spread of board members and 6.) ethnical variety are merged to one parameter. This is done by

measuring the values of each of these six parameters for each company and determine if each diversity variable of this company is below or above the median within the company’s industry-year. When one of these variables is above median it will get a value of ‘1’. For example, if a company has a value which is above median for all six parameters, the value of the sum of its diversity parameter (‘divsum’) will have a value of 6. When they are all below median, this divsum parameter will get a value of 0. The value of ‘divsum’ for each company will have a value between zero and six by following this methodology.

This will have an advantage but also a disadvantage: the advantage is that by ‘equalizing’ each parameter by giving them a value of 0 or 1, we are enabled to design an aggregated board diversity parameter. The disadvantage is that we arbitrary equalize all the six diversity parameters. Maybe, the magnitude of influence of each parameter differs from one and each other. This difference in power is not known by us, however, we are not sure if all parameters are equally affecting good governance either. By equalizing the six board diversity parameters we are maybe assigning faulty quantitative influences to some parameters on effectiveness of governance related to CEO pay.

We will then divide the sample in quartiles regarding market value and regarding their level of the ‘divsum’ variable. We will test model (8) on the full sample and on each quartile partition. By doing this, we examine the conditions which prior research suggests may be more prone to excessive compensation (Hill et. all, 2016).

The implication of using this method is that it has an inherent bias against finding excessive compensation. We assume a positive difference of total compensation between the test firm and the benchmark firm to be ‘normal’, because we assume that larger firms pay-out higher salaries. While only a negative difference, this happens when the benchmark (smaller) firm rewards a higher compensation than the test firm, is marked as ‘excessive compensation’.

(30)

30 Test outcomes which are against the excessive compensation hypotheses should therefore be interpreted with caution. Results which confirm the presence of excessive compensation will therefore be of sufficient magnitude to overcome this bias.

Thirdly we will test the effect of board diversity on the odds of having excessive

compensation. Here we cluster the sample per industry-year-size quartile5_{. Than we split these}

clusters in quartiles based on the log of total compensation. Since we want a pure

measurement in each cluster, we clean for clusters having 4 or more firms, so every cluster has a firm within each ‘total compensation quartile’. We assume that excessive compensation exists, but we also admit we do not know exactly how to measure it. Therefore, we assume that all firms belonging to the highest total compensation cluster within each industry-year-size quartile is subject to having excessive compensation. We assign an ‘excessive

compensation dummy’ to all these firms having value 1 if the firm belongs to the highest total compensation quartile. Then we will perform a binary logistic regression on this excessive compensation dummy, the proxy/control variables and the diversity variables. We prefer to select the logistic regression above the probit regression since the outcomes of a logistic regression are more intuitive and easier to interpret. The outcomes of the logistic regression could be defined as the effect of the diversity variables on decreasing the odds of the firm to be a firm belonging to the highest total compensation quartile. The model we will test will have the following form:

Excessive compensation dummynt = β0 + β1 Revenuen (t-1) + β2 ROAn t(-1) + β3 Stock return n (t-1)

+ β4 σStock return over last 5 yearsnt-1 + β5 σROA over last 5 yearsnt-1 + β6 divsumnt + ε

(9)

Testing this model is a pure measurement of the effects of diversity variables on excessive compensation. By clustering per industry, year and size quartile we correct for macro effects within each year and industry and we largely overcome heteroscedasticity issues. We expect a larger variation of total compensation between larger firms and a smaller variation of total compensation between smaller firms.

(31)

31

Chapter 4: Quantitative analysis

In the following chapter of this paper we will perform quantitative research regarding board diversity and its relation to (excessive) CEO compensation. In paragraph 4.1 we will shortly describe the sample and its characteristics. In paragraph 4.2 we will perform a Pearson correlation analysis which presents us a view on internal correlations between the different variables of our regression models. In paragraph 4.3 we will perform univariate analysis on the mean differences between the tested variables for the above median board diversity group out of the sample, versus the below median board diversity group. In paragraph 4.4 we will perform several linear regressions on multiple models. The first linear regression model is tested to uncover the relation between several board diversity characteristics on CEO compensation level. Secondly, we will perform multiple regressions on the quantitative difference of the variables between matched firms to get a view on plausible presence of excessive compensation. We will also test this for different groups within the sample where we will separate firms based on size quartile and board diversity quartile. Lastly, in paragraph 4.5. we will perform a logit regression on a defined excessive compensation parameter to get a view on the effects of the diversity variables on the odds of excessive compensation by a firm.

4.1 Descriptive analysis

Table 1. Descriptive statistics Descriptive Statistics:

The table shows an overview of the sample statistics for 3,041 firm years over the period 2014 to 2016 of 1,095 unique firms

Variable Mean Std. Dev. Min. Max.

Total compensation ($) 6,966,667 5,210,383 410,520 32,377,350

Revenue ($) $9.1b $26.2b $22.6.m $483.5b

Market value ($) $14.6b $39.5b $43.8m $615.3b

# of directors 9.3 2.0293 4.0 20.0

Ratio of females 16.1% 10.9% 0.0% 75%

Ratio financial expertise 24.2% 13.5% 0.0% 83.3%

Ratio of independent directors 80.7% 10.5% 0.0% 100%

Age spread 178.95 6,309.66 0.8 348,005.10

Ethnical diversity in % to size 19.3% 7.7% 6.3% 50.0%

N=3,041

Table 1 shows a summary of the descriptive statistics we used for this research. The sample consists of 3,041 company years, in the period of 2014 to 2016 of 1,095 unique companies out

(32)

32 of the S&P 1500. The overview shows the mean, standard deviation, minimum value in the sample and maximum value in the sample, of the main variables analyzed in this research.

The average total compensation (which is the sum of salary, bonus, other annual compensation, total value of restricted stock granted, total value of stock options granted using Black-Scholes, long-term incentive payouts and all other compensation for year t) is just below $7 million with a standard deviation of $5.2m. The average revenue of the

companies in this sample is $9 billion of which the maximum revenue is $482.5 billion. The average market value of firms in the sample is $14.6b with a maximum firm value of $615.3b. The average amount of board of directors consists of 9 directors, of which almost 1 out of 6 is female. Almost a quarter of the average board consists of people with financial expertise. The majority of the directors of the average board is independent to the firm of which most boards consist of people with multiple ethnical backgrounds (almost 2 at average).

4.2 Pearson correlation analysis

Table 2. on the next page is the representation of the performed Pearson correlation analysis. We performed this analysis on the dependent variable, independent variables, proxy variables and cross-term variables for the full sample period 2014-2016 of all 3,041 sample firms. We measure the correlation coefficients between each variable, these values are represented by their value r, and when this value is r > 0.5 we consider the correlation to be strong. The correlations are measured with a two tailed confidence interval, of which the p-value is presented underneath the value of the correlation coefficient.

Two interaction terms (cross-terms) are added based on the literature review. This is the variable ‘ind x eth’ which is an interaction term between ‘Indep’ and ‘Ethnic’. This cross-term is added because one of the research papers in the literature review measures a

significant negative effect on CEO pay when board members with a different ethnical background are also independent directors. Another interaction term is added ‘Ind x finan’ which is a cross-term between ‘Finanexp’ and ‘Indep’. This one is added since multiple papers in the literature review stated that independent directors with financial expertise could have a positive effect on an increase of CEO remuneration.

There are two correlations of higher absolute value than 0.5 which we would like to mention. First the correlation between total compensation of the current period and revenue of the period before the measured period is 0.6588 and is significant to the 1% level. This

finding is confirmed in the research literature. Hallock & Torok (2010), also find that larger firms tend to pay substantially more to their CEOs than smaller firms do.

(33)

Table 2. Pearson correlation matrix

Totalcomp Rev Stock Roa Roa5 Stock5 Ndirec Female Finanexp Indep Age Ethnic Ind x eth Totalcomp 1.0000 Rev 0.6588*** 1.0000 0.0000 Stock 0.1428*** 0.2494*** 1.0000 0.0000 0.0000 Roa 0.0289 0.0858*** 0.4365*** 1.0000 0.1114 0.0000 0.0000 Roa5 -0.0212 -0.1181*** -0.3334*** -0.4991*** 1.0000 0.2460 0.0000 0.0000 0.0000 Stock5 0.0197 0.0504*** -0.0423** -0.0421** 0.0406** 1.0000 0.2777 0.0054 0.0195 0.0203 0.0267 Ndirec 0.4287*** 0.5763*** 0.1390*** 0.0025 -0.0793*** 0.0334* 1.0000 0.0000 0.0000 0.0000 0.8886 0.0000 0.0652 Female 0.2402*** 0.3044*** 0.0524*** 0.0739*** -0.0322* 0.0229 0.2776*** 1.0000 0.0000 0.0000 0.0039 0.0000 0.0784 0.2071 0.0000 Finanexp -0.0140 0.0068 0.0020 0.0076 0.0119 0.0032 -0.1319*** -0.0062 1.0000 0.4398 0.7072 0.9123 0.6761 0.5147 0.8603 0.0000 0.7308 Indep 0.2563*** 0.1966*** 0.0378** -0.0542*** -0.0027 -0.0168 0.2412*** 0.2319*** 0.1245*** 1.0000 0.0000 0.0000 0.0373 0.0028 0.8838 0.3532 0.0000 0.0000 0.0000 Age -0.0004 0.0142 0.1085*** 0.0270 -0.0029 0.0022 0.0146 0.0026 0.0402** -0.0001 1.0000 0.9823 0.4344 0.0000 0.1371 0.8724 0.9052 0.4215 0.8846 0.0267 0.9939 Ethnic 0.1318*** 0.0963*** 0.0501*** 0.0354* -0.0203 -0.0094 -0.0733*** 0.1124*** 0.0353* 0.0962*** 0.0398** 1.0000 0.0000 0.0000 0.0058 0.0507 0.2677 0.6058 0.0001 0.0000 0.0514 0.0000 0.0282 Ind x eth 0.1989*** 0.1591*** 0.0601*** 0.0205 -0.0208 -0.0133 0.0204 0.1755*** 0.0653*** 0.3792*** 0.0371** 0.9489*** 1.0000 0.0000 0.0000 0.0009 0.2577 0.2559 0.4632 0.2598 0.0000 0.0003 0.0000 0.0407 0.0000 Ind x finan 0.0390** 0.0449** 0.0071 -0.0060 0.0139 0.0004 -0.0683*** 0.0438** 0.9727*** 0.3199*** 0.0380** 0.0492*** 0.1365*** 0.0315 0.0134 0.6965 0.7424 0.4475 0.9842 0.0002 0.0158 0.0000 0.0000 0.0360 0.0067 0.0000 N = 3,041 Variable Totalcomp Rev Stock Roa Roa5 Stock5 Ndirec Female Finanexp Indep Age Ethnic Ind x eth Ind x finan *: **: ***:

the natural logarithm of the sum of salary, bonus, other annual compensation, total value of restricted stock granted, total value of stock options granted (using Black-Scholes), long-term incentive payouts and all other compensation for year t,

The natural logarithm of the revenue of the firm for year t – 1, The return of assets for year t – 1,

The annual return to shareholders for the firm for t – 1,

The average standard deviation of the return on assets over 5 years for the firm, prior to year t, The average annual return to shareholder for the firm over 5 years, prior to year t,

The natural logarithm of the natural logarithm (double log) of the number of directors in year t, The ratio of females in the board of directors of the firm in year t,

The ratio of directors with financial expertise in the board of directors of the firm in year t, The ratio of independent directors in the board of directors for the firm in year t, The age spread of the directors in the board of directors of the firm in year t,

The ratio of the count of unique ethnical backgrounds of the directors in the board to the number of members of board of directors of the firm in year t, Cross-term between Inden and Ethnic for the firm in year t,

Cross-term between Inden and Finanexp for the firm in year t Two tailed p-value < 0.1

Two tailed p-value < 0.05 Two tailed p-value < 0.01

(34)

This is according to the literature. Revenue could be seen as a proxy for firm size and CEOs of larger firms tend to receive a higher compensation compared to CEOs of smaller

companies. This higher pay could be justified when we assume that a higher ability is required to let a larger firm perform better, relative to letting a smaller firm perform better. Within lays the assumption that CEOs have influence on firm performance.

Another notable strong significant correlation is the one between the number of directors and revenue. This is according to the literature. Larger firms tend to have larger boards. Another correlation we will note on is the one between total compensation and the number of directors in the board. This correlation has a coefficient of 0.4287 and is highly significant. The number of directors could have a positive relation with compensation level because increasing the number of directors could lead to coordination difficulties in between the board. This could lead to less effective governance which could lead to higher or

excessive compensation.

There are several other correlations with coefficient above 0.4, however those make sense because of their nature, so we do not elaborate on those broadly. Those correlation are between different performance parameters or between variables and cross-terms which are composed of those regarding variables.

4.3 Univariate analysis

In the following part we will conduct univariate analysis on the mean difference of the tested variables between two groups in our sample. The first group consist of companies with above median board diversity. We measure the value of the six diversity characteristics and test if they are above or below median. When they are above, we assign a number 1 to that above median outcome for the regarding variable, 0 otherwise. We will aggregate the 6 outcomes, which will create a sum with a minimum of outcome 0 and a maximum of 6. Of this

aggregated value (0 to 6) we measure for the company if the outcome is below or above median. When above, the company is assigned to the group ‘above median board diversity’ and otherwise to ‘below median board diversity’. Outcomes of the t-test regarding the two-group mean difference tests are shown in Table 3.

The group with companies with above median board diversity pay out a higher average remuneration than the group of below median board diversity. This could be caused by size of the company playing a role. Larger, more complex organizations tend to have larger board with more board diversity. It is inconclusive which dynamic plays a role in

The influence of board diversity on excessive CEO compensation

Thesis: