IB&M Supervisor: dr. K. van Veen Co-assessor: dr. P. Rao Sahib IE&B Supervisor: dr. P. Rao Sahib Co-assessor: dr. K. van Veen

(1)

Master thesis International Business & Management and

International Economics & Business

University of Groningen

Faculty of Economics and Business

The effect of shareholders on the

interlocking directorate network in

Europe

Annefleur Clercx

a.l.c.clercx@student.rug.nl

S2349027

June 20

th

2017

IB&M

Supervisor: dr. K. van Veen

Co-assessor: dr. P. Rao Sahib

IE&B

(2)

Abstract

This study analyses the development of the interlocking directorates network in European countries over 2013, 2014, and 2015, and what determines these changes. Additionally, the influence of several ownership structures (i.e. concentrated-,

institutional investor-, insider-, family-, state-, and foreign ownership) on the number of interlocking directorates is examined. Data with regard to interlocks has been collected for 2013, 2014, and 2015 including 226 firms in 10 European countries (i.e. Austria, Belgium, Denmark, Finland, France, Germany, the Netherlands, Spain, Sweden and the UK). Ownership data has been collected for 2015 on these same companies. Expected is that the network of interlocks has decreased on a national level and increased on an international level. The results contradict expectations, showing a decrease on international level. Large differences in the average number of interlocks per company can be found between countries. With regard to the ownership study, it is found that in Denmark concentrated ownership is positively related to the number of interlocks. Besides that, foreign ownership has a positive influence on the number of interlocks in Finland, France and Germany. In Austria and Spain this influence is negative.

(4)

1. Introduction

Have you ever looked into how many links there are among boards of directors of companies, and whether or not members are potentially overlapping? Gerhard Cromme, a German businessman, is said to have been one of the most influential German leaders (Stern, 2003). Having served on the boards of Allianz SE, BNP Paribas, E.ON AG, Hochtief AG, Lufthansa AG, Saint-Gobain, Siemens, Suez AG, Thales AG and Volkswagen there can be said he has occupied a significant number of board positions (Stern, 2003). Another person who serves on many boards is the Swedish director Jacob Wallenberg, he has been and still is on the boards of Investor AB, Ericsson, ABB Ltd, SAS, Atla Copco AB, The Coca-Cola Company, Knut and Alice Wallenberg Foundation, Stockholm School of Economics and Skandinaviska Enskilda Banken (Investorab, 2017). These overlapping board memberships establish connections between companies. Another example shows the links between

corporations as AXA, Schneider Electric and BNP Paribas share 13 directors, being highly connected with each other in that way (Maclean, Harvey, & Press, 2006). There are more links among people than might be thought initially. These are characterised as interlocking directorates; “links between corporations that occur when a director who is affiliated with one corporation sits simultaneously on the board of directors of another corporation” (van Veen & Kratzer, 2011, p. 2).

A number of studies have showed the existing interlocking directorate network among countries in Europe (Fennema, 1982; Fennema & Schijf, 1985; van Veen & Kratzer, 2011; Heemskerk, 2011). Investigating interlocking directorates of a company can help to understand their company performance, company governance, and actions of directors (Ma, 2014). Moreover, a number of studies have analysed whether or not diverse ownership structures are parallel to the density of an interlock network (Windolf & Beyer, 1996; Scott, 1997). These studies were based on Germany, UK, France, Belgium and Italy and significant relations were found (e.g. Germany was found to have a concentrated ownership structure and a dense network of interlocks (Windolf & Beyer, 1996). Furthermore, economic theory respects ownership structures among companies as one of the main determinants of understanding corporate behaviour and governance (Pedersen & Thomsen, 1997). The contribution of this thesis is the extension of the interlocking directorate network analysis in European countries with multiple years and why this has developed in that way. This is not available yet and the findings contradict expectations. With regard to the previous research on ownership and interlocks, this has been performed on interlocks within country borders. Shown was that a high portion of the interlock network reflects ownership structures and enhances power of shareholders. Unfortunately, these studies have been performed a long time ago (i.e. up to the 1990s and 1992) and are based on a limited number of countries. This thesis uses a larger group of

(5)

illustrate which type of shareholder group leads to more interlocks, and in that way which type of shareholder potentially drives up firm performance.

The interlocking directorate network will be analysed through a unique database containing data on the boards of directors of 226 companies in 10 European countries during 2013, 2014 and 2015. With this database an analysis of the interlocking directorates can be done. Data of diverse ownership structures (i.e. concentrated-, institutional investor-, insider-, family-, state-, and foreign ownership) will be hand-collected using of Orbis.

(6)

2. Theory and hypotheses

2.1 Interlocking directorates

Previously performed studies show that there is a growing international corporate elite and that the network coherence increases in Europe (Carroll, Fennema, & Heemskerk, 2010; Heemskerk, 2011; Heemskerk, 2013, van Veen & Kratzer, 2011). Fennema (1982) was the first to analyse the national and especially international network of interlocking directors. More recently, found by van Veen and Kratzer in 2011, up to 2006 Europe had an interlocking directorate network spanning fifteen countries, 362 European companies, and 684 interlocking directorates, with a variation in

interlocking directorates among countries. Also discovered in this same study was that in 2006 just under 40% of all the board interlocks among the largest companies in European countries crossed national borders, and in this way, a corporate elite

network was created. Heemskerk (2013) demonstrates that the building blocks for the social field of the European corporate elite consist of board interlocks as an elite formation- and communication device. Additionally, if a firm shares more than one director with another firm, the strength of their tie is characterised as a durable and strong relationship (Heemskerk, 2011). “The boards of directors at large European companies overlap with each other to a sizable extent both within and across national borders” (Heemskerk et al. 2013, p. 1). Also, Heemskerk et al. (2013) point out that in the interlocking directorates network in Europe from 2005-2010 there was an

indication of homophily, meaning firms which are well connected to other firms prefer to connect with firms having the same level of connectedness. The network is not based anymore on the small corporate elite, which used to run the show, a growing number of individual linkers have an increasing influence on interlocking directorates (Heemskerk, Fennema, & Carroll, 2016). A shift is suggested from fewer national linkages to more transnational linkages, thus crossing country borders

(Heemskerk et al., 2016). How and due to which characteristics has the network previously identified by van Veen and Kratzer (2011) developed over the years? Previous research initiates that this has become more dispersed across national borders and less within country borders (Heemskerk, 2011; Heemskerk, et al., 2013; Heemskerk, et al., 2016).

Before taking into account determinants of interlocking directorates, benefits for companies that are interlocked are examined. Many aspects of interlocking

(7)

transfers through interlocks, valuable information is obtained (Haunschild & Beckman, 1998). An interlock is seen as a useful channel in that different types of influences transfer through this such as control of others and general co-optation (Haunschild & Beckman, 1998). Useem (1984) proposed that because of interlocks managers are able to get insights in the general business environment and the latest business methods. Worthy knowledge of how other companies do things can be gained. The ability to get insights in the business environment can reduce

environmental uncertainty faced by organizations (Shropshire, 2010). This due to the experience directors have had on other boards that can be applied in situations of uncertainty at other companies (Shropshire, 2010). With regard to international interlocks, different nationalities on a board can be seen as a strategic asset (van Veen & Elbertsen, 2008). This can aid the firm when developing and/or expanding

internationally (van Veen & Elbertsen, 2008). When taking into account the board of directors, diversity is seen as a positive asset since this signs at a large and wide base of wisdom (van der Walt & Ingley, 2003). This stimulates working together

successfully and in that way makes a contribution that is valuable since it provides different perspectives on strategic issues. Diverse nationalities on a board enhance creativity, the supply of novel ideas, performance, quality of decision-making and the availability of knowledge (van Veen et al., 2013). Contrary to the positive aspects, diverse nationalities in boards can also be a possible disadvantage with regard to collective performance (Erhardt, Werbel, & Shrader, 2003). This due to the fact that greater effort and time is needed to come to decisions when there are many different nationalities (Erhardt et al., 2003) Lastly, but yet very important, firms with boards that are well connected have a high return on assets ratio (Larcker, So, & Wany, 2013). Interlocked directors are found to be significant drivers of improving firm performance (Pombo & Gutierrez, 2011).

Having investigated the benefits interlocks can bring companies; determinants starting with country-level determinants followed by firm-level determinants are discussed. Interlocking directorates are influenced by multiple country-level factors, which in turn determine the density of the interlocking directorates network. The length of EU membership of a country is found to positively influence the centrality of a country in the network (van Veen & Kratzer, 2011). Additionally, differences in governance regimes between countries have shown to have a significant influence of the density of the interlocks between firms (van Veen & Kratzer, 2011). This involves the

varieties of capitalism approach of countries. Firms in coordinated market economies (CMEs) engage in strategic interactions with one another (Hall & Soskice, 2001). Companies in liberal market economies (LMEs) interact with one another through competitive relations (Hall & Soskice, 2001). Companies in CMEs have denser interlocking directorate networks than those in LMEs (van Veen & Kratzer, 2011). As can be seen in table 1 on page 8 Austria, Belgium, Denmark, Finland, France,

(8)

score of 0.57, which could matter for the extent to which they are interlocked. The UK is expected to have a less dense network of interlocking directorates (score of 0.07). These differences in governance regimes can also have more specific effects, influencing the recruitment processes of board members in countries. There are countries in which companies tend to be more open to outsiders and developing a career within multiple companies, whereas others are less prone to these

developments (van Veen & Marsman, 2008). Lastly, competition occurring within the economic markets of countries drives interlocks (Ma, 2017). To reduce competition directors take on multiple board positions in order to instead of compete with one another, coordinate actions beneficial for both (Ma, 2017).

Table 1: Country scores on LME-CME index (0=LME, 1=CME)

Country LME-CME index

Austria 1 Belgium 0.74 Denmark 0.7 Finland 0.72 France 0.69 Germany 0.95 The Netherlands 0.66 Spain 0.57 Sweden 0.69 UK 0.07

Source: (Hall & Gingerich, 2004)

Narrowing down the characteristics that can be of influence on interlocking

directorates from a country-level to a firm level, interesting determinants have been found in previous studies (Windolf & Beyer, 1996; Bizjak, Lemmon, & Whitby, 2009; Nielsen & Nielsen, 2010; Heemskerk, 2011; Ma, 2017). On a firm level, nationality diversity within boards has caused an expansion of the interlocking directorate network across national borders (Heemskerk, 2011). This due to a tendency to recruit board members who are different but not too different from their existing board members (Nielsen & Nielsen, 2010). Moreover, between firms there is found to be a limited pool of board candidates, which does not make it abnormal for a member of a certain board to sit on the board of more than one firm, creating an interlock (Bizjak et al., 2009). These candidates are found to occupy multiple board positions due to their political, legal and business experience found worthy for companies (Bizjak et al., 2009). As mentioned earlier, a benefit of interlocks is

(9)

meetings, where the shareholders are deciding on who to select as potential new members. Through this preferences of shareholders can be included. This final determinant of interlocking directorates will be analysed more extensively in this thesis. This because shareholders of companies are found to have a significant influence on multiple aspects of the company through control and collaborative aspects (Sundaramurthy & Lewis, 2003).

Before doing so, the first hypothesis is presented with regard to the expected

developments of the interlocking directorate network. Combining these expectations with the interesting aspects interlocks can present to companies there is reason to investigate this. Multiple determinants at country-level and firm level are of influence, which will be taken into account when discussing the actual developments the

network has undergone. The analysis has led to the first hypothesis, which is as follows:

Hypothesis 1: The number of interlocking directorates has decreased on a national level and increased on an international level over the years 2013-2015.

The following section discusses the influence shareholders, and so different types of ownership, have on the creation of interlocking directorates.

2.2 Shareholders and the interlocking directorate network

The ownership analysis in this thesis will be performed on a firm- and country level. It will be taken into account what the influence of the ownership structure is on the number of interlocking directorates per company in a certain country. Ownership involves shareholders who own at least a part of the companies’ stock and have rights and obligations (Investopedia, 2017). Prior to examining the influences of

(10)

According to Thomsen & Conyon (2012) the main feature of ownership structure is the ownership concentration, which implies ownership by one or a few large owners. A company with concentrated ownership can be defined as “one in which a single shareholder owns a large portion of total equity” (Ragazzi, 1981, p. 263). In contrast, dispersed ownership occurs when the shares of a company are more widely held, it can be defined as “one whose shares are owned by a large numbers of individuals none of whom is in a position to obtain direct or indirect benefits per share greater than those available to other shareholders” (Ragazzi, 1981, pp. 262-263). Specific types of ownership are all dispersed or concentrated in nature.

The first of many types of shareholders is the institutional investor defined as “a nonbank person or organization that trades securities in large enough share quantities or dollar amounts that it qualifies preferential treatment and lower commissions” (Investopedia, 2017). This ownership type occurs when an (large) amount of the company’ stock is owned by for example mutual funds, pension funds, sovereign wealth funds or hedge funds, and is found to be positively related to firm performance (Boyd & Solarino, 2016). Institutional investors are found to be of influence on the firm’s choices since they are not constrained by their business relationships (Brickley et al., 1988). Next to this, insider ownership occurs when an insider in a company, for example the CEO, a director, board member or officer, holds part of the shares; this is found to have a positive influence on firm performance (Brickley et al., 1988).

Family ownership is found to be one of the most occurring ownership types (La Porta, Lopez-de-Silanes, & Shleifer, 1999). This type of ownership is different from other types since family members have high involvement in the company. The family potentially offers a buffer in turbulent times, it is possible that minority shareholders are exploited with regard to private benefits (Boyd & Solarino, 2016). Finally, state owned enterprises are owned by the state, which can be favourable since it can provide access to resources or legitimacy, however social and political goals can take preference (Boyd & Solarino, 2016). These diverse types of shareholders will be included again in a later stadium of this thesis. The following section will first get to the core of the influences shareholders have on companies in general.

Ownership structure is of key importance in corporate governance since it determines who has the ultimate decision-making power in a company (Kumar & Zattoni, 2015). A main role of the board of directors is to serve in the best interest of the

(11)

shareholders have incentives to monitor for managerial opportunism (Attig, El Ghoul, Guedhami, & Rizeanu, 2013). With regard to the strategy of companies, shareholders can have an influence on that through the process of restructuring (Connelly,

Hoskisson, Tihanyi, & Certo, 2010). This results in higher firm performance through for example cutting costs, separate unprofitable business or adapt a company in a way that it is prone to changing market conditions (Connelly et al., 2010). Next to that, shareholders can have an impact on the acquisition strategies firms exhibit, leading to either less risky acquisitions or more risk taking through acquisitions (Connelly et al., 2010). Lastly, especially shareholders with a larger holding in a company are

interested in the allocation of resources. If necessary they provide resources required, delivering the firm a wider base of resources (Connelly et al., 2010).

The specific influence shareholders have on the selection of new board members and thus who takes place in the board of directors is analysed in closer detail in the following section.

Shareholders are part of a concentrated or dispersed ownership structure. These have different influences on companies and likewise the selection of new board members (Windolf & Beyer, 1996). If ownership is dispersed many individual investors invest in the company. However the chance that they can make a substantial difference in the selection of the board of directors, and in that way establish an interlock is likely to be small. This because there are many shareholders with different ideas about directors (Windolf & Beyer, 1996). This will result in indecisive situations since the ability to take collective action is limited. Contrary to dispersed ownership;

concentrated ownership means that only one or multiple large shareholders exercise effective control over the company (Windolf & Beyer, 1996). Shareholders that have a significant part of the stock have sufficient influence and control to steer in their preferred direction. Taking into account benefits interlocking directorates can bring companies; there can be a focus on recruiting board members from others boards in order to derive these advantages. Whether this is the case and a concentrated ownership structure increases the number of interlocks will be investigated through the second hypothesis:

Hypothesis 2: Concentrated ownership is positively related to the total number of interlocking directorates.

(12)

concentrated ownership (as can be seen in table 1 on page 8 Austria has a LME-CME index of 1). In contrast, the UK, with an LME-CME index of 0.07, is characterised by dispersed ownership. Taking into account the extent to which a country is

characterised as a LME or CME allows to analyse which degree of concentration of ownership is expected to be of influence. Austria is expected to have a positive relation between concentrated ownership and the number of interlocking directorates. In the UK this relationship will be expected to be negative as the companies are expected to have dispersed ownership structures. The remaining countries all have a score that points at a coordinated market economy. Taking this into account a positive relation is expected between concentrated ownership and the number of interlocking directorates in Belgium, Denmark, Finland, France, Germany, the Netherlands, Spain and Sweden.

Referring back to the specific types of ownership, these are found to have different influences on the selection of new board members. Firstly, institutional investors are capable of influencing the choices made in a company, and thus the decisions made with regard to the board of directors. Additionally, institutional investors are found to be one of the most active types of shareholders, often seeking to change the structure and memberships of boards and top management teams (Dai, 2007). Connelly et al. (2010) studied that institutional investors are in favour of internationalization.

Combining the findings that these investors are often aiming at changing the board of directors, favour internationalization, and can influence the choices made, it is

expected that the presence of an institutional investor will drive up the number of international interlocking directorates, and so the total number of interlocks. Therefore, the third hypothesis is:

Hypothesis 3: Institutional investor ownership is positively related to the total number of interlocking directorates.

Insider ownership, stock owned by people internal to the firm, can result in less risk taking (Morck, Shleifer, & Vishny, 1988). Insiders are rooted within the company due to their ownership shares. They are not willing to take on high risks potentially at the cost of their profit. Even though interlocking directorates are found to reduce risk, potential risks can arise from them as well. Taking on a new board member means giving access to information internal to a company to an outside director. This can be seen as risky and potentially not favoured by inside shareholders. An external director gets access to company specific, precious, information. The expectation therefore is that the significant presence of insider owners will not be in favour of interlocking directorates, leading to our fourth hypothesis:

(13)

Studies performed on family ownership have shown that families tend to keep their family within the business for a rather long time (Boyd & Solarino, 2016).

Furthermore, due to their concern about long-continuity of the company, outsiders can be a potential threat and risk to the firm and are preferred to be kept at a distance (Connelly et al., 2010). This can imply that the tendency to interlock is rather low with family ownership leading to the fifth hypothesis:

Hypothesis 5: Family ownership is negatively related to the total number of interlocking directorates.

The final type of ownership analysed in this thesis is that of state owned companies. State owners tend to favour innovation within the company and are not that likely to expand their business. The government can provide a wide array of resources and the potential social and political goals can take over the company interests. It is therefore questionable whether interests will lie at who will fill the positions on the board of directors. The state owner itself can partly fulfil the gains otherwise derived through interlocking directorates. The tendency of state owners seems to keep processes rather internalized, not likely to take on outside directors. Therefore the sixth hypothesis is: Hypothesis 6: State ownership is negatively related to the total number of interlocking directorates.

The last aspect this thesis will take into account is the nationality of shareholders, implying the foreigners that hold shares of a company. Research performed by Greenaway et al. (2014) has found that companies partly owned by foreign shareholders and partly by domestic investors tend to have a higher corporate performance than those that are fully domestic owned. This because of the

combination of knowledge on the legal and political environment in the home country from the domestic owners and a contribution with regard to international networking, managerial skills and better corporate governance from foreign owners (Greenaway et al., 2014). Thus, foreign ownership can lead to higher corporate performance. Besides this, foreign ownership can increase the financial condition of a firm (Wang & Wang, 2015). More specifically, firms with parts being owned by foreigners are less

financially restrained due to access to international financial markets, which advances their financial condition (Wang & Wang, 2015). Because the financial condition is improved, wages, employment and output can increase due to higher sales and market shares (Wang & Wang, 2015). Ongore (2011) has also found that performance

increases with foreign ownership due to the ability foreign owners have to monitor managers. In that way activities and behaviours undermining the wealth of the firm are prevented and the transfer of management practices and new technologies occurs. Noticing the potential increase in corporate – and financial performance when part of the owners have a nationality different from the country of origin of the company gives reason to investigate this. When there are many international owners the

(14)

performance can be driven up. People are tending to recruit people like themselves; internationals will therefore see reason to recruit international board members. Combining this tendency, benefits derived from interlocks, and the ability of owners to exercise control over selection of directors can result in more international

interlocking directorates. Our seventh and final hypothesis therefore is: Hypothesis 7: Foreign ownership is positively related to the total number of interlocking directorates.

Country level differences with regard to foreign owners can influence the extent to which foreign investments are allowed into a county. Certain countries are very restrictive with regard to foreigners entering a country and investing in companies (OECD, 2017). Restrictions influence a country’ capability to attract foreign direct investment (OECD, 2015). Restrictiveness will result in fewer investments in companies in a certain country and in that way less foreign owners, as investments will be more domestically.Table 2 below displays the degree of foreign direct investment restrictiveness, where the closer to 1 implies the more restricted that country is to foreign investments. The numbers would imply that in Austria the lowest levels of foreign owners are to be found since they have the highest level of

restrictiveness. There would be expected to be a negative relationship between foreign ownership and the number of interlocking directorates in Austria. Contrary, the

Netherlands will be expected to have high levels of foreign ownership, as they are extremely open to foreign investments. This means foreign ownership is expected to relate positively to the number of interlocking directorates in the Netherlands.

Table 2: Foreign direct investment restrictiveness (0=open, 1=closed)

(15)

4. Conceptual framework

The conceptual framework presented below gives a clear overview of the hypotheses developed in section 3 and what will be tested in this research. The hypotheses will be tested on firm level, controlling for the countries of origin of the companies in the sample. Hypothesis 1 is not displayed in the conceptual framework due to an expectation with regard to the overall interlocking directorate network and not a relationship between the network and a dependent variable. With the other hypotheses it is expected that concentrated ownership, institutional investor ownership and

foreign ownership will be positively related to the number of interlocking directorates. Insider ownership, family ownership and state ownership are predicted to be

negatively related to the number of interlocking directorates.

(16)

5. Data & methods

5.1 Data and sample selection

To test the hypotheses a sample of sizable companies from ten European countries (Austria, Belgium, Denmark, Germany, Finland, France, the Netherlands, Sweden, Spain and the UK) has been constructed. The choice for these countries was made based on their EU entry, all having entered in the first waves of entry (van Veen & Kratzer, 2011). The remaining five countries that have entered the EU in the first waves (Greece, Ireland, Italy, Luxembourg, and Portugal) are unfortunately not included due to a lack of data collected since the data collection is a rather labour-intensive process. The companies were chosen based on their listing in the stock market in their country of origin. Over the years 2013-2015 a sample of 226

companies (a list of companies per country can be found in appendix 1) will be used with regard to the interlocking directorate network analysis. The choice for these three years will present the most recent developments with regard to the interlock network and displays changes. Firms that have been delisted or merged with other firms in one of the years in our analysis will be excluded from the sample. This is due to the impossibility to gather data for these companies. An overview of the firms that have been left out can be found in appendix 2. Information on the board structures was hand-collected from the annual reports the companies have publicly available. This includes directors from one-tier board structures as well as two-tier board structures. To measure the diverse types of ownership taken into account in the hypotheses data was hand-collected from Orbis per company in the sample. Information on the nationalities of stockowners was present in Orbis; if this was not available a

secondary source (i.e. Google) was used to identify the nationality of an owner. The data on ownership is only available for the year 2015, therefore hypotheses 2-7 will be analysed using data solely from the year 2015.

5.2 Measures

5.2.1 Dependent variable

The dependent variable in the hypotheses focuses on the total number of interlocking directorates among companies, within and across national boarders. Consequently, the dependent variable is “the total number of interlocking directorates”, including

(17)

includes executives as well as non-executive directors. This data is derived through the European Top Managers Project database.

5.2.2 Independent variables

Concentrated ownership. To test the expected relationship in hypothesis 2 it is essential to measure the ownership concentration of a company. This is done through determining if there is an ultimate owner or not, which has shown to be a proper measure. For the definition of ultimate ownership we use the following: a firm or person that has at least 20% of stock of a company (Faccio & Lang, 2002; Tsao & Chen, 2012; Carney & Child, 2013). The distinction will be made between the presence and absence of an ultimate owner through the use of a dummy variable which takes the value 1 if present and the value 0 if absent. This data is derived through Orbis.

The following four variables (institutional investor -, insider-, family-, and state ownership) are included as present if the share they hold is sizeable enough, this threshold is set at a 5% share of company stock (i.e. influence and control is expected to be of potential significance on or after the 5% threshold) (Carney & Child, 2013). Institutional investor ownership. Multiple investors are categorized under the

definition of institutional investors ownership. The presence of an institutional investor will be taken into account if there is one of the following that has a 5% ownership share: financial institution (e.g. bank or insurance company), mutual-, pension- or hedge- fund (Boyd & Solarino, 2016). This variable will be used through a dummy variable, taking the value of 1 when an institutional investor is present and the value of 0 if absent. This data is derived through Orbis.

Insider ownership. This variable will be operationalized as the stock that is owned by people internal to the firm (e.g. directors, CEOs or board members). The variable will be used through a dummy variable, taking the value of 1 when an inside owner is present and the value of 0 if absent. This data is derived through Orbis.

Family ownership. The family ownership is measured as the stock held by the family members of the company with the 5% threshold. The variable will be used through a dummy variable, taking the value of 1 when family membership is present and the value of 0 if absent. This data is derived through Orbis.

(18)

Foreign ownership. To test hypothesis 7 the dependent variable foreign ownership will be taken into account. The operationalization of this variable is done in two different ways:

- (1) The number of foreign owners: this is measured as the percentage of owners that have a nationality different from those in the home country of the company. This implies the higher this percentage, the larger the number of foreign owners

- (2) The percentage of stock held by foreigners: measured as the percentage of stock the foreign owners hold compared to the total stock. The higher this percentage, the more stock held by foreign owners.

This data is derived through Orbis.

5.2.3 Control variables

The inclusion of control variables allows getting more precise results of the observed relationships; they are intended to confirm the relationship between the independent and dependent variables. In this thesis we control for the several variables that can be of potential influence:

Firm size. Firm size is controlled for since this can be of influence on the diversity in nationalities in top management teams as larger companies are more international (Greve, Nielsen, & Ruigrok, 2009). Firm size will be measured as the number of employees per firm per country in 2015 (Nielsen, 2010).

Board size. It is found that larger boards tend to fill vacancies on a more frequent basis, and so have more room for new board members (Ruigrok & Greve, 2008). The firms in the database have board sizes that differ in number quite a lot (e.g. ranging from 6 to 31 directors per board). Thus, the second control variable will be included as number of directors in the board of each firm in 2015. This will be included as the board size per firm per country in 2015.

Country of origin. This control variable will be taken into account in order to control for country effects since the study performed in this thesis is cross-national. This will be done by using the LME-CME index score developed by Hall & Gingerich (2004), where differences in governance regimes and institutional settlements are taken into account (Greve et al., 2009). This score will take values 0-1 where the closer to 1 is characterised as more CME oriented. These country specific values are derived from the research performed by Hall & Gingerich (2004).

5.3 Econometric methodology

(19)

In order to test hypotheses 2-7 the least squares method will be used for regression. Appendix 3 gives an overview of all the variables that will be used including their measurements and indexes. The following estimation equation will be used:

(1) ILDi = β0 +β1OWN_CONi + β2OWN_INSINVi + β3OWN_INSi +

β4OWN_FAMi + β5OWN_SOi + β6OWN_FOR1i + β7OWN_FOR2i +

β8FIRM_SIZEi + β9BOARD_SIZEi + β10COOi + e

Where ILDi is the total number of interlocking directorates of firm i; the independent variables are

concentrated ownership of firm i: β1OWN_CONi; institutional investor ownership of firm i:

β2OWN_INSINVi; insider ownership of firm i β3OWN_INSi; family ownership of firm i: β4OWN_FAΜi;

state ownership of firm i: β5OWN_SOi; foreign ownership (1) of firm i: β6OWN_FOR1i, foreign

ownership (2) of firm i: β7OWN_FOR2i; the control variables are firm size of firm i: β8FIRM_SIZEi;

board size of firm i: β9BOARD_SIZΕi; country of origin of firm i: β10COOi. β0 is the intercept, e is the

error term.

In order to test the hypotheses two steps will be taken. To study the developments proposed in hypothesis 1 the number of interlocking directorates will be examined for the years 2013, 2014 and 2015. After this, regressions are executed in order to test hypotheses 2-7.

5.4 Verification of the variables

The empirical study in this thesis will use an Ordinary Least Squares (OLS)

regression in order to determine the potential influence the independent variables have on the dependent variable. Before doing so, the trustworthiness of the variables and dataset have to be examined in order to detect unusual data, normality,

homoscedasticity, and multicollinearity.

In order to detect observations that are different from all the others potential outliers can be detected. Outliers are detected through the use scatter plots. The scatter plots do not show extreme observations that should be taken into account during the regression.

Secondly, checking for normality of residuals will ensure that outcomes with regard to the p-values for F-tests and t-tests will be valid. With OLS regression this implies that the residuals (errors) should be normally distributed. This is done through the “Shapiro-Wilk test for normal data”, the results of this are presented in appendix 4 and shows that the residuals are distributed normally.

(20)

Chi-score is rather high (12.47), in order to control for the heteroskedasticity in the regressions robust standard errors will be used.

Finally, to check for multicollinearity the variation inflation factor (VIF) is used. When multicollinearity occurs variables tend to move together in systematic ways, not explaining enough about individual effects of the explanatory variables (Hall,

Griffiths, & Lim, 2011). Multicollinearity can lead to probability intervals that are less reliable (Adkins & Hall, 2011). A rule of thumb for the VIF value is that it may not exceed 10 and the 1/VIF may not be lower than 0.1. Appendix 6 shows the VIF values for the variables used. All VIF values are far below 10 with a mean of 1.26, all 1/VIF values are above 0.1, thus there are no problems related to multicollinearity. Furthermore, to detect potential correlation problems a correlation matrix is used. If there exists a significant correlation that is >70% the variables tend to be too

(21)

6. Empirical results

6.1 Interlocking directorate network

The interlocking directorate network can be analysed on country- and firm-level with specific characteristics influencing at both levels. When examining the data on the boards of directors and the interlocks between firms, substantial differences are found in the extent to which interlocks are present.

Table 3 below presents the network statistics for companies within their country border for the year 2015. The number of companies per country, executive and non-executive positions of all the companies, the number of interlocking directorates between the firms in the country, and the mean number of interlocking directorates per company are presented. The number of executive and non-executive directors, and the number interlocking directorates vary rather substantially, which is because some countries have more companies than others. Taking into account the mean number of interlocking directorates per company significant differences can be found varying from a mean of just above 1 interlocking directorate per company to a mean of 6.1 interlocking directorates per company.

Table 3: Network statistics – interlocking directorates within countries (2015)

Country Number of companies Executive & non-executive positions Interlocking directorates Mean per company Austria 17 274 36 2.1 Belgium 16 213 18 1.1 Denmark 26 344 46 1.8 Finland 20 341 38 1.9 France 30 481 120 4 Germany 28 676 170 6.1 Netherlands 20 253 32 1.6 Spain 23 294 30 1.3 Sweden 24 502 88 3.7 UK 25 324 48 1.9

Table 4 on page 22 presents the networks statistics of the interlocking directorates on national and international level, thus within and across country borders. In total 226 companies have a total number of 821 national and international interlocks. However, as was expected these are not evenly distributed across countries since companies in some countries tend to interlock more than in other countries. The number of total interlocks per country differs severely from 40 interlocks to 196 interlocks.

(22)

vary severely. Despite these logics, the mean number of interlocks per company in each country is of noteworthy difference as well, noticing higher means in countries where the total number of interlocks is higher and lower means where fewer

interlocking directorates occur. Comparing the national to the international interlocks, countries show differences in the ratio national to international interlocks. The

Netherlands for example has over 50% of international interlocking directorates whereas their national connectedness is average compared to the other countries. This would imply that companies in the Netherlands favour international board members. The same holds for Belgium, who has almost 60% of international interlocks. Consequently their national mean of interlocks is lower (1.1) than the international mean of interlocks (1.6). In contrast, Germany has an extremely low mean score of international interlocks per company (0.9) whereas their mean per company on a national level is rather high (6.1).

Table 4: Network statistics – interlocking directorates across country borders (2015)

Country Number of companies Total interlocking directorates (mean per company) National interlocking directorates International Interlocking directorates (mean per company) % International interlocking directorates Austria 17 40 (2.4) 36 4 (0.3) 10 Belgium 16 43 (2.7) 18 25 (1.6) 58.1 Denmark 26 71 (2.7) 46 25 (0.96) 35.2 Finland 20 59 (3) 38 21 (1.1) 35.6 France 30 160 (5.3) 120 40 (1.3) 25 Germany 28 196 (7) 170 26 (0.9) 13.3 Netherlands 20 69 (3.5) 32 37 (1.9) 53.6 Spain 23 36 (1.6) 30 6 (0.3) 16.7 Sweden 24 131 (5.5) 88 43 (1.8) 32.8 UK 25 82 (3.3) 48 34 (1.4) 41.5

In general it can be seen that on a national and international level France, Germany and Sweden appear to be the most interlocked whereas Austria, Belgium, and

Denmark, Finland, the Netherlands and the UK have a significantly lower amount of national and international interlocks. Spain is at the bottom with the lowest mean number of interlocks. On a national level the mean numbers of interlocks per

company are in comparable order where Germany, France and Sweden belong to the top, Austria, Belgium, Denmark, Finland, the Netherlands and the UK are in the middle and Spain is again the country with the lowest mean number of interlocks. The international mean numbers of interlocks changes this order. Belgium, the

Netherlands and Sweden have the highest means. The middle sector includes

(23)

in total, national and international interlocking directorates per company. Figure 3 on page 24 gives a visual representation of the interlocking directorates network on a transnational level in 2015. The visuals in appendix 8 present the interlocks between countries in 2015, differences in countries can be seen clearly from these.

Figure 2: Interlocks per country (2015)

0,00 1,00 2,00 3,00 4,00 5,00 6,00 7,00 8,00

Interlocks per country as mean per company

(24)

Figure 3: Visual representation transnational interlocking directorate network on a firm level (2015)

Note:

a) Each nodes represents a firm

b) A link between two nodes means at least 1 national or international link is present, the thicker the line the more connections between two firms

With regard to hypothesis 1 it is expected that the network has decreased on a national level and increased on an international level. Shown in table 5 on page 25 the national interlocking network has indeed decreased, rather significantly with a total decrease of 21.2% from 2013-2015. The mean per company has also decreased with

(25)

Table 5: Interlocking directorate network within and across country borders All countries 2013 2015 Δ% 2013-2015 Companies with interlocks 202 204 1 % Of all firms 89.4 90.3 1 National interlocks 792 624 -21.2 International interlocks 270 197 -27 % International 34.1 31.6 -7.3 Total interlocks 1062 821 -22.7 Average national interlocks per firm 3.5 2.8 -20 Average international interlocks per firm 1.2 0.9 -25

Figure 4: % Changes number of interlocks, per country (2013-2015)

6.2 Shareholders and the interlocking directorate network

Table 6 on page 26 present the descriptive statistics for the regression analyses taking into account the year 2015 for all variables. The number of observations is 226 for each variable; this includes firms from 10 different countries. The total number of interlocking directorates is 3.6 on average, ranging from companies with 0 to 18 interlocks. The figure in appendix 10 gives an overview of how the interlocks are

-50,00 -40,00 -30,00 -20,00 -10,00 0,00 10,00 20,00 30,00

% Changes interlocks, per country

(26)

distributed per country. 62% of the firms have an ultimate owner; implying 38% of the companies do not. With regard to institutional investor ownership, 69% of the firms have an institutional investor as owner whereas 31% of the firms do not. Insider ownership is found in only 4% of the companies, which is a rather small percentage. The same holds for family ownership, which is found in only 5% of the companies. State ownership occurs more often since this is found in 17% of the firms in the sample. Foreign ownership 1 (i.e. the percentage of foreign owners) is high on average, since the mean is 73.6%. This value ranges from 0 to 100, as the percentage of foreign owners can be 0% or 100%. Foreign ownership 2 (i.e. the percentage of stock held by foreign owners) is 50% on average, ranging from 0% of stock held by foreigners to 100% of stock held by foreigners. The average number of employees per firm is 64265, ranging from 109 to 626715. The large difference between these

numbers might come across as abnormal, however since the sample includes very large firms and rather small firms these differences are logical. Boards of directors have an average of 16.4 directors (including executive and non-executive directors); the minimum number of directors is 6 whereas the maximum number of directors is 39. Lastly, the LME-CME score measures the country of origin; the average is 0.67 indicating that the companies in the sample tend to be located in countries having a coordinated market economy. The minimum of 0.07 is the country that is present in a LME; the maximum of 1 indicates a CME.

Table 6: Descriptive statistics (2015)

Variable Observations Mean Std. Dev. Minimum Maximum

ILD 226 3.63 3.13 0 18 OWN_CON 226 0.62 0.49 0 1 OWN_INSINV 226 0.69 0.46 0 1 OWN_INS 226 0.04 0.20 0 1 OWN_FAM 226 0.05 0.22 0 1 OWN_SO 226 0.17 0.38 0 1 OWN_FOR1 226 73.63 18.03 0 100 OWN_FOR2 226 49.98 28.40 0 100 FIRM_SIZE 226 64265 89364 109 626715 BOARD_SIZE 226 16.43 5.91 6 39 COO 226 0.67 0.24 0.07 1

(27)

ownership and the number of interlocking directorates is positive, however it is not significant thus there is no support for hypothesis 3. With regard to hypothesis 4, model (4) shows a relationship that is negative as expected, however again no

(28)

Table 7: Regression results Variable (1) (2) (3) (4) (5) (6) (7) (8) OWN_CON -0.33 OWN_INSINV 0.37 OWN_INS -0.56 OWN_FAM 0.43 OWN_SO -0.61 OWN_FOR1 0.02** OWN_FOR2 0.004

FIRM_SIZE 3.16e-06 3.13e-06 3.13e-06 3.13e-06 3.06e-06 3.17e-06 2.72e-06 3.20e-06

(29)

Besides the regressions including all countries, country-level regressions have been performed with regard to the variables of concentrated ownership and foreign ownership. This is done to get a thorough understanding of the presence of variables per country and the potential relations among these. The control variable “country of origin” is omitted since the regressions are on a country level. The detailed descriptive statistics per country are presented in appendix 11.Explanation per independent variable is presented in the section below.

As can be seen in figure 5 below the number of companies per country differs, ranging from 16 companies in Belgium to 29 companies in France.

Figure 5: Number of companies per country

With regard to concentrated ownership the extent to which an ultimate owner is present differs substantially per country (see figure 6 of page 30). In Austria, as was expected; all companies have an ultimate owner and thus concentrated ownership. Furthermore, Belgium, Denmark, Finland, France, Germany, the Netherlands and Sweden are all strongly characterised as a CME. In Belgium this corresponds to the presence of concentrated ownership, as in 94% of the companies an ultimate owner is present. Denmark has an ultimate owner present in 65% of the companies, not as high as Austria and Belgium but in more than 50% of the companies there is a

concentrated ownership structure. Having a closer look at Finland shows a percentage contradicting expectations; in 45% of the companies an ultimate owner is present. This potentially points at a negative relation between concentrated ownership and interlocking directorates in Finland. In France 69% of the companies have

concentrated ownership, which is in line with expectations. Germany was highly characterised as a CME (index of 0.95), 71% of the companies appear to have an ultimate owner and so a concentrated ownership structure. In the Netherlands the presence of an ultimate owner was expected more often since it is characterised as CME country, in 47% of the companies a concentrated ownership structure is found. Taking into account Spain, the presence of an ultimate owner is as expected. In 52% of the companies an ultimate owner is present. Likewise for Sweden where there is a

0 5 10 15 20 25 30 35

Number of companies per country

(30)

presence of an ultimate owner in 65% of the companies. Finally, the UK has by far the lowest percentage of ultimate owners present (20%), though this was expected as it is highly characterised as a LME.

Figure 6: Mean concentrated ownership per country

With regard to foreign ownership (1), thus the average number of foreign owners, the mean values varies per country differ as anticipated (displayed in figure 7 on page 31). However, Austria has a high percentage of foreign owners whereas this was not expected since they have the highest score of foreign direct investment restrictiveness. Belgium has a percentage that was expected (84%). The same holds for Denmark who was expected to have a high percentage of foreign owners, which shows to be the case (76%). Finland has a 58% foreign ownership; this was expected to be higher as their restrictiveness score was one of the lowest. France has as expected 69% of foreign ownership. Likewise in Germany, the percentage is 82%, which matches expectations since it was ranked as rather open to foreign direct investment. The Netherlands has the highest percentage of foreign ownership (91%) and this matches their openness to foreigners. Contrary to expectations, Spain has an average of 68% of foreign owners. Sweden is in line with expectations with an average of 65% of foreign ownership. Lastly, the UK has on average 74% of foreign ownership, which is expected since it is characterised as open to foreign direct investment.

(31)

Figure 7: Mean percentage of foreign owners per country

Moreover, the second measurement of foreign ownership (average stock held by foreign owners) is shown in figure 8 on page 32. However, these numbers show large differences with the percentage of foreign owners. On average the stock held by foreign owners in Austria is 28%, which is quite low. This is not surprising since Austria has a high score on the restrictiveness index. This means that the foreign owners do not hold large amounts of stock, and there might be a negative relationship between the stock held by foreign owners and the number of interlocking directorates in Austria. Belgium has a higher percentage (49%), although still rather low with regard to what was expected. The same holds for Denmark, the country is open to foreign direct investment however on average just 52% of the stock is in hands of foreign owners. Finland, France and Germany were all expected to have a rather high share of the stock in hands of foreign owners. Despite this expectation all three countries score under 50% (respectively 44%, 45%, 43%). This suggests there is a negative relation between the stock held by foreigners and the number of interlocking directorates in these three countries. The Netherlands is characterised as the most open to foreign investments, this corresponds in the percentage of stock held by foreigners which 67%. With reference to Spain on average 59% of the stock is held by foreign owners, which was expected to be slightly higher. Sweden has an extremely low percentage (40%) especially since their economy is found as having a low score with regard to the restrictiveness index. Finally, the UK matches the expectations having an open economy and on average 72% of stock held by foreigners.

(32)

Figure 8: Mean percentage stock held by foreign owners

The following section will discuss the regressions results per country. The dependent variable “total number of interlocking directorates per country” is used in all models. For all countries model (1) regresses solely with control variables, models (2), (3) and (4) regress with the three independent variables. Table 8 below shows the regression results for Austria. In all models except model (3) the control variable board size is significant (respectively 5%, 5%, and 10% significance). The firm size is not significant in all models. The relationship between concentrated ownership and the number of interlocks was expected to be significant. However, the regression

indicates a non-significant negative relation. With regard to foreign ownership it was expected to have a negative relation with the number of interlocks. Both measures of ownership confirm this; the relation in model (3) is not significant. In model (4) there is a significant relation (10% significance) between foreign ownership (2) and the number of interlocking directorates.

Table 8: Regressions results Austria

Variable (1) (2) (3) (4)

OWN_CON -0.20

OWN_FOR1 -0.04

OWN_FOR2 -0.03*

FIRM_SIZE -4.78e-06 -4.78e-06 0.00002 -8.93e-06

BOARD_SIZE 0.16** 0.16** 0.10 0.17* Constant -0.20 -0.20 3.05 0.51 N 17 17 17 17 F 2.44 9.77** 1.66 1.83 R2 0.15 0.68 0.21 0.30 *** p<0.01, ** p<0.05, * p<0.1 0 10 20 30 40 50 60 70 80

% Of stock held by foreign owners

(33)

With regard to the regression results of Belgium in table 9 below, there are no significant relations present in any of the models. Thus, in Belgium concentrated ownership and foreign ownership have no relation with the number of interlocking directorates.

Table 9: Regressions results Belgium

Variable (1) (2) (3) (4)

OWN_CON -2.18

OWN_FOR1 -0.01

OWN_FOR2 0.04

FIRM_SIZE -7.04e-06 -4.06e-06 -6.44e-06 -0.00002

BOARD_SIZE 0.14 0.12 0.14 0.21 Constant 0.86 2.97 1.49 -1.49 N 16 16 16 16 F 0.53 0.98 0.43 0.72 R2 0.04 0.10 0.04 0.15 *** p<0.01, ** p<0.05, * p<0.1

Table 10 below present the regression results for the Danish relationships. In model (1) firm size and board size are both significant at a 5% significance level. The regression in model (2) shows that there is a significant positive relations between concentrated ownership and the number of interlocking directorates, as was expected. With regard to model (3) the relation is not significant and positive. In contrast to expectations model (4) shows a negative relation, however this is not significant.

Table 10: Regressions results Denmark

Variable (1) (2) (3) (4) OWN_CON 1.60*** OWN_FOR1 0.009 OWN_FOR2 -0.01 FIRM_SIZE 0.00003** 0.00003** 0.00003** 0.00003** BOARD_SIZE 0.20** 0.20** 0.20** 0.19** Constant -0.40 -1.51 -1.02 0.31 N 26 26 26 26 F 11.88*** 23.47*** 8.31*** 8.69*** R2 0.33 0.46 0.34 0.36 *** p<0.01, ** p<0.05, * p<0.1

(34)

between foreign ownership and the number of interlocking directorates. Model (3) at a 5% significance level, model (4) at a 10% significance level. Model (4) was

expected to show a negative relation as the percentage of stock held on average was rather low.

Table 11: Regressions results Finland

Variable (1) (2) (3) (4)

OWN_CON -0.76

OWN_FOR1 0.11**

OWN_FOR2 0.05*

FIRM_SIZE 8.98e-06 7.90e-06 -0.00002 -0.00001

BOARD_SIZE 0.22 0.25 -0.10 0.11 Constant -1.23 -1.30 -1.60 -1.27 N 20 20 20 20 F 1.23 1.15 2.73* 2.38 R2 0.07 0.10 0.35 0.27 *** p<0.01, ** p<0.05, * p<0.1

Analysing the regression results for France in the table below it can be seen both control variables are not significant in model (1). The expected relationship with concentrated ownership is not positive, it is negative and not significant. The relation with foreign ownership in model (3) was expected to be positive; model (3) indicates that there is indeed a positive significant relation (10% significance).

Table 12: Regressions results France

Variable (1) (2) (3) (4)

OWN_CON -1.93

OWN_FOR1 0.03*

OWN_FOR2 0.004

FIRM_SIZE 3.90e-06 -1.43e-06 3.88e-06 4.00e-06

BOARD_SIZE 0.10 0.14* 0.11 0.11 Constant 2.99* 4.35 0.42 2.69 N 29 29 29 29 F 1.19 1.89 3.30** 0.78 R2 0.04 0.10 0.09 0.04 *** p<0.01, ** p<0.05, * p<0.1

The results for Germany in table 13 on page 35 show that in model (1) board size is significant at a 1% significance level, firm size however is not significant. The

(35)

Table 13: Regressions results Germany

Variable (1) (2) (3) (4)

OWN_CON -0.89

OWN_FOR1 0.03**

OWN_FOR2 0.008

FIRM_SIZE -6.34e-06 -5.28e-06 -6.23e-06 -5.84e-06

BOARD_SIZE 0.41*** 0.38** 0.40*** 0.40*** Constant -2.84 -1.72 -5.33 -3 N 28 28 28 28 F 4.51** 3.98* 4.45** 3.06** R2 0.34 0.35 0.36 0.34 *** p<0.01, ** p<0.05, * p<0.1

The results for the regressions taking into account the Netherlands in table 14 below do not show any significant relations. This implies concentrated ownership and foreign ownership have no relation with the number of interlocks.

Table 14: Regressions results the Netherlands

Variable (1) (2) (3) (4)

OWN_CON 0.40

OWN_FOR1 0.03

OWN_FOR2 0.004

FIRM_SIZE -7.06e-07 -3.21e-08 -2.33e-07 -5.28e-07

BOARD_SIZE 0.13 0.10 0.12 0.14 Constant 1.73 1.94 -1.14 1.43 N 19 19 19 19 F 0.18 0.16 0.14 0.12 R2 0.02 0.03 0.03 0.02 *** p<0.01, ** p<0.05, * p<0.1

Table 15 on page 36 shows that for Spain the control variable board size is significant at a 5% level in each model; firm size is not significant in any of the models. The relationship between concentrated ownership and the number of interlocks is negative as expected, however it is not significant. Foreign ownership was expected to

(36)

Table 15: Regressions results Spain

Variable (1) (2) (3) (4)

OWN_CON -0.20

OWN_FOR1 -0.00008

OWN_FOR2 -0.02***

FIRM_SIZE 3.87e-06 3.51e-06 3.87e-06 3.69e-06

BOARD_SIZE 0.25** 0.29** 0.25** 0.25** Constant -2 -1.60 -2 -0.94 N 23 23 23 23 F 6.26*** 4.04** 3.97** 10.82*** R2 0.21 0.22 0.22 0.30 *** p<0.01, ** p<0.05, * p<0.1

Regarding table 16 below the control variables for firm size are significant in the models. Model (2) shows a positive relation between concentrated ownership and the number of interlocks, unfortunately this is not significant. Models (3) and (4) display a non-significant negative relation between foreign ownership and interlocks, as partly opposing expectations.

Table 16: Regressions results Sweden

Variable (1) (2) (3) (4) OWN_CON 1.57 OWN_FOR1 -0.02 OWN_FOR2 -0.005 FIRM_SIZE 0.00003*** 0.00003*** 0.00003*** 0.00003*** BOARD_SIZE 0.09 0.11 0.09 0.09 Constant 1.86 0.60 3.08 2.14 N 23 23 23 23 F 13*** 7.87*** 8.27*** 8.82*** R2 0.29 0.33 0.30 0.29 *** p<0.01, ** p<0.05, * p<0.1

(37)

Table 17: Regressions results UK

Variable (1) (2) (3) (4)

OWN_CON -0.97

OWN_FOR1 -0.01

OWN_FOR2 -0.003

FIRM_SIZE -5.45e-07 -4.07e-07 -8.57e-07 -6.08e-07

BOARD_SIZE 0.28** 0.31** 0.29* 0.29* Constant -1.08 -1.27 -0.44 -0.94 N 25 25 25 25 F 3.06* 1.85 1.96 1.97 R2 0.14 0.18 0.14 0.14 *** p<0.01, ** p<0.05, * p<0.1

(38)

7. Discussion

This thesis analysed the developments the interlocking directorates network in Europe has gone through over the years 2013-2015. In line with expectations this network has decreased on a national level, within country borders. Contrary to expectations the international network has also decreased, whereas an increase was expected. Additionally, the influence of different types of owners have of the number of interlocking directorates is analysed. It is found that the presence of foreign owners has a positive and significant effect on the number of interlocking directorates. Country specific there was found that foreign ownership is positively related to interlocking directorates in Finland, France and Germany and negatively in Austria and Spain. Concentrated ownership is positively related to the number of interlocks in Denmark.

7.1 Interlocking directorate network

In general the network has seen a severe decrease in the number of national and international interlocks with an overall decrease of 22.7%. Based on the LME-CME index it was expected that Austria, Belgium, Denmark, Finland, France, Germany, the Netherlands, Spain, and Sweden would have dense networks of interlocking

directorates. The UK was not expected to have a dense interlocking directorates network. As predicted Austria, Belgium, Denmark, Finland, France, Germany, the Netherlands and Sweden have rather interlocked networks taking into account the national and international interlocks. The big linkers in the data are Germany (mean of 7 per company), Sweden (mean of 5.5 per company) and France (mean of 5.3 per company). Spain does not have a dense network of interlocking directorates with the lowest mean of all countries in the sample. Spain was characterised as CME, but not to a large extent. Thus potentially this can explain the dispersed network, assuming their economy would tend to have characteristics of LMEs. The UK was expected to have a small number of interlocks, however this is certainly not the case as the only countries having more interlocks are France, Germany and Sweden and these are the big linkers in the network.

These countries differences can be explained due to differences in the average board sizes firms have. Germany for example has an average board size of 25 directors per board whereas Spain only has an average of 13 members per board. Logically, the chance that Germany has more board members than Spain is almost twice as high. Another aspect that can be of influence on the country differences is the size of the companies in the sample. Research by Croci & Grassi (2014) showed that in Italy firm size has a positive effect on the extent to which companies are interlocked. This could hold for the countries in our sample as well. For example Belgium has a

average company size of 40,168 employees and an average of 2.7 interlocks per firm. France has an average of 111,691 employees per firm and an average of 5.3

(39)

Lastly, the extent to which a country is said to be open might have an influence. Openness of a county means the extent to which foreign investments and trade are allowed into the country (Jong, Smeets, & Smits, 2006). Several cultural aspects are found to have a significant effect on the openness; individualism increases openness whereas uncertainty avoidance and power distance decrease the openness of a country (Jong et al., 2006). Comparing for example Austria and Sweden, which differ quite a lot in the average number of interlocks (Austria 2.4, Sweden 5.5) this could be true. Uncertainty avoidance in Sweden is rather low (29) whereas in Austria this is substantially higher (70) (Hofstede, 2017). This suggests Austria is less open than Sweden, potential of influence of the number of directors allowed to enter countries. The national network of interlocking directorates has undergone a rather severe decrease from 2013-2015, having 21.2% fewer interlocks in 2015 than in 2013. All countries noticed a decrease or no change in the national interlocking directorate network. As discussed by Heemskerk et al. (2016) the small corporate elite that used to run the show with regard to interlocking directorates is diminishing. There are more individual directors who occupy the board positions (Heemskerk et al., 2016). This can be a reason for the decrease in national interlocks, meaning potential big linkers have lost their multiple board positions because directors occupying less interlocked positions replace them. Another explanation can be derived from potentially many (old) directors who have retired. If there are many directors that occupied multiple board positions but retired, these interlocks disappear as well. Whether or not this is the case can be investigated in potential future research.

With regard to the international interlocking directorate network, Belgium, Denmark, the Netherlands, Sweden and the UK experienced an increase in the number of international interlocks, whereas Austria, Finland, France, Germany and Spain have experienced a decrease or no change in the number of international interlocks. The overall decrease of the international network is 27%, which is considered severe. This international increase in Belgium, Denmark, the Netherlands and Sweden can be due to a preference for international board members. All four countries are characterized as a coordinated market economy, which involves strategic interactions with other companies (Hall et al., 2011). This could be a potential reason why they had a

significant increase in the number of international interlocking directorates, in order to gain access to resources and managers with experience from other countries. In that way strategic connections with companies outside the home country borders are established. Additionally, the extent to which a culture is characterised as

IB&M Supervisor: dr. K. van Veen Co-assessor: dr. P. Rao Sahib IE&B Supervisor: dr. P. Rao Sahib Co-assessor: dr. K. van Veen

Master thesis International Business & Management and

International Economics & Business

University of Groningen

Faculty of Economics and Business

The effect of shareholders on the

interlocking directorate network in

Europe

Annefleur Clercx

a.l.c.clercx@student.rug.nl

S2349027

June 20

2017

IB&M

Supervisor: dr. K. van Veen

Co-assessor: dr. P. Rao Sahib

IE&B

Table of Contents

Abstract

1. Introduction

2. Theory and hypotheses

4. Conceptual framework

5. Data & methods

6. Empirical results

Interlocks per country as mean per company

% Changes interlocks, per country

7. Discussion