WHAT EXPLAINS THE

W

C

C

B

?

A

D

Master Thesis prepared for MSc IE&B, Faculty of Economics and Business, University of Groningen

AUTHOR RESEARCH SUPERVISOR METHODOLOGY SUPERVISOR

C.C.M.A. Revis Prof. dr. H. van Ees Prof. dr. H.W.A. Dietzenbacher

s1258184 Faculty of Economics Faculty of Economics

Lange Leidsedwarsstr. 106-1 Landleven 5, Landleven 5,

1017 NP, Amsterdam 9747 AD, Groningen 9747 AD, Groningen

c.c.m.a.revis@student.rug.nl

ABSTRACT

INDEX

1. INTRODUCTION 4

1.1 Corporate Elite Awareness 4

1.2 Executive board composition and its drivers 4

1.3 Board Diversity 6

1.4 The Netherlands as a case study 7

1.5 Paper composition 7

2. LITERATURE REVIEW: THEORY & HYPOTHESES 8

2.1 Board diversity and its impact on firm performance 8

2.2 Board members‘ background characteristics & professional experience 9

2.3Board diversity and the link with contextual factors of influence 11

2.3.1 Organizational factors: ‗Company complexity‘ 12

2.3.2 Environmental factors: ‗Industry membership‘ 14

2.3.3 Corporate Elite factors 16

CEO Dominance 16

Supervisory Board composition 18

3. BOARD COMPOSITION OUTSIDE THE NETHERLANDS 20

4. THE MODEL 23

5. THE DATA 24

5.1 The Sample 24

5.2 Background characteristics and professional experience 24

5.3 The Variables 24 5.3.1 Dependent Variables 25 Background Characteristics 25 Experience variables 25 5.3.2 Independent Variables 26 5.3.3 Control Variables 26

7. THE EMPIRICAL RESEARCH 28

8. RESULTS 31

8.1 Descriptive Results 31

8.1.1 Index Dispersion 31

8.1.2 Industry Descriptives 32

8.1.3 Board member background characteristics: Executives vs. Supervisory 34

8.1.4 Board member Experience: Executives vs. Supervisory 36

8.1.5 Complexity & CEO Dominance Descriptives 37

8.2 Analytical Results 38

9. DISCUSSION AND CONCLUSIONS 42

REFERENCES 47

APPENDICES 50

Appendix One: Data Encoding 50

Appendix Two: Descriptive Statistics & Correlations 53

Appendix Three: Industry Categorization 55

Appendix Four: Company Sample selection 56

1. INTRODUCTION

1.1 CORPORATE ELITE AWARENESS

With the explosive rise in companies‘ media exposure on numerous different platforms, information on companies has never been more accessible to the general public. This exposure not only leads to an increased scrutiny of the company, but it also leads to more attention being paid to the corporate elite. The corporate elites are no longer perceived as anonymous entities but now have recognizable faces which are subject to comparison, critique and evaluation.

Corporate elites are in control of the business sector. In the Netherlands, most companies have a two-tier board system consisting of an executive board and a supervisory board. Supervisory board members (all non-executive directors) are individuals who provide a monitoring role; they ensure that the policies implemented by the executive board are in line with the interests of the company. In addition, supervisory board members have an advisory role and provide important network resources to the company (Zahra & Pearce, 1992). Finally, members of the supervisory board are responsible for hiring, firing and compensating the executive board members. The executive board members, which will form the main focus of attention in this paper, are the group of people responsible for the day-to-day functioning of the company and who have the task of ensuring the effective and rational execution of a firm‘s strategy.

Information on these elites is made available through the annual reports of Dutch companies as well as through other sources like corporate websites. This information generally includes their remuneration, curricula vitae and general job descriptions. However, they also reveal interesting demographic characteristics such as age, ethnicity and gender; all in the name of transparency. Hence, this information not only allows for a thorough examination of who exactly form a board but also for testing hypotheses about the determinants and consequences of its constitution.

1.2 EXECUTIVE BOARD COMPOSITION AND ITS DRIVERS

paper was that organizational outcomes can be partially predicted from managerial backgrounds. More precisely, the executives‘ actions, that directly affect the organization, are believed to be driven by their own personal interpretations vis-à-vis their strategic situation. Furthermore, their interpretations are subject to the executives‘ professional experience, background characteristics, values and personalities. In short this describes the very important role that the personal characteristics of individual members of a board have on organizational outcomes.

It has thus been established that executives‘ characteristics determine their personal interpretation of situations, which in turn determine the way they respond to situations, which ultimately determine organizational outcomes. But what determines the executives‘ characteristics?This question was recently raised in a 2007 paper by Hambrick, which brings new and interesting research directions. As Hambrick says “…there is a need to turn upper

echelons theory on its head by considering executive characteristics as consequences rather than as causes… Why do top management teams look the way they do? …” (2007). Certainly, if board members

directly influence organizations‘ outcomes, then it would logically follow that it is in the best interest of the organization to ensure that this board exists of the most competent persons for the job. Nevertheless, what exactly determines the mix of executive background characteristics and experiences has rarely been researched.

Though the importance of context as a determinant of composition was also initially addressed in earlier Upper Echelons work, these thoughts of ‗causality‘ were merely to note that the occurrence of a particular set of executive backgrounds in a firm is not a random process. Although these thoughts should not detract from the theory that executive backgrounds are reflected in strategic outcomes, the interpretations of research results should keep them in mind (Hambrick & Mason, 1984).

Officer (CEO) has with regard to determining board composition as well as the role that the

supervisory board will play in determining the board‘s composition (Keck & Tushman, 1993; Haleblian & Finikelstein, 1993; Finkelstein & Hambrick, 1996).

1.3 BOARD DIVERSITY

At first glance the typical disposition of most boards worldwide remains one of elitist male homogeneity. However,when taking a closer look at corporate boards, diversity within them can be found. In the context of this paper two questions are of interest. First of all, what would classify a diverse/heterogeneous board? As Van der Walt and Ingley (2003) argue, diversity relates to the varied combination of attributes, characteristics and expertise contributed by individual board members. In this paper, these attributes, characteristics and expertise will be either in the form of demographics such as nationality, age, gender and educational background, or in terms of board members‘ professional experience within the company, the industry or abroad. All of which clearly can set board members apart from one another.

Secondly, why promote diversity in the boardroom at all? In 1978 Pfeffer and Salancik introduced Resource Dependence theory. This theory is based on the notion that environments are the source of scarce resources and that organizations are dependent on these finite resources for survival. According to Pfeffer and Salancik (1978) boards can be viewed as boundary spanners between the firm and its environment as their backgrounds and professional experience are a potentially important strategic resource. Coupled with the Upper Echelons theory, where board member characteristics have a direct effect on organizational outcomes, it follows that a diverse board may have a positive effect on organizational outcomes as they are better equipped to tackle diverse or complex environments.

1.4 THE NETHERLANDS AS A CASE STUDY

Most of the research on group composition has attained its proof from the United States where it seems that diverse boards are more and more the trend. Research that investigates the makeup of Dutch boards is relatively meagre. With a population representing many different cultures and backgrounds and a long standing history of international trade and partnerships, one could expect that Dutch executive boards have a natural tendency towards greater diversity within their ranks. However, such an expectation is not justified. A large portion of literature has demonstrated that boards are no reflection of the diverse societies within which they operate (e.g. Westphal & Milton, 2000). All the more reason to investigate the exact driving forces behind their composition.

Using cross-sectional data on 100 listed Dutch companies and their 821 board members this paper attempts to identify to what degree Dutch executive board composition is determined by contextual influencing factors found in their environment. Industry, company and corporate elite variables are considered as possible triggers for their constitution. The influence of these variables on the background characteristics and professional experience of board members will be examined. By studying this complex relationship between companies‘ environments on the one hand and the composition of their boards on the other, this paper not only researches that which has not yet been studied rigorously, but it also moves away from the more established literature which predominantly focuses on the consequences of board composition. Moreover, by focusing on Dutch listed companies, this paper attempts to add to the little knowledge existing on non-US boards in this field of research.

1.5 THE STRUCTURE OF THE PAPER

2. LITERATURE REVIEW: THEORY & HYPOTHESES

Links with the corporate environment are important as it has become an acknowledged fact that an organization will find itself subject to external pressures. It is assumed in this paper that the composition of company boards should represent the optimum board structure that best matches its configurations to its environment (Keck & Tushman, 1993). Having briefly introduced Upper Echelons theory and the idea of context, some reasons why a company would take an interest in the configuration of its board will be highlighted. Hereafter, the paper will continue with the discussion why a board‘s composition can be seen as an outcome of the three identified contextual influences.

2.1.BOARD DIVERSITY AND ITS IMPACT ON FIRM PERFORMANCE

According to Westphal and Milton (2000) boards have traditionally been viewed as a homogenous group of elites with similar socioeconomic backgrounds, educational degrees and professional training. As a result board members develop very similar views about business practices. Several authors claim this unveils deficiencies in corporate governance and, especially, missed opportunities (Carter et al., 2003; van der Walt & Ingley, 2003; Daily & Dalton, 2003; Singh & Vinnicombe, 2004; Brammer, et al., 2007).

A diverse board is said to positively affect company outcomes in a number of ways. Composed of qualified individuals who reflect a diversity of gender, ethnicity and experience, boards can take advantage of their members‘ differences to productively work together in support of the organization. Therefore, boards are thought to be able to enhance both their team performance as well as their company‘s performance by diversifying their membership (Catalyst report, 2004; Singh & Vinnicombe, 2004; van der Walt & Ingley, 2003; Krishnan & Park, 2005).

fact, the board is seen as a potentially important strategic resource for the organization, particularly by providing connections to competitors or by providing market and industry expertise (Van der Walt & Ingley, 2003). Furthermore, it means the company will maintain more effective global relationships due to well understood social and cultural sensitivity. Erhardt et al. (2003) also suggest that rapid or dynamic changes in the market will require boards to effectively deal with organizational change, signifying that heterogeneous groups may be better equipped to do so. While heterogeneity may initially produce more conflict, diverse boards are also believed to be capable of more effective decision making; a diverse board will bring a variety of perspectives entailing more alternatives that need to be explored before a decision can be reached. Additionally, a diverse board will mean increased creativity and innovation as well as a broader corporate management view. Finally, this will also mean an enhanced effectiveness of corporate leadership (Baysinger & Butler, 1985; Finkelstein & Hambrick, 1996; Carter et al., 2003; Carpenter et al., 2002, 2004).

It is a well researched fact that diverse boards, in more ways than one, have an advantage over homogenous boards when it comes to many company performance measures. Therefore, it is not surprising that board composition has become a much discussed topic. It is clear that there are definite incentives for a company to have the right mix of members on the board. The following sections will further investigate what characteristics and professional experience this ‗right mix‘ can entail and what it will depend upon.

2.2BOARD MEMBERS’ BACKGROUND CHARACTERISTICS & PROFESSIONAL EXPERIENCE

have generated substantial evidence that they are highly related to strategy and performance outcomes (Millikin & Martins, 1996; Erhardt et al.; 2003, Hambrick, 2007; Tacheva, 2007).

A distinction will be made between two sets of observable variables. The first category is background characteristics which will form the umbrella term for all variables that describe members‘ backgrounds and demographics like nationality, age, educational background and gender. These are thought to have formed the initial reference point or means of identification when asserting if someone will be an appropriate member of the board. The second category is made up of experience variables which account for specific experience that members have attained throughout their career such as international, industry or company experience. These are thought to not only further project the individuals career path and choices made but may also explain why the member is of importance to a specific board. Distinguishing between board members‘ background characteristics and their professional experience will therefore allow for both the observations on ‗fixed‘ pre-career characteristics and the choices that executives have made to acquire additional professionally relevant experience throughout their careers.

It is believed that different nationals will bring along country and region specific knowledge, expertise and network contacts as well as general understanding of foreign markets. Nationality diversity will therefore allow board members to better understand and cope with the complexity related to a firm‘s international operations. (Milliken & Martins, 1996; Erhardt et al.; 2003, Hambrick, 2007; Tacheva, 2007).

Gender is another variable which is highly related to strategy and performance

Educational background is considered a variable depicting the relevant educational level

(masters, phd etc) or educational type (finance, economics, law etc) that has already been completed by the board member by the time he or she reaches the board. It therefore reflects the amount of variation in terms of information and knowledge within the board (Williams and O‘Reilly, 1998; Simsek et al., 2005; Tacheva, 2007).

The professional experience variables will reflect board members‘ valuable sources of knowledge and expertise. In the case of international experience, this entails knowledge and expertise about foreign markets and cultures. It is often positively associated with the performance of multinational firms as it has been shown that international assignment experience positively affects managerial skills (Carpenter et al., 2001). As Tacheva (2007) explains, top executives‘ with international experience are expected to have a better understanding of the complexity and dynamics of managing a firm‘s (international) operations. Likewise, industry experience is expected to give board members increased opportunity for boundary-spanning activities through industry specific knowledge, expertise and access to external organizations or network contacts within the industry (Geletkanycz & Hambrick, 1997). Finally, company experience, which is a variable first described in this paper, will illustrate if board members have prior experience in the same company before becoming part of its executive board. Members with prior company experience will have been with the company longer and will therefore have a greater understanding of the inner workings of the company itself.

2.3 BOARD DIVERSITY AND THE LINK WITH CONTEXTUAL FACTORS OF INFLUENCE

Both the Resource Dependence theory and the Upper Echelons theory have shown the importance of the board when it comes to organizational outcomes. This section will now look at the importance of context as a determinant of executive board composition.

power of the Chief Executive Officer (CEO) as well as the role of the supervisory board (Keck & Tushman, 1993; Haleblian & Finikelstein, 1993; Finkelstein & Hambrick, 1996).

All three contextual factors will have their own specific requirements and effects on the composition of a company‘s board. For instance Sanders and Carpenter (1998) claim that in the case of company internationalization increased information-processing demands and agency concern are partially isomorphic throughout an organization and therefore it is the role of governance to accommodate the increased complexity. This entails boards comprising of the optimum mix of characteristics and professional experience that best match their configurations to their environment (Keck & Tushman, 1993).

2.3.1 Organizational factors: ‘Company complexity’

In this paper the ‗complexity‘ a company faces is thought to have an influence on the composition of its board. By this is meant the degree of a company‘s environmental turbulence or instability. Proxies often used to indicate complexity are net annual sales, or corporate diversification measures. For example, larger sales revenue or diversification could mean the company is active in more product and/or geographical markets. As Markarian and Parbonetti (2007) point out, the identification of new product opportunities, often in new markets, and anticipated changing customer needs, is essential for a sustained competitive advantage. As these types of proxies signify the company‘s strategy vis-à-vis

Organizational Factors  Company complexity

Environmental Factors  Industry membership

Corporate Elite Factors  CEO dominance

 Supervisory Board composition Executive Board Composition

Background Characteristics & Experience

Source: Author

product diversification or level of international involvement they also incorporate the dynamics of managing the company‘s‘ operations and the degree of complexity it faces. The higher the complexity of firm operations, the higher are the demands posed on the board of directors (Michel & Hambrick, 1992; Daily & Schwenk, 1996; Sanders & Carpenter, 1998; McNamara et al., 2002). According to theory a more diverse board would be needed to deal efficiently with this increased amount of complexity. Helmich and Browne (1975) conclude that a successful corporate leader must be flexible in order to harness and respond appropriately to the changing forces of the environment. In the same respect, the executive board must be flexible in order to attain the organization's goals. Creating a more diverse board in terms of board member attributes will raise the board‘s flexibility as it increases a company‘s access to valuable resources and becomes better equipped to deal with the higher management demands (Pfeffer & Salancik, 1978; Carter et al., 2003; Stiles, 2001). In the case of international operations for instance, nationality diversity or international experience are believed to increase network contacts as well as a general understanding of foreign markets and will therefore allow board members to better understand and cope with the complexity related to international operations.

Hence, top management teams with diverse background characteristics and more professional experience will be better able to manage complex environments compared to homogeneous top management teams (Carpenter, 2002; Keck, 1997). The first hypotheses will test whether the predicted company‘s need for a diversified and well equipped board is indeed positively related to the degree of its environmental turbulence or instability.

Hypothesis 1a: A company’s degree of complexity will be positively related to board diversity in terms of background characteristics.

2.3.2 Environmental factors: ‘Industry membership’

According to the Resource Dependence theory, a firm is dependent on its external environment and its ability to secure critical resources in this context will determine its chances of survival (Pfeffer & Salancik, 1978). Board members are thought to form the ‗link‘ between the firm and its environment by using their valuable knowledge and expertise to access important information and resources (Pfeffer & Salancik, 1978; Zahra & Pearce, 1992; Stiles, 2001; Carter et al., 2003; Tacheva, 2007). The environmental context in which a firm operates is said to be determined by the dynamism and the characteristics of the industry to which it belongs (Hambrick & Mason, 1984; Markarian & Parbonetti, 2007; Boone et al., 2007; Tacheva, 2007). Whether a company is growing or simply surviving as an organization, its demands for specialized board services are also likely to change accordingly. DiMaggio and Powell (1983) introduce three mechanisms through which institutional isomorphic change occurs across organizations operating in the same industry. This is interesting as it suggests that there should be distinct differences visible between industries. First of all, organizations tend to model themselves after more legitimate or successful organizations in their field. This reaction is also known as mimetic institutional

isomorphism, often a standard response to uncertainty. In uncertain environments, firms may

carry out activities to make environmental threats more manageable (Pfeffer & Salancik, 1978). One way of doing so is by changing the composition of its board. Pegels et.al (2000) argue that in the U.S. domestic airline industry, the closer the top management teams‘ heterogeneity of a firm, in terms of age, tenure, educational level and functional background, was to that of the dominant player in the industry group, the better it performed.

Conflicting levels of functional heterogeneity that lead to greater financial performance between the two industries were found. For instance, broad and high-level experience of top executives was more beneficial to the minicomputer firms whereas specialized expertise proved valuable to the cement industry. These differences were attributed to different environmental conditions and/or industry turbulence.

A company‘s external environment may also have an influence on its board composition through coercive institutional isomorphism. This is a type of isomorphic change that is due to pressure exerted on organizations by cultural expectations, regulations and by other organizations upon which they are dependent. For instance, complex banking regulations require bank executives to have significant banking experience. This immediately sets restrictions to who is considered for an executive position.

Finally, through normative institutional isomorphism individuals who occupy similar positions across a range of organizations become almost identical. Within many organizational fields the hiring of individuals is done from firms within the same industry. This results in boards comprising of individuals who have been sorted on a common set of backgrounds and attributes such as university degrees, schools and even personal characteristics.

All these forces are unique to the particular organizational field a company is part of. Hence firms within different industries will have experienced isomorphic change distinctive to that industry and set board member requirements or preferences accordingly. Therefore, it is expected that boards of companies within one industry will show different degrees of background diversity or professional experience to those in other industries.

Hypothesis 2a: Industry membership is related to the degree of board diversity in terms of background characteristics

2.3.3 Corporate Elite Factors CEO dominance:

Environmental conditions are believed to influence the distribution of power and control within an organization. These conditions affect the process of selection and departure of executives which in turn impacts organizational actions and structures (Keck & Tushman, 1993; Boone et al., 2004). However, in addition to these environmental factors a well developed body of research has determined that the CEO plays an important role in selecting new directors.

In the Netherlands the Dutch corporate governance model (Code Tabaksblat, 2004) calls for a dual board structure with the underlying principles of consensus, cooperation, a stakeholders approach and codetermination (Goodijk, 2007). This structure is intended to increase the value of the firm as a result of enhanced monitoring of management and meeting the claims of share- and stakeholders. It would seem that in the Netherlands there is little room for a CEO to exercise much influence on board composition. In general however, literature repeatedly proves otherwise stating that the CEO plays an important role in selecting new directors despite the prevalence of nominating committees (Zahra & Pearce, 1992). In fact, a model has been proposed in which board structure is the outcome of the negotiation between the CEO and outside directors (Hermalin & Weisbach, 1998).

The studies cited above show that the involvement of the CEO in the selection process of board members cannot be neglected. The question now is who will the CEO select, and why?

Scholars claim that specific social psychological mechanisms incline CEOs to prefer demographically similar colleagues. This is often referred to as ‗law of homo-social reproduction‘ (Boone et. al., 2004) which over time leads to a homogeneity equilibrium. Zajac and Westphal (1996) give an example where CEO‘s possessing master‘s degrees from Harvard Business School are inclined to favor candidates with the same qualifications, regardless if educational background is an important criterion for selection. In general, executives with similar demographics such as nationality, age or educational backgrounds to that of the CEO are said to be more likely to share the same views and beliefs and are thus expected to agree more with the suggestions and decisions that the CEO makes. This serves to reduce friction between board members and the CEO, and to reduce resistance to a CEO‘s objectives and the proposed manner in which they are attained. One manifestation of this influence is the fact that CEO tenure is positively related to the income of the CEO (Hermalin & Weisbach, 1998). Hence, the more powerful a CEO, the greater his influence in selecting these ‗indistinguishable‘ executives will be, and the more homogenous the executive board will become.

Moreover, if indeed the CEO has a say in the makeup of the board, it could also be argued that the standards the CEO sets for the new members of the board will be high in terms of professional experience. The CEO will want fellow board members to possess those complimentary resources, whether they are industry, international or company experience, which will most likely contribute to the success of the company and the board. This will then serve to enhance the reputation and credibility of the CEO. In addition, Callahan et. al. (2003) conclude that management participation in the director selection process is even positively associated with stockholder wealth. They go so far as to suggest that the existence of a separate nominating committee and lack of involvement by the CEO adversely impact performance.

whether or not a CEO has a significant impact on the composition of the board and, if so, whether the sign of its coefficients will describe if this influence leads to a homogenous board in terms of background characteristics and an increased amount of professional experience.

Hypothesis 3a: CEO tenure will be negatively related to board diversity in terms of background characteristics

Hypothesis 3b: CEO tenure will be positively related to the percentage of board members with attained international, industry or company experience

Supervisory Board Composition:

Research on board composition that considers the executives and supervisors simultaneously is relatively new to this field of literature (Daily & Dalton, 1996; Jensen & Zajac, 2004; Tacheva, 2007). As the two boards work closely together and are part of the same social environment the connection between them in terms of board composition is an interesting one.

Traditionally the agency theory has been used to describe the importance of independent directors (i.e. supervisors) in protecting shareholder interests from the self-interests (the agency costs) of management by monitoring and incentivizing the executives (Jensen & Meckling, 1976; Zahra & Pearce, 1992). The larger the agency costs, for example in larger or more complex companies, the more important this monitoring role becomes (Boone et al., 2007). Supervisory board members are therefore often chosen by the Annual Meeting of Shareholders for their expertise in a certain field or market, their networks or their position in the business community. This is also in line with the resource dependence framework (Pfeffer & Salancik, 1978) where boards must comprise of individuals that are best able to acquire necessary resources from the external environment in order to be effective.

new board members. First, one could argue that similar logic applies to the supervisory boards as with the CEO in selecting new members: It would be in the interest of the supervisory boards to elect similar executive members to themselves as it facilitates a cooperative relationship between the two boards and hence enhances the task of monitoring and advising the executive board. This would lead the two boards to consist of similar background characteristics and professional experience.

Second, the supervisory board members are chosen by the shareholders based on attributes most likely to contribute to the success of the company. This means that supervisory board members, and hence those taking seat in the nominating committee, in effect represent the optimum mix of characteristics and professional experience necessary for company success, considering the environment in which the company operates. The Resource Dependence theory would predict that nomination committees with the interests of the company at heart will in turn focus on matching new board members as best they can to the company‘s environment and so fulfill part of their fiduciary role.

In both cases, one could expect a large degree of similarity between the two boards in terms of background characteristics and professional experience. Therefore, the fourth hypothesis describes the relation between the compositions of the executive board with that of the supervisory board, i.e. does the composition of both boards indeed resemble each other in terms of background characteristics and in the level of professional experience.

Hypothesis 4a: The executive boards’ degree of diversity in terms of board member characteristics will be positively related to the supervisory boards’ degree of diversity in terms of their board member characteristics. Hypothesis 4b: The percentage of executive board members with professional

3. BOARD COMPOSITION OUTSIDE THE NETHERLANDS

To bring a bit of perspective and initial back-drop to the findings of this study, this section will briefly take a look at some trends with respect to board composition outside the Netherlands.

Using reports from Spencer Stuart1 _{a brief comparison was made across a few board}

characteristics mentioned throughout all the reports. The results are presented below in Table One. Incorporated are some results for two other European and three Anglo-Saxon countries. As the countries have different board structures, one- or two- tier, it was chosen to present numbers reflecting the entire board; both executive and non-executive members.

Table One: Board Composition outside the Netherlands

Year

of Report Average B. size CEO tenure Average Women % Foreigners % United States 2006 10.7 6.5 16 6* United Kingdom 2006 10.8 4.5 9.9 27 Canada 2006 12 5 13 17 Sweden 2004 8 6 20 19 Italy 2006 11 7.1 4.3 24 The Netherlands 2006 8.8 3.8 5.1 24.6

*non-US nationals; does not account for different ethnicities.

First of all, average board size across these countries ranges between 8 and 12 members. Sweden and the Netherlands represent the smallest boards. Canada, where the average board size is 12, has the largest boards. As was predicted in 1999 by ‗Corporate Board‘, among other things, European boards of the future were to have fewer members. What is not apparent from the table but was also included in these predictions were larger proportions of truly independent directors, and a greater number of board committees, such as nominating and compensation committees; Indicating that the role of Dutch supervisory boards would become of greater significance.

Secondly, average CEO tenure ranges from around 4 to 7 years among the countries. Interestingly, it is the Netherlands that represents the country with the shortest CEO tenure.

1 _{Spencer Stuart is a privately owned executive search consulting firm. The reports used can be found at}

According to the Spenser Stuart reports the overall trend in terms of CEO tenure has also been declining across all countries whereby Italy still lags behind with a 7 year average. This is interesting as it has a direct affect on CEO dominance, or the amount of power they will be able to accumulate while in power.

Thirdly, there is a clear disparity in terms of percentage female board members among these countries. Overall though, the figures are quite low. Sweden forms a relative exception where 1 in 5 board members is female. The lowest ratio is found in Italy where 1 in 23 board members is female. The Netherlands does not do much better with a ratio of 1 in 20. Moreover, it should be kept in mind that the numbers of female executives are drastically lower and in particular female CEO‘s. These women are still a rarity in all countries. Though to this end the reports did not all share the same measurements, which were therefore not included in the table, they do provide some interesting information. In the U.S. and the Netherlands for example only 2.4% and 1.8% of the executive members are female, respectfully. Furthermore the reports show that in Canada 1% of the CEO‘s are female and that in the UK only 0.39% of all board members are female executives. When considering either the amount of companies or the amount of board members that were taken into account in these 2006 reports, the results for the Netherlands are in fact the lowest. These results are pretty much in line with an earlier study by Bolte (2005) who depicts the global situation vis-à-vis female participation in top management teams. The Dutch lag far behind the best in class. Within Europe, the Netherlands is put in the 50th

percentile range where it falls behind the U.K. and Switzerland. However, even Europe as a whole comes in third place behind the U.S. and Canada.

corporate legislation and the Code Tabaksblatt2 _{made Dutch companies much more}

attractive to foreign investors. Currently all the large AEX companies are majority foreign owned. As a result also the nationality diversity within these boards has increased after 2004. Overall these results are quite interesting. In fact, studying why these countries differ so much from each other may be very helpful in finding out what determines board composition. However, this is outside of the scope of this thesis. This paper will continue with its focus on Dutch boards and will now proceed by presenting the model used to test the previously developed hypotheses.

2 _{The Code Tabaksblatt is the Dutch corporate governance code. It is a code of conduct for listed companies}

4. THE MODEL

In this paper the composition of Dutch executive boards, in terms of executive background characteristics diversity and professional experience will be examined by setting industry, company and corporate elite variables as possible triggers for its formation. This will entail a multiple regression model with on the left hand side a dependent variable representing the diversity in all background characteristics and professional experience and on the right the independent and control variables indicating the company, corporate elite and industry factors of influence.

BC_i = β1 + β2CPi + β3CEOi + β4SCi + β5CSi + β6XSi + β7IMi

Background Characteristics and Professional Experience:

BC: Board composition. BC will represent each separate dependent variable. In effect this means there are eight different BC‘s, namely: nationality diversity (ND), age diversity (AD), gender diversity (GD), educational level diversity (ELD), educational type diversity (ETD), level of international experience (INT), level of industry experience (IND), and finally the level of company experience (CMP).

Contextual Factors:

CP: Company complexity measure

CEO: CEO tenure used to illustrate a CEO‘s dominance.

SC: Supervisory board composition variable which, as with BC, represents each background characteristic and attained experience separately and will be used to depict the similarity between the two boards.

Control Variables

CS: Company size XS: Executive board size

IM: Environmental control variable which will control for the possible effects on board composition due to the different industry memberships.

5. THE DATA

5.1THE SAMPLE

The initial sample for this study consisted of randomly selected Dutch companies listed on the AEX, AMX, AScX, and on the local market indexes. There were a number of companies that were eliminated from this group for several reasons. First of all, companies that did not have websites or annual reports available to the public for the year 2006 were neglected. Companies that did have annual reports or websites but which lacked the adequate information needed on the members of the boards were excluded from the data set as well. Finally, companies that during 2006 ceased to exist in their past state (due to mergers or bankruptcies which could lead to complete management overhauls) were also eliminated. This process left a final sample of 100 listed companies. For all the companies data was collected on both executive and supervisory board members. This resulted in observations on a total of 821 board members.

5.2BACKGROUND CHARACTERISTICS AND PROFESSIONAL EXPERIENCE

The distinction between background characteristics and professional experience variables is made as their composition has slightly different effects on the effectiveness of the board. Studies have shown that boards may benefit from members with different background characteristics, in this paper it is assumed that board will also benefit from all members having some rather than no experience. This will entail that the way in which both variables are computed will also be different; Background characteristic variables will be good candidates for the computation of diversity indexes. These will show the ‗…distribution of a

population among groups in terms of a nominal parameter‖ (Blau, 1977), i.e. the variety of one

demographic attribute within a group; different ages, nationalities etc. However, the purpose of the latter experience variables is not to form a diversity measure but rather to provide a simplified reference of a member‘s career path choices and to give an idea of their accumulated knowledge.

5.3 THE VARIABLES

member in regards to the company in question. This variable had four possibilities allowing for both two tier board membership (i.e. executive or supervisory) and one tier board membership (i.e. dependent or independent board members). Their current position or function within the board was another variable, coded using six possible values that allowed a distinction to be made between functions such as chief executive officer or chairman of the supervisory board. The year in which they were first appointed was simply recorded as that year.

5.3.1 Dependent Variables

Background characteristics variables:

Gender was encoded as a dummy variable; zero for males and a one for females. Nationality

was recorded as stated in annual reports and reflected the country of origin of the particular board member. In some cases a second nationality, using the same encoding, was recorded.

Year of birth was recorded as found in the annual reports. At a later stage these years were

translated into ages by subtracting them from 2006. Board members' highest attained level of

education was recorded as a categorical variable with six possible values up to a PhD level.

Finally, the type of education a board member has received was also measured as a categorical variable ranging from law and general economics to informatics and political science. Again here some members had more than one educational type background; these were recorded as well using the same encoding.

Experience variables:

For these variables dummies were used to denote the different possibilities. For the variable

industry experience a ‗one‘ distinguished board members with some prior experience in the

same industry sector in which he/she currently operates and a ‗zero‘ if otherwise. Similarly,

company experience was denoted with a ‗one‘ for those with a prior commitment to the same

5.3.2 Independent Variables

A company‘s complexity was illustrated using its net sales in millions of Euros, for the year 2006. The understanding here is that net sales will give an indication of how diverse or complex the environment of a company is. Larger net sales may indicate a presence in more markets and a more dynamic customer base thus increasing the information processing capacity that is needed to respond to a larger number of variables. Moreover, by using net sales, this variable was obtained from the same annual reports and websites used for the above variables. CEO tenure was used to depict the above described CEO dominance. Here the year of the board member‘s first appointment as CEO was subtracted from the year 2006.3 _{The composition of supervisory boards (SC) was also recorded to investigate the link}

with the composition of executive boards (BC). In the model these two terms will always represent the same dependent variable: when regressing for executive boards‘ nationality diversity, SC will represent the nationality diversity for the supervisory boards and so forth. 5.2.3 Control Variables

Large teams are said to have an advantage when it comes to dealing with uncertainty. In fact the costs associated with the communication and coordination problems that also arise in large teams is believed to be outweighed by the team‘s enhanced capabilities (Haleblian & Finkelstein, 1993). Furthermore, the larger a team, the higher are the chances of diversity in them. Therefore in this paper executive board size will be used as a control variable. Similarly, company size will constitute another control variable, as a larger employee base will increase the chance of diverse individuals making it into the boardroom. For this variable the year end average total number of firm employees, for 2006 was used. Industry membership was also translated into a control variable. It controlled for possible effects on board composition due to different industry memberships. Dummy variables were used to differentiate between the different industry categories outlined in Appendix Three.

3 _{Because annual reports are written with a delay; members that joined the company in 2006 on average already}

6. GROUP AGGREGATION AND DIVERSITY INDICES

Per company a distinction was made between the executive board members and the supervisory board members. Thus, per company the above mentioned variables that are based on individuals were aggregated to these two board levels using several different measures.

First of all, the categorical variables: educational type, educational level and nationality needed to be aggregated to a group level in such a way that the diversity in terms of individual characteristics is visible. As Harrison and Klein (2007) point out, choices on diversity indexes should be driven by the theoretical specification of diversity type. Therefore, as prior researchers specifying diversity as variety have most commonly done, the chosen form here too was Blau's index (Finkelstein & Hambrick, 1996; Carpenter, 2002; Tacheva, 2007). The functional form is shown below;

Blau = [1- Σ (pi2)]

Where ‗p‘ is the percentage of team members in one of the categorical groups ‗i‘; nationality, educational level etc.4_{If all group members differ per category, the board will approximate}

unity (1-0.00=1). If all group members are exactly alike, there is no heterogeneity (1-1.00=0). Therefore, the closer Blau‘s value is to 1, the greater the board diversity on a particular variable (Blau, 1977).Similarly a board Blau index was used to measure the diversity of board members with respect to gender. Secondly, to measure age diversity within the board the standard deviation of age was computed per board. Finally, percentages were used to indicate the accumulated amount of board members‘ international experience, industry experience and company experience.

4 _{An example board consisting of four Germans, six Dutchmen and 2 Belgians, totalling 12 board members}

7. THE EMPIRICAL RESEARCH

Multivariate regression analysis was undertaken as the primary test of the hypotheses. This statistical methodology was preferred as it allowed for cross section analysis while controlling for the effects of other firm characteristics such as company- and board size and industry membership. The four hypotheses are tested using the following eight multivariate regression equations: 1. NDi = β1 + β2CPi + β3CEOi + β4SCi + β5CSi + β6XSi + β7IMi + ε 2. AD i = β1 + β2CPi + β3CEOi + β4SCi + β5CSi + β6XSi + β7IMi + ε 3. GD _i = β₁ + β₂CP_i + β₃CEO_i + β₄SC_i + β₅CS_i + β₆XS_i + β₇IM_i + ε 4. ELD i= β1 + β2CPi + β3CEOi + β4SCi + β5CSi + β6XSi + β7IMi + ε 5. ETD i= β1 + β2CPi + β3CEOi + β4SCi + β5CSi + β6XSi + β7IMi + ε 6. INT i = β1 + β2CPi + β3CEOi + β4SCi + β5CSi + β6XSi + β7IMi + ε 7. IND i= β1 + β2CPi + β3CEOi + β4SCi + β5CSi + β6XSi + β7IMi + ε 8. CMP _i= β₁ + β₂CP_i + β₃CEO_i + β₄SC_i + β₅CS_i + β₆XS_i + β₇IM_i + ε

Equation one for example, tests for each individual company, whether the nationality diversity (ND) of its board can be described as a function of the company‘s complexity (CP), CEO dominance (CEO) and the nationality diversity of its supervisory board (SC), while controlled for company size (CS), executive board size (XS) and its industry membership (IM). The other equations do the same for age diversity (AD), gender diversity (GD), educational level diversity (ELD), educational type diversity (ETD), international experience (INT), industry experience (IND), and company experience (CMP). Where β₁ is an unknown constant, β2, 3-7 are unknown parameters and ε is the error term. Note, that hypothesis two

(influence of CEO dominance) predicts β3 to have a negative sign with the background

characteristics ND, AD, GD, ELD and ETD, and a positive sign for the experience variables INT, IND and CMP. The other hypotheses predict positive signs for all betas.

what are called the residuals. These residuals are also known as estimates of the unknown

disturbances inherent to the data set. The estimator generating the set of values of the

parameters that is most popularly used among researchers is one that minimizes the sum of squared residuals and is called the ordinary least squares estimator. Consequently, in this paper a simple ‗ordinary least squares‘ model was used.

A major objective when choosing a functional form, or transforming the variables, was to create a model for which the error terms associated with each of the regressions were normally distributed about their mean. Using a Jarque-Bera measure of normality the errors were tested for normality. First of all, the regression for age diversity initially showed non-normal distributed residuals. Using a log-log functional form here resolved the issue. Secondly, the regression for gender diversity also showed non-normal residuals. However, due to the lack of variation in the sample in terms of gender diversity using a different functional form was not an option and it will therefore not be included in the regression analyses. The other six regressions showed normal distributions of their residuals. The Jarque-Bera values are included in the regression outputs that can be found in Appendix Five.

These error terms are further assumed to have a uniform variance (the variance measures the uncertainty in the regression model) or to be homoskedastic. In this cross-sectional data set there is no immediate reason to believe the variances would change throughout the sample (as one might predict in a time-series, or when observations are from data that is organised in a specific manner). In other words heteroskedasticity, or non-uniform variances, was not a main concern. Nevertheless, it was tested for using White‘s test for heteroskedasticity.5

Though it was expected that serial correlation would not be an issue, as in this model there is no variable which provokes specific ordering of the observations as can be the case with time series, serial correlation was controlled for as a precautionary measure. Durbin-Watson statistic was used to measure the possible presence of serial correlation in the residuals. None of these statistics had a value significantly less that 2, therefore there was no evidence of positive serial correlation

5 _{White‘s test includes an estimator for the variances and covariances of the least square coefficient estimators}

8.

THE

RESULTS

8.1DESCRIPTIVE RESULTS

This part is meant to provide a first insight on the board characteristics found in this study. The investigated variables will be described and initial findings will be enlightened. Descriptive statistics and correlations for all variables are given in Appendix Two.

8.1.1. Index Dispersion

In Graph One the total amount of board members is portrayed along the y-axis and along the x-axis are the four different indexes (AEX=1, AMX =2, AScX=3, Local=4). One can see that the dataset is reasonably well divided between the four indexes. There were a few reasons for this particular set of board members coming about. First of all, all of the Dutch listed companies were looked at but only those that provided enough information on their board members were included in the dataset. The larger companies, which are commonly found in the largest index, usually provided more informative WebPages and also included more complete ‗personalia‘ (personal particulars) in their annual reports. This translated itself directly in there being more observations of board members in the largest Dutch index; AEX. Care was taken to ensure that the smaller indexes, AMX, AScX and Local funds, were sufficiently represented as well.

Graph One: Index Dispersion

A next challenge was to create a dataset that included as many different industry sectors as possible to facilitate industry comparisons as well as to avoid an industry bias to begin with.

8.1.2. Industry Descriptives

The 100 companies in this dataset were arranged into 7 appropriate industries. Please see Appendix Three for a complete description of the industry classification, and Appendix Four for the precise list of companies and the industry sectors they belong to.

In Table Two the 7 categories are broken down into smaller sectors for a more precise overview of the industry dispersion. Though clearly the Industrials sector has the largest amount of companies and board members, it is interesting to see that out of the subgroups the retail wholesale sector takes the lead with 14 companies in the sample. It is closely followed by the financial sector with 13 firms, which coincidentally also has the largest boards in terms of board members; 139 in total.

Table Two: Industry Dispersion

CIGS6_{-Code INDUSTRY SECTOR #FIRMS #BOARD MEMBERS}

10&15 Energy & Materials 9 88

Oil/Petroleum 1 6 Chemicals 4 43 Steel 4 39 20 Industrials 25 204 Construction 9 66 Electronics 9 75 Engineering 3 26 Logistics 4 37 25 Consumer Discretionary 17 131 Media 8 65 Services 9 66 30 Consumer Staples 21 160 Food/Beverages/Consumption 7 69 Retail/Wholesale 14 91 35 Health Care 4 32 Pharmaceuticals 40 Financials 13 139 Financial

45&50 IT & Telecommunication Services 11 67

IT 8 45

Telecommunications 3 22

Total 100 821

Table Three demonstrates a few interesting details in terms of demographics and professional experience. Nationality diversity (ND) seems to be the most apparent in the

Energy and Materials sector and lowest for the Telecommunications and IT sector. This is not

surprising as the latter is a sector where 6 of the 11 companies are local; specialized at catering the Dutch market. The Energy and Materials sector on the other hand is characterized by many large internationally active and sometimes world leading companies.

Table Three:Average Levels of Executive background diversity and professional experience per Industry

Industry Sector NDB _ADσ _GDB _ELDB _ETDB _INT% _IND% _CMP%

Energy and Materials 0.42 6.69 0.08 0.37 0.51 0.38 0.56 0.20 Industrials 0.16 4.94 0.04 0.17 0.25 0.23 0.56 0.36 Cons. Discretionary 0.20 3.46 0.05 0.11 0.28 0.25 0.60 0.40 Cons. Staples 0.17 4.28 0.00 0.13 0.32 0.32 0.58 0.29 Healthcare 0.18 6.63 0.00 0.21 0.44 0.65 0.50 0.00 Financials 0.16 6.62 0.00 0.20 0.30 0.34 0.65 0.24 Telecom. and IT 0.13 4.02 0.04 0.13 0.30 0.31 0.55 0.23

B_{-Blau index, σ -standard deviation, %-percentages.}

Again the Energy and Materials sector scores the highest age diversity (AD) and the Consumer

discretionary sector scores the lowest spread of ages. What may be even more interesting, apart

from the very apparent low levels of gender diversity (GD) for all sectors, is the complete lack of women in the Consumer Staples, Healthcare and Financials sectors. Though almost all companies have a lot of female employees in lower and mid-level company ranks, they seem to have the most trouble getting to the executive level of these sectors.

In terms of educational level diversity (ELD) boards with the most diversity are found in Energy and Materials sector whereas board members in the consumer discretionary sector have the least dissimilar educational levels. Educational type diversity (ETD) is interesting as all sectors score relatively highly with Industrials as the most homogeneous group. This is a quite reasonable finding as one would imagine the least amount of diversity to be found in the sector that requires very specific or specialized education.

Table Four shows the average board size, company size in terms of employees and net sales (which is later used as a measure of complexity) across all industry sectors.

Table Four: Average board size, company size and net sales

Industry Sector Board Size Employees (tth) Net Sales (mln)

Energy and Materials 3.44 6.72 49585.82 Industrials 2.9 3.37 3501.05 Cons. Discretionary 2.82 2.87 1827.77 Cons. Staples 2.67 2.53 7922.75 Healthcare 3 0.17 627.18 Financials 4.31 2.20 31192.71 Telecom. and IT 2.2 0.68 138067.38

tth= ten thousand, mln= million

It seems that on average most boards in all sectors have 3-4 executive members. The largest and smallest executive boards are found in the Financial and Telecommunications and IT sectors respectfully. The size of a board is apparently not a reflection of the company‘s size in terms of employees: Companies in the Energy and Materials sector with by far the largest amount of employees do not have a much larger board compared to the companies in the Healthcare sector with, in stark contrast, almost 40 times less employees. Neither does board size seem to signify the complexity of the company‘s environment: The companies with the most ‗complex‘ environments (highest net sales) are found in the Telecommunications and IT sector, which also happens to be the sector with the smallest boards. Thus larger boards are not by definition more capable of dealing with complex environments associated with more extensive company operations or with a larger company workforce.

8.1.3 Board Member Background Characteristics: Executives vs. Supervisory

In Table Five the background characteristics per board type are given. For most of these background characteristic variables there is a distinct trend of supervisory boards surpassing the executives in terms of diversity.

Table Five: Background Characteristics

EXECUTIVE BOARDS

Mean Max Min S.D. Exec. B. Nationality D.B _{0,19 0,78 0,00 0,27} Exec. B. Age D.σ _{5,38 21,28 0,00 3,69}

Exec. B. Gender D.B _{0,03 0,50 0,00 0,10} Exec. B. Educational Level D.B _{0,21 0,72 0,00 0,23} Exec. B. Educational Type D.B _{0,31 0,78 0,00 0,28}

SUPERVISORY BOARDS

Mean Max Min S.D. Sup. B. Nationality D.B _{0,26 0,84 0,00 0,29} Sup. B. Age D.σ _{6,16 17,17 0,00 2,78}

Sup. B. Gender D.B _{0,09 0,49 0,00 0,15} Sup. B. Educational Level D.B _{0,24 0,67 0,00 0,24} Sup. B. Educational Type D.B _{0,24 0,85 0,00 0,29} B _{-Blau index, σ -standard deviation}

Furthermore, with age diversity, there seems to be a smaller spread of ages in the executive boards than on the supervisory boards (5.38 vs. 6.16). Worth mentioning may be that the extremely high maximum standard deviation of 21.28 on the executive board is due to one women born in 1980; the daughter of the co-founder of the company (Ms. Mittal Bhatia, at Arcelor-Mittal), subsequently the youngest board member in this dataset. For the supervisory boards, the 17.17 maximum standard deviation was due to a gentleman at Pharming who was born in 1929, making him the oldest board member in this dataset.

The supervisory boards also score higher for gender diversity. Both groups score very poorly here, 0.09 for the supervisory boards and a mere 0.03 for the executives. In fact, of the total 43 women found in this dataset, only 6 of them took up seat on executive boards.

Graph Two: Educational Level Dispersion Graph Three: Educational Type Dispersion

For instance, there was information available on educational level for 430 members which can be found in Graph Two. 81% of them had a level 3 education; equivalent to a masters title. According to Graph Three, the most popular educational type was level 2: general economics, with 15.3% of the available 314 members. This was closely followed by educational types 9 (technical/engineering with 14.6%) and 3 (financial or accounting studies with 14%).

8.1.4 Board Member Professional Experience: Executives vs. Supervisory

As can be seen from Table Six the executive boards scored considerably higher levels than the supervisory boards for all three experience variables.

Table Six: Professional Experience

EXECUTIVE BOARDS

Mean Max Min S.D. Exec. B. International E.% _{0.31 1.00 0.00 0.37}

Exec. B. Industry E.% _{0.58 1.00 0.00 0.40}

Exec. B. Company E.% _{0.30 1.00 0.00 0.37}

SUPERVISORY BOARDS

Mean Max Min S.D. Sup. B. International E.% _{0.21 1.00 0.00 0.27}

Sup. B. Industry E.% _{0.23 1.00 0.00 0.27}

Sup. B. Company E.% _{0.03 0.71 0.00 0.09} %= percentages

For international experience diversity for example, the executive boards scored 31% compared to the supervisory boards that scored lower at 21%. In terms of prior experience in the industry the executives scored a staggering 58% versus a mere 23% for the supervisory members. Another extreme was found for company experience where the executives scored 30%, whilst only 3% of the supervisory board members had prior company experience. The first two

variables could lead one to believe that executives, when dealing with the day-to-day management of a company, are perhaps benefited more by a profounder knowledge of international markets and the industry it belongs to. In terms of company experience, on the one hand, it could be presumed that many executives have worked their way up the ‗corporate ladder‘ to get into the executive board. In this case it is not so surprising that they would score so highly. On the other hand, supervisory directors are chosen for their distinct role as, preferably, independent director; having no prior ties with the company would therefore make sense.

8.1.5 Complexity and CEO Dominance Descriptives

This table gives a preliminary descriptive look at the explanatory variables complexity and

CEO dominance.

Table Seven: Descriptives for complexity and CEO dominance

# NDB _ADσ _GDB _ELDB _ETDB _INT% _IND% _CMP%

Larger Complexity 42 0.33 5.33 0.04 0.42 0.20 0.49 0.68 0.24 Smaller Complexity 42 0.09 5.19 0.02 0.26 0.14 0.22 0.56 0.35 CEO 2003-2006 49 0.20 5.44 0.03 0.29 0.18 0.27 0.53 0.28 CEO before 2003 51 0.18 5.31 0.02 0.34 0.16 0.34 0.63 0.31

B_{-Blau index, σ -standard deviation, %-percentages.}

First of all, the sample of companies was divided into two equal groups according to their environmental complexity. For all but one experience variable it seems that the more complex group of companies does indeed have more diversity in terms of background characteristics and more international and industry experience. This already gives some support for the first hypothesis which stipulated that more complex environments will need boards with more background characteristic diversity and more professional experience to cope.

Secondly, the 100 CEOs were split up to make a distinction between longer and shorter tenured CEOs. This was done according to a division in years that ensured both groups would have about the same number of observations. On average, boards with longer tenured CEOs show less diversity for four of the five background characteristics (all but

educational type diversity) and score higher for all three experience levels. Here too it seems that

have a negative influence on their board‘s diversity (hence their members should look more alike in terms of background characteristics than the boards of newly appointed CEO‘s) along with 3b which predicted a positive relationship between longer tenured CEOs and the levels of professional experience held by their board members.

8.2ANALYTICAL RESULTS

A summary of the regression results is shown in Table Eight. Each column denotes the separate regression results per dependent variable: ND= nationality diversity, AD= Age diversity, ELD= educational level diversity, ETD= educational type diversity, INT= international experience, IND= industry experience and CMP= company experience. As explained before, gender diversity was not included. The more complete regressions can be found separately in Appendix Five.

Hypothesis one predicted a positive relationship between a firm‘s complexity and its