Board Composition and CSR: The Influence of Board Diversity on CSR Performance of European MNEs and the Moderating Effects of Board Independence and CEO Duality

(1)

Master Thesis

Board Composition and CSR: The Influence of Board Diversity

on CSR Performance of European MNEs and the Moderating

Effects of Board Independence and CEO Duality

Mirjam Brandl S3023915

e.m.brandl@student.rug.nl

Supervisor: Dr. O. Lindahl Co-assessor: Dr. C.H. Slager

MSc International Business & Management

University of Groningen, Faculty of Economics and Business Date of Submission: June 18th_{, 2018}

(2)

II

Abstract

Diversity in corporate boards has become a topic of interest in politics, business practice and research over the past decades. More recently, the debate started to link board diversity to corporate social responsibility (CSR). However, to date, the results are inconclusive and show positive, negative or non-significant relationships. The key objective of this master thesis is to contribute to the debate by exploring effects of gender, age, nationality and tenure diversity in boards of directors on CSR performance of the European top 100 multinational enterprises (MNE). Based on agency, resource dependence and stakeholder theory, it is argued that a diverse board will focus more on CSR related issues to fulfill societal and stakeholder expectations. Furthermore, the moderating role of board independence and CEO duality is assessed in this context. The analysis conducted through OLS regression supports a positive impact of gender diversity on CSR performance. However, while the influence of age and nationality diversity remain insignificant, tenure diversity leads to differing results among the main analysis and the robustness tests. Moreover, contrary to expectations, board independence and CEO duality do not moderate the board diversity effects on CSR performance. The findings of the study challenge existing theories and methods and hence provide important implications for research on board diversity and CSR and managerial practice.

Keywords: Corporate social responsibility (CSR), board diversity, corporate governance,

(3)

III

Acknowledgments

(4)

IV

List of Figures

Figure 1: Conceptual Model……….17

List of Tables

Table 1: Descriptive Statistics, 2014……….24 Table 2: Correlation Matrix, 2014………26 Table 3: Regression Analysis, 2014……….………….28

List of Abbreviations

CG Corporate Governance

CSR Corporate Social Responsibility CEO Chief Executive Officer

e.g. exempli gratia

ESG Environmental, Social, Governance

EU European Union

MNE Multinational Enterprise ROA Return on Assets

ROE Return on Equity SD Standard Deviation

UK United Kingdom

(7)

1

1. Introduction

Over the past decades, the debate about the corporate role in society in the context of corporate social responsibility (CSR) has become a globally pressing issue (Amran, Lee, & Davis, 2014; Zhang, 2012). Corporate scandals and governance failures as well as societal and environmental damages have drawn attention towards corporate governance (CG) and raised the question of how CG and CSR are interrelated (Jain & Jamali, 2016). Companies, especially multinational enterprises (MNE), are increasingly expected to not only fulfill economic objectives but also to balance the interests of different stakeholders and to consider social and environmental impacts (Jamali, 2006). One of the major decision-makers in CSR context is the board of directors who drives the CSR goals and objectives (Jamali, Safieddine, & Rabbath, 2008; Rao & Tilt, 2016). Generally, the board fulfills two key roles. First, following agency theory, the board monitors the top management team to prevent that managers act in self-interest instead of shareholders’ interest (Fama & Jensen, 1983). Second, following resource dependence theory, the board’s primary role is to ensure and facilitate access to essential resources by maintaining the firm’s legitimacy within the external environment as well as ensuring support of important stakeholders (Pfeffer & Salancik, 1978). It has been argued that diversity can facilitate all of the board’s functions (Ali, Ng, & Kulik, 2014). By considering more diverse stakeholder interests, monitoring may be improved (Kang, Cheng, & Grey, 2007). Further, diversity may bring in a broader range of perspectives, increase creativity and the number of considered alternatives which can lead to improved strategic decisions and facilitate the access to vital resources through a wider network (Kosnik, 1990; Miller & Triana, 2009). A diverse board can additionally have a positive impact on reputation as it signals social justice, diversity, and inclusion to the public (Miller & Triana, 2009). Hence, companies are increasingly under public pressure to promote diversity in their boards, especially regarding gender and nationality to represent more diverse stakeholder groups and to reflect the growing demographic diversity of society (Van der Walt & Ingley, 2003; Zhang, 2012). Following this logic, a more diverse board will be more concerned with diverse stakeholder interests and may, therefore, be in favor of or promote CSR.

(8)

2 still limited and the results to date are rather mixed finding positive, negative as well as non-significant effects (Hafsi & Turgut, 2013; Harjoto, Laksmana, & Lee, 2015; Jain & Jamali, 2016). The main body of research focusses on gender diversity and generally shows positive effects of women on boards (Byron & Post, 2016). Other diversity characteristics such as age, nationality, educational background, functional expertise and experience, political orientation, religion or sexual preferences can influence group processes as well (Van Knippenberg, de Dreu, & Homan, 2004). However, there are only a few studies linking those characteristics to CSR (Rao & Tilt, 2016). The relationship between age diversity and CSR performance is still not clear and the research on this topic is inconclusive finding positive (Hafsi & Turgut, 2013) and negative (Harjoto et al., 2015) effects. While studies on cultural differences have been considered to be of high importance in managerial contexts in the past decades and there is common agreement that national culture shapes the individual’s knowledge and values, the number of studies linking national differences to CSR performance remains sparse (Estélyi & Nisar, 2016). Similarly, tenure diversity has rarely been addressed in a CSR context and findings vary among insignificant (Hafsi & Turgut, 2013) and positive links (Melo, 2012; Krüger, 2009). While board independence (e.g. Johnson & Greening, 1999; Rahman & Rubow, 2011) and CEO duality (e.g. Jizi, Salama, Dixon, R., & Stratling, R., 2014; Webb, 2004) have been found to directly impact CSR performance, their separate moderating role on the relationship between board diversity and CSR does not seem to be explored yet.

The key objective of this project is to systematically contribute to this research field by exploring the influence of board diversity in terms of gender, age, nationality and tenure diversity on CSR performance of Europe’s 100 largest MNEs. As the understanding of the interrelationship between board diversity effects and CSR performance is still limited, this master thesis combines several group-level diversity characteristics and assesses their effects empirically. While most prior studies focus on gender diversity and board independence, this thesis adds less examined factors such as age, nationality, and tenure diversity to improve the understanding of diversity impacts. As opposed to prior studies and novel approach, board independence and CEO duality are not incorporated as direct diversity effects but as moderators building on the attempt of Hafsi and Turgut (2013) who aggregated several structural board characteristics as a moderator on the effect of board diversity on CSR performance.

This master thesis explores this interrelation with the following research question: Does

(9)

3 Following the advice of Harrison and Klein (2007), the study distinguishes among the dimensions of diversity instead of calculating a general diversity index to explore each effect separately. Additionally, the setting of the study does not solely focus on MNEs from one country but include MNEs which are headquartered in the European Union, which has a leading role in CSR topics. In the first chapter, the theoretical background of the topic and a review of literature on CSR, its interrelationship with CG, and to date research on impact of board diversity on CSR is provided to introduce the underlying key concepts. Based on stakeholder, resource dependence and agency theory, the hypotheses and a corresponding conceptual model are developed. Subsequently, the methodology including a description of the data and measures of the project is presented. Thereafter, the results of the study are described, explained and discussed. To conclude, reflections regarding limitations and contributions of the study as well as recommendations future research are proposed. Beside the theoretical contributions, this master thesis is of practical relevance by providing advice on how to compose boards more effectively to achieve better CSR performance results.

2. Theory

The following section introduces the CSR construct, its relevance for corporations as well its interrelationships with CG and board diversity. Accordingly, the hypotheses development section combines the underlying concepts to explain the relationship between board diversity dimensions and CSR as well as the moderating effects of board independence and CEO duality. Finally, the derived hypotheses are presented in a conceptual model.

2.1 Literature Review

2.1.1 Corporate Social Responsibility

Definition of the CSR Construct & CSR Performance

(10)

4 concerns in their business operations and in their interaction with their stakeholders on a voluntary basis.” This definition allows to operationalize CSR performance in terms of a company’s social and environmental assessment which it is later analyzed in and set it in context with the study of EU firms.

Evolution of CSR in context of Stakeholder Theory

Although first ideas about the corporate role in society emerged during the 1930s and 1940s, the awareness for CSR started to rise in the 1950s (Carroll, 2008). In 1953, Bowen argued that business is responsible beyond shareholder’s interest and to do good to contribute to social welfare to be considered legitimate in society. In his view, business was obligated to respond to social failure caused by laissez-faire economy. However, his work raised numerous questions about the role of business in society and the understanding for linking doing good for society and business was still scant (Carroll & Shabana, 2010). Further, Bowen’s approach was not without criticism. His main opponents, Levitt (1958) and later Friedman (1962; 1970), warned that engaging in social issues exceeds the obligations, competencies, and responsibilities of business and should be left to politics. In the 1960s, while Frederick (1960) argued rather broadly that corporations should use their resources also for social goals, McGuire (1963) pointed out the need for businesses to do good beyond legal and economic requirements as a duty towards its employees and society. Davis (1960) even emphasized social responsibility as a managerial responsibility to maintain a powerful position because business can only be successful in a healthy society. In the 1970s, scholars started to make the definition of the construct more concrete based on a new rationale. Under the assumption that businesses need to do something for society to maintain support and their customers (Wallich & McGowan, 1970), an era of ‘enlightened-self-interest’ began (Lee, 2008). Scholars started to link CSR and financial performance but did not yet logically explain their interrelationship (Weick, 1977). At the same time, there was growing criticism on the CSR construct. According to Friedman (1970), who established the shareholder perspective, businesses only have one social responsibility which is to generate profits. He argues that allocating resources towards CSR would conflict with the pursuit of profit maximization, would not be in line with the shareholders’ interest and creates agency problems.

(11)

5 into economic, legal, ethical and discretionary (philanthropic) categories which are fundamental for business to exist (Carroll, 1979; 1991). The economic dimension builds the fundament of the model and addresses the responsibility to make profit and in order to maintain the business operations. Further, the legal responsibilities concern complying with ground rules such as laws and regulations. These often reflect the ethical values of society (Carroll, 2016). Additionally, the normative component of society’s expectations are the ethical responsibilities such as operating within ethical standards even if there is no legally binding force. Lastly, discretionary or philanthropic responsibilities cover all voluntary activities to be a good corporate citizen. While the first two responsibilities are required by society, ethical responsibility can be considered as expected and philanthropic responsibility as desired by society (Carroll, 1979; 1991).

A fundamental challenge for measurability of the CSR construct was that ‘society’s expectations’ were too broad to be measured (Lee, 2008). Freeman (1984), arguing in favor of the stakeholder theory, suggested that a firm should not only consider the owner’s interests but the interests of all groups or individuals which are affected by or affect the business. Stakeholders can be distinguished into two groups: First, there are primary stakeholders who are essential to a firm’s survival such as customers, governments, suppliers, investors, and employees. Secondary stakeholders are not directly involved in transactions with the firm. However, media, activist groups or NGOs can have a strong influence on the firm’s surrounding and reputation (Clarkson, 1995). The idea behind this theory is that a firm’s ability to sell its products is affected by its ability to adjust its values to stakeholder and societal expectations (Freeman, Wicks & Parmar, 2004). Following this approach, publications by Clarkson (1995) and Jones (1995) applied stakeholder theory in their CSR views and exploited it as a strategic instrument for management. Thereafter, it became the dominant view in CSR research and a complementary construct to CSR (Carroll, 2015; Lee, 2008). Thus, this master thesis argues in favor of the importance of stakeholders and their impact on MNEs’ CSR strategies.

(12)

6 practices complemented by rising external pressure on MNEs to operate transparently, CSR and consideration of social and environmental impacts have evolved to an important part of management and to a vital part of business success (Vogel, 2005; Zhang, 2012).

Current Research on the Business Case for CSR

The impact of CSR on firm performance is still under scientific debate. In a comprehensive review, Aguinis and Glavas (2012) show that there are many different perspectives in which scholars have aimed to capture CSR drivers, impacts and outcomes. Berger, Cunningham, and Drumwright (2007) describe how the business case for CSR can be seen from two perspectives: In a narrow view, there is only a business case for CSR if it can be directly related to firm performance and especially to financial performance. However, if a broad view is considered, a business case for CSR also exists for indirect impacts on firm performance. The latter approach has the advantage of including a broader array of relationships between CSR and firm performance and further acknowledges interrelationships between firms and their environment (Berger et al., 2007; Carroll & Shabana, 2010).

A rather narrow and probably most common approach to measure CSR and to legitimize it on economic reasoning was to consider it in direct context to financial performance (Carroll & Shabana, 2010; Lee, 2008; Weber, 2008). Yet, due to varying CSR definitions, CSR performance measures and datasets, the findings of the respective studies are mixed (Margolis, Elfenbein, & Walsh, 2007; Vogel, 2005). In a meta-analysis, Margolis et al. (2007) investigated 167 studies from a period of 35 years (1972-2007). Their results showed an overall positive, however rather modest, effect of CSR on financial performance.

Due to the differences in financial performance studies, researchers who suggest that CSR has positive effects on the general firm performance, often argue in favor of the broader perspective on the business case for CSR and include more dimensions to analyze the value of CSR for firms (e.g. Carroll & Shabana, 2010; Kurucz et al., 2008; Weber, 2008).

(13)

7 reputation by the claim for legitimacy. By fulfilling society’s and stakeholders CSR expectations, a firm may enhance the legitimacy of its operations. As a result, the firm may benefit from increased credibility (Pfau, Haigh, Sims & Wigley, 2008), improved evaluation through customers, trust and brand loyalty (Pivato, Misani & Tencati, 2008; Sen & Bhattacharya, 2001) which may lead to increased sales, revenues and market shares (Weber, 2008).

Studies have also found that a positive CSR reputation increased the firm’s attractivity as an employer (Behrend, Baker & Thompson, 2009; Turban & Greening, 1997). Furthermore, engaging in CSR may also positively affect the motivation and retention of existing employees (Bhattacharya, Sen, & Korschun, 2008) and may strengthen corporate values which may create a stronger organizational identity (Carmeli, Gilat, & Waldman, 2007). If employees identify with their employer and feel integrated into the corporation by shared values, employee commitment can be increased (Peterson, 2004) which may lead to improved productivity.

Investing in CSR may further create cost saving potentials and reduce CSR related risks. Kurucz et al. (2008) argue that investing in environmental and social issues may mitigate negative aspects of potential threats through stakeholder demands. Especially in the sustainability domain, development of more sustainable, environmentally friendly products and production processes may contribute to the creation of more efficient, time-saving and energy-saving value chains (Epstein & Roy, 2001). Beside lowering operating cost, it can also prevent sanctions by regulators in the present or future penalties and lower social concerns in the communities (Bergman, Wicks, Kotha & Jones, 1999). Improving community relations can lead to advantages such as tax reductions (Kurucz et al., 2008). Further, a firm may also reduce the risk to be involved in CSR related scandals which would harm its reputation (Weber, 2008). In summary, although direct effects of CSR on financial performance are mixed and some scholars argue that CSR efforts may negatively impact the firm by reducing financial resources (Shen & Chang, 2007), research has shown that CSR can have various positive effects which are likely to exceed the costs. Hence, it may be argued that a corporation that is able to combine its own financial objectives with CSR performance and its stakeholders’ expectations may create a win-win situation and a competitive advantage (Kurucz et al., 2008).

2.1.2 CSR, Corporate Governance & the Board of Directors

(14)

8 CG can be narrowly defined as the regulatory mechanisms which direct and control a corporation (Saravanamuthu, 2004). Other definitions take a broader view into account and include the corporation’s responsibilities towards all stakeholders who provide vital resources for the firm’s survival and success (MacMillan, Money, Downing, & Hillenbrad, 2004). Since CSR can be an important tool which can improve a corporation’s reputation, mitigate CSR related risks and can promote credibility for important stakeholders (Bear, Rahman, & Post, 2010; Pfau, et al., 2008), it can be argued that CSR and CG are two closely interrelated concepts which are both based on maintaining positive stakeholder relationships. Studies linking CSR and CG are of rather recent nature. Jain and Jamali (2016) found CG mechanisms influencing CSR in studies on institutional level (e.g. legal and political factors; norms, values, and culture), firm level (e.g. ownership structure), group level (e.g. board structure, board demography, board social capital), and on individual level (e.g. CEO demography, CEO socio-psychological capital). Although factors from these different levels may be interrelated, the link between CG an CSR has strongly been suggested in the context of the board of directors given the assumption that CSR is the outcome of the board’s decisions and that the board is highly involved in stakeholder management (Jain & Jamali, 2016; Rao & Tilt, 2016).

Functions of the Board

(15)

9 the board to provide resources may help to reduce resource dependence from single external sources which reduces uncertainty, transaction cost and may ensure vital resources (Hillman & Dalziel, 2003). Finally, scholars have emphasized on the role of the board as strategic advisors (Johnson, Daily, & Ellstrand, 1996; Zahra & Pearce, 1989). Further, it has been suggested that a diverse board which represents different experiences, backgrounds, and personal attributes may be able to achieve more effective strategic outcomes (Hillman et al., 2001). Following this argumentation, several studies proposed that the board has an important role in decision-making processes and has a key role in formulation of CSR strategies (Ali et al., 2014; Jamali et al., 2008). In this context, the debate of diversity of boards gained attention in public and in research.

2.1.3 Board Diversity Debate

Diversity can be broadly defined as heterogeneity among board members and can be measured in diverse dimensions such as gender, age, nationality, religion, professional experience, education, organizational membership and others (Kang et al., 2007; Van Knippenberg et al., 2004). As modern societies tend to become more diverse, it has been argued that boards should reflect this societal diversity to represent more stakeholders’ interests (Van der Walt & Ingley, 2003). However, the diversity debate is two-sided. According to social identity theory (Tajfel, 1978) and social categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), individuals tend to categorize themselves based on demographic attributes. This may lead to a more positive attitude towards ‘in-group’ members with the same demographic attributes in contrast to ‘out-group’ members who are different. This may hamper the decision-making process in a group through stereotyping and lack of communication or cooperation (Ali et al., 2014). Therefore, this theory may explain why several studies on board diversity have found negative or null-effects of diversity (e.g. Ali et al., 2014; Westphal & Milton, 2000). In contrast, a homogeneous board may lack external connections, variety in skills, information and knowledge, have limited perspectives and may hence have less creativity in problem solving (Miller & Triana, 2009; Nielsen, 2010). Therefore, diversity can improve a group’s creativity, finding solutions, and promote the decision-making process (Nielsen & Huse, 2010; Rose, 2007).

(16)

10 CSR performance may be realized (Fama & Jensen, 1983; Hillman & Dalziel, 2003; Kang et al., 2007). According resource dependence theory, board diversity may positively influence the board’s ability to ensure access to important resources for the firm by balancing different stakeholders’ interests including social and environmental concerns, maintaining more external contingencies, creating a positive public image and increasing the firm’s legitimacy (Hillman & Dalziel, 2003; Luoma & Goodstein, 1999; Pfeffer & Salancik, 1978). A diverse board may have a broader strategic perspective which may positively support the strategic role of the board (Rose, 2007). As CSR basically is a form of effective stakeholder management, it is likely that board diversity will positively impact CSR performance. Based on this argumentation and prior studies, it is suggested that board diversity has the potential to lead to competitive advantages for the corporation and to outperform homogenous boards in CSR performance (e.g. Bear et al., 2010; Harjoto et al., 2015; Zhang, 2012).

Following upper echelons theory (Hambrick & Mason, 1984), strategic leadership is influenced by personal attributes such as personal values, knowledge, and experiences of the individuals in charge. Therefore, the hypotheses are developed based on individual demographic and cognitive characteristics, namely gender, age, nationality and board tenure. Furthermore, the potential moderating effects of the board structural attributes board independence and CEO duality are introduced.

2.2 Hypothesis Development

2.2.1 Board Gender Diversity & CSR Performance

(17)

11 Dietz, & Kalof, 1993). This suggests that women promote a more democratic and inclusive leadership style which enhances decision-making (Rudman & Glick, 2001). In line with this logic, many studies have found a positive effect of women in boards on CSR performance as the presence of female directors increases the focus on diverse stakeholder interests and as women generally focus more on business ethics and CSR topics (e.g. Bear et al., 2010; Byron & Post, 2016; Hafsi & Turgut, 2013). However, it can be argued that men may have better access to the traditional ‘old boy network’ with stronger ties which may be beneficial for the relationships to other stakeholders (Miller & Triana, 2009) and have more leadership experience in large organizations (Hillman, Canella, & Harris, 2002). Further, they may focus more on financial resources which are also necessary to ensure the MNE’s ability to engage in CSR practices. Therefore, a diverse mix of both genders within the board may provide inclusion of more diverse stakeholder interests and may be the most effective combination for a strong CSR performance. Female directors may foster sensitivity towards CSR related topics and stakeholder concerns while male directors may ensure the link to corporate performance and the firm’s ability to engage in CSR. Hence,

H1: The higher the gender diversity within the board, the higher the MNE’s CSR

performance.

2.2.2 Board Age Diversity & CSR Performance

(18)

12 Jhunjhunwala & Mishra 2012). This would mean that younger directors are more sensitive towards CSR issues. However, considering that different age groups will represent reflect the needs of multiple stakeholder groups, it can be proposed that age diversity within the board will balance different interests of stakeholders and CSR issues. While older directors may be more likely to consider the wellbeing of society and to make more mature decisions, younger directors may be more familiar with technological possibilities to do so and may further focus on different issues than their older colleagues due to generational effects. Therefore,

H2: The higher the age diversity within the board, the higher the MNE’s CSR

performance.

2.2.3 Board Nationality Diversity & CSR Performance

(19)

13 about the domestic market and may have an advantage in accessing important domestic stakeholder networks. Based on social identity theory, directors with a domestic nationality may be rather perceived as in-group members by stakeholders in the MNE’s home country and may be perceived with more benevolence (Feather, 1994; Tajfel, 1978). Additionally, non-foreign directors are more familiar with the national culture, values, informal and formal institutions which does not only improve the understanding for the domestic stakeholder’s needs but may also support a more effective communication with the MNE’s external domestic environment (Masulius et al., 2012). Given these findings, it can be suggested that nationality diversity with non-foreign and foreign directors in boards can advance and broaden stakeholder relationships and CSR performance. Hence,

H3: The higher the nationality diversity within the board, the higher the MNE’s CSR

performance.

2.2.4 Board Tenure Diversity & CSR Performance

(20)

14 In contrast, directors with a shorter tenure are less biased by the firm-specific perspective and bring in new opinions, knowledge and skills which may enhance the board’s decision-making processes (Van Knippenberg et al., 2004). Therefore, shorter tenured directors might positively contribute to the firm’s CSR performance through inclusion of broader views and stakeholder interests.Additionally, according to the findings of Li and Wahid (2017), they are more distant to the CEO which positively influences the board’s monitoring capabilities because shorter-tenured directors may be less likely to follow the CEO’s strategy which is essential if the CEO is not in favor of promoting CSR. Following this argumentation, tenure diversity may lead to improved CSR performance. While longer tenured directors have more firm-specific experiences with more sensitivity for its stakeholders needs, shorter tenured directors bring in new perspectives and are less captured by the top management which might enhance critical thinking over management strategies. Therefore, a tenure diverse board may benefit from inclusion of more perspectives, firm knowledge and sensitivity for a broader range of stakeholder interests and hence foster CSR performance. Therefore,

H4: The higher the tenure diversity within the board, the higher the MNE’s CSR

performance.

2.2.5 Moderating Effect of Board Independence

Often studies incorporate the independence of the board members as diversity indicator or as an independent effect on CSR (e.g. Post et al., 2011; Shaukat, Qui, & Trojanowski, 2016; Zhang 2012). However, the independence of the board can also be a measure for the structural composition instead of an individual diversity measure. Following the suggestions of Hafsi and Turgut (2013) who clearly differentiate between diversity in boards (demographic, cognitive characteristics) and diversity of boards (structural characteristics), this thesis incorporates this variable as a moderator which may impact the diversity effects rather than a diversity variable itself.

(21)

15 percentage of independent directors on CSR performance (e.g. Johnson & Greening, 1999; Post et al., 2011; Webb, 2004). Following the agency perspective, whether a director is an independent or not may influence his perspective and behavior (Hillman & Dalziel, 2003). An inside director who is dependent has a personal interest in the firm’s short-term financial performance and might hence be biased in his decisions from internal MNE perspectives. Therefore, the positive effects of a diverse board may be weaker if the insider percentage is higher as the information basis and the interests of the directors may be more similar despite their demographic background. In contrast, the positive diversity effects may even be stronger for independent directors because they may not only bring in new perspectives and a broader network into the board but through their personal independence, they also have no incentive to only operate in the short-term, profit-oriented interest of the MNE. Therefore, it can be suggested that the previously introduced board diversity effects of gender, age, nationality and tenure diversity of boards will be stronger if a higher percentage of directors are independent because dependent directors may be more in favor the top management’s strategies and short-term profit maximization than independent directors. A more independent board may have a greater ability to act on the directors’ true perspectives and values in the company’s interest rather than personal interest. Hence,

H5a: A higher proportion of independent directors on the board will positively

moderate the positive effect of board gender diversity on an MNE’s CSR performance.

H5b: A higher proportion of independent directors on the board will positively

moderate the positive effect of board age diversity on an MNE’s CSR performance.

H5c: A higher proportion of independent directors on the board will positively

moderate the positive effect of board nationality diversity on an MNE’s CSR performance.

H5d: A higher proportion of independent directors on the board will positively

moderate the positive effect of tenure diversity on an MNE’s CSR performance.

2.2.6 Moderating Effect of CEO Duality

(22)

16 value are a CEO’s main responsibilities. Shareholder theorists argue that allocating firm resources to CSR, including financial means, can lead to a reduction in financial resources and may hence be not in line with the aim for profit maximization (Friedman, 1970). However, from a stakeholder theory perspective, it can be argued that CSR may be also beneficial for the MNE’s financial performance due to positive reputational effects and improved stakeholder relationships (Freeman et al., 2004). To decide whether and how much CSR efforts are favorable for the shareholders’ and MNE’s interest, it is important that the board is fully capable of its monitoring task. However, if the CEO holds a dual role, it is more likely that the board will act in line with the top management’s and the CEO’s interests as a dual CEO may also be in favor of selecting a more CEO friendly board (Shaukat et al., 2016). Therefore, although the board may still consist of directors with diverse cognitive and demographic attributes, the directors may be more likely to have a similar mindset and perspective. Analog to the argumentation how the ratio of independent directors relates to diversity, this may affect the board’s monitoring capabilities, ability to act on its perspectives and weaken the attention for more diverse stakeholder interests. Thus, even a theoretically more diverse board may be less powerful if the CEO blocks its decisions or if the positive impact of diversity is weakened by too similar mindsets. Hence, CEO duality may decrease the positive effects of gender, age, nationality and tenure diversity on CSR performance.

H6a: CEO duality will negatively moderate the positive effect of board gender diversity

on an MNE’s CSR performance.

H6b: CEO duality will negatively moderate the positive effect of board age diversity on

an MNE’s CSR performance.

H6c: CEO duality will negatively moderate the positive effect of board nationality

diversity on an MNE’s CSR performance.

H6d: CEO duality will negatively moderate the positive effect of tenure diversity on an

MNE’s CSR performance.

2.3 Conceptual Model

(23)

17 negative moderating effect of CEO duality (H6a-d) on the positive influence of the board diversity factors on CSR performance are integrated into the model.

Figure 1: Conceptual Model

3. Research Methodology

This section introduces the research methodology to test the hypotheses. First, the information about the data collection is provided. Subsequently, the sample is described, and the measures used for this study are explained. Lastly, the data analysis is outlined.

3.1 Data Collection

To explore the previously presented relationship between board diversity on CSR performance and the moderating effects of board independence and CEO duality, a quantitative approach is applied. Databases provided to students of the University of Groningen as well as publicly available data are used as sources for the secondary data. The data for the dependent variable

CSR Performance is obtained via the Thomas Reuters Asset4 ESG database. The board data for

(24)

18 Studies working with similar data assume that board diversity affects corporate outcomes a lag of time and not immediately since changes due to board composition are expected to develop over time (Carter, D’Souza, Simkins, & Simpson, 2010). Additionally, it has been argued that research on board data is confronted with endogeneity and reverse causality effects because firms who engage more in CSR activities may naturally attract a more diverse board in favor of CSR issues and select a more CSR friendly board (Hermanlin & Weisbach, 2003). A commonly used suggestion to counteract these effects is to use lagged variables by lagging the dependent variable in comparison to the independent, moderating and control variables (Carter et al., 2010). Following these argumentations, the data for the independent variables, moderators and control variables is collected for the years 2014 and for robustness purposes for 2015 and the data for the dependent variable for 2016 creating a 2- and 1-year time lag. Including the data of two years may be a control for these effects. Although board composition is not likely to change considerably between two years, comparing the samples of the two years may offer important implications for the data interpretation and for the robustness of the results in case of changes.

3.2 Sample

The sample for this study focusses on the European largest MNEs given the assumption that Europe has a leading role in environmental engagement and CSR issues (Post et al., 2011). Only MNEs which are headquartered in the EU are included because the EU gives a rather common framework regarding CSR practices and disclosure of corporate boards (Harjoto et al., 2015). The approach to integrate a region instead of single countries is based on the following thoughts. First, including several countries with a similar legal framework may create a more diverse sample than focusing on one country. Some countries have mandatory and country-specific quotas for female representation on boards. A too homogenous sample might hide the effect of such factors. Second, it can be expected that the largest firms are legally obligated to disclose information which ensures sufficient data availability. Lastly, the suggested sample provides insight in several industries which may allow for more comprehensive results than focusing on one single area.

(25)

19 ranking published in 2017 included 120 companies which are headquartered in the EU. From this initial sample, one MNE is excluded caused by unrealistic data variations and 19 MNEs due to missing Asset4 data. The sample of 100 MNEs still fulfills the minimum requirement and qualifies as the final sample.

3.3 Measures

3.3.1 Dependent Variable

Various approaches have been used to measure and operationalize CSR (Margolis et al., 2007). Although it is under debate which method is the most effective, using social, environmental and governance (ESG) data provided by third-party agencies such as the MSCI ESG Index (formerly known as KLD), Thomson Reuters Asset4 ESG scores1 and Sustainalytics (ESG indicators) for CSR performance ratings seem to have become the most common approach in academics. These ratings aim to make the firm’s activities more transparent to stakeholders who may lack the expertise to relate the firm’s activities to environmental or social impacts or may not be aware of all activities (Chatterij, Levine, & Toffel, 2009). Although all these databases are based on ESG data, the conceptualization and measurements differ. Therefore, results of academic studies may depend on the underlying data source (Chatterij, Dourand, Levine, & Touboul, 2016). While the MSCI (KLD) data seems to be the most commonly used data source for CSR performance (Harjoto et al., 2015), it is measured based on ESG scores of the Asset4 database in this thesis. Since 2002, Asset4 research specialists collect ESG data from various sources and evaluate firms based on more than 700 firm level ESG and data points. As this is one of the most comprehensive and objective databases on CSR, Asset4 provides appropriate data for this analysis (Shaukat et al., 2016). Since the governance pillar measures structural and strategy indicators and includes data on board composition which overlap with the independent variables, this category is excluded from this study. The environmental category which Asset4 defines as “a company's impact on living and non-living natural systems, including the air, land, and water, as well as complete ecosystems”, contains indicators for resource use, emissions, and innovations. The social pillar is defined as “a company's capacity to generate trust and loyalty with its workforce, customers, and society, through its use of best management” and comprises indicators for the workforce, human rights, community relations and product responsibility (Asset4, 2013). Both categories, environmental and social, create a score from 0 (low performance) to 100 (high performance). Using the average of both dimensions to measure

1_{Asset4 is relaunched as Thomson Reuters ESG Scores in 2018 (Thomson Reuters ESG Scores, 2017).}

(26)

20

CSR performance is an approach which has been previously applied in a similar manner in

research (e.g. Cheng, Ioannou, & Serafeim, 2014; Shaukat et al., 2016). 3.3.2 Independent Variables

This study uses dichotomous data for gender and nationality diversity which allows calculating diversity using simple percentages to create values between 0 and 1 (e.g. Ararat, Aksu, & Cetin, 2015). This methodological approach has is simple to apply and the variables do not need to be rescaled for testing. Further, it compares the differences among boards relative to the board size (Post et al., 2011). Following this approach, the independent variable Gender can have the characteristic male or female. The value for Gender is calculated by the percentage of female directors on the board though dividing the number of female directors by the total board size. With the same logic, the independent variable Nationality with the categories non-foreign, which means that the director has the nationality of the country the MNE is headquartered in, and foreign is calculated through the percentage of foreign directors on the board. For Age and

Tenure, a different approach is used. Age is measured in total years of the directors in the year

of examination. Consistent with prior studies (e.g. Ali et al., 2014), the coefficient of variation formula is applied to calculate the diversity value for this variable by dividing the standard deviation by the mean for each board. The same method is used for the variable Tenure measured by years which each director has spent on the respective board.

3.3.3 Moderating Variables

Board Independence is measured following the same approach as Gender and Nationality. Each

director can either be an independent director if he has no personal or professional dependence from the firm (Pearce & Zahra, 1991) or a dependent director if he is employed by the respective MNE or has any other financial interest in the firm. By dividing the number of

independent directors on the board by the total number of directors on the board the ratio of

(27)

21 3.3.4 Control Variables

Board Size

Studies argue that larger boards may have more independent members and have more external links which may indicate relationships to more stakeholders and access to more resources (Post et al., 2011). Simultaneously, Benson, Davidson, Wang, and Worrell (2011) suggested that smaller boards spend less on CSR and have a less effective stakeholder management. Following these assumptions, board size may affect an MNE’s CSR performance. Hence, Board size is controlled for by the total number of directors on the board (Hafsi & Turgut, 2013).

MNE Size

It has been proposed that larger firms have more resources. Greater financial resources have been associated with increased engagement in CSR activities and are confronted with a higher external stakeholder pressure to foster their CSR performance (Hafsi & Turgut, 2013). Further, Vogel (2005) argues that the larger a firm, the greater its social and environmental responsibility. Therefore, MNE size may affect CSR performance and is measured by the total number of employees of the respective MNE (e.g. Ali et al., 2014).

MNE Profitability

A more profitable MNE may also have more financial resources to invest in CSR activities. Profitability is commonly measured in return on assets (ROA) and return on equity (ROE). Scholars have found positive a positive relationship between ROA and CSR performance (e.g. Campbell, 2007). However, since this thesis includes MNE from different industry sectors with various levels of capital-intensity, ROE which measures how effectively a firm generates profits from its shareholders’ investments seems to be more appropriate and is used to measure MNE

Profitability consistent with prior research (e.g. Harjoto et al., 2015; Zhang, 2012). Industry Type

(28)

22 3.4 Empirical Data Analysis

The suggested hypotheses are evaluated using IBM SPSS Statistics (Version 24). As the study examines multiple independent variables to explain one continuous dependent variable, a multiple linear regression analysis based on the ordinary least square (OLS) method is applied. Since moderation effects are included in this study, the independent variables and the continuous moderator Board Independence are mean centered for the entire analysis. Before conducting the main regression analysis, the preliminary requirements for linear regression are tested to ensure an adequate interpretation of the regression results.

Normal distribution is tested by using a Shapiro-Wilk test and checking for skewness and kurtosis. These tests show that Gender is the only normally distributed variable in the dataset. Since normal distribution of the variables is not an essential requirement for linear regression and is further not expected from the nature of the data, this should not be an issue for the analysis (Lumley, Diehr, Emerson, & Chen, 2002). To ensure that the skewed data does not create non-parametric, non-linear effects, the correlation among the variables with Pearson and Spearman correlations are compared. Since both correlation analyses deliver very similar results, this does not seem to be the case. However, the normality of the residuals is an essential requirement for linear regression which is not fulfilled. Therefore, the dependent variable is transformed by the following formula to correct this result:

CSR Performance= CSR raw data / (100-CSR raw data)

(29)

23 tests are insignificant for all regression models, heteroskedasticity does not seem to disturb the results (details in Appendix B).

3.5 Robustness Test

To ensure that the results have sufficient validity, the regression analysis is first repeated by comparing the consistency of the results with the 2015 data and additionally through a modification of the independent variables by the calculation of the Blau indices for both years. Blau’s diversity index (Blau, 1977) is commonly used in diversity research (e.g. Ali et al., 2014; Harjoto et al., 2015). Instead of using simple percentages, this index is an alternative method to explore diversity based on categorical data and uses the following formula:

𝑘 Blau Index

=1− ∑𝑝²

i 𝑖=1

K stands for the number of categories and p for the proportion of directors in each category. The index can have values between 0 and (k-1)/k where 0 stands for a perfectly homogeneous group and (k-1)/k for a completely heterogeneous, diverse group. Therefore, the value depends on the number of categories (Harrison & Klein, 2007; Coding in Appendix C).

4. Results

The results of the analysis are presented in this section. First, the descriptive statistics and the correlations of the data are shown. Second, the results of the main regression to test the hypotheses of this thesis are described. Finally, the results of the robustness tests are disclosed. 4.1 Descriptive Statistics

(30)

24 In the 2014 sample which includes data of 1,553 directors, the Gender variable measured in the percentage of women in boards ranges between 0% and 54.54% (mean= 25.37%, SD=10.4%) and shows that female directors are underrepresented in corporate boards. This assumption is supported by the absolute numbers of 383 female and 1,170 male directors. The coefficient of variation measuring Age ranges from 6.45% to 24.51% (mean=13.01%, SD= 3.35%). The age of the individual directors’ ranges between 28 years and 90 years with an average of 58.45 years. Nationality includes extreme values such as 0% of foreign directors to 100%. However, the mean (30.45%; SD=22.46%) demonstrates that most directors on the boards are non-foreign. In general, the sample includes 435 foreign and 1118 non-foreign directors from a total of 46 different countries. Tenure has a broad value range between 0% and 134.94% (mean=72.99%, SD=19.2%). This distribution indicates a high dispersion among the boards with very similar to highly diverse tenure differences. The individual tenure of directors varies between 0 and 39.8 years with an average of 5.97 years. According to these statistics, the sampled boards seem to be most heterogeneous in Tenure. The ratio of independent directors ranges between 0% and 92.31% with an average of 53.99% (SD=25.77%). 797 directors in the sample are dependent and 756 are independent. In 52 boards, the CEO fulfills a dual role as the chairman of the board. Further, board size ranges between 5 and 30 directors with an average of 15.53 directors (SD=5.72). The average MNE in the sample has 135,136.39 employees (SD= 115,306.12) and with an MNE Profitability measured in return on equity of 12.26%.

The 2015 sample used for testing the robustness comprises data of 1,558 directors. 417 directors are female while the majority of 1141 directors remain male. The individual ages range from 29 to 88 years. 453 directors are foreign and 1,105 nonforeign who in total come from 45 different countries. Board tenure varies between 0 and 40.8 years. With 793 dependent

Variable Min Max Mean SD Min Max Mean SD

CSR Score (raw data) 28.71 95.70 89.96 10.47 28.71 95.70 89.96 10.47

CSR Performance .40 22.23 12.47 4.76 .40 22.23 12.47 4.76 Gender (%) .00 54.54 25.37 10.40 .00 57.89 27.35 10.72 Age (%) 6.45 24.51 13.01 3.35 6.48 23.78 12.56 3.23 Nationality (%) .00 100.00 30.45 22.46 .00 100.00 31.30 22.73 Tenure (%) .00 134.94 72.99 19.20 .00 135.57 72.37 18.63 Board Independence (%) .00 92.31 53.99 25.77 .00 100.00 54.53 25.73 CEO Duality 0 1 .52 .50 0 1 .48 .50 Board Size 5 30 15.53 5.72 6 30 15.58 5.78 MNE Size 1,069 592,586 135,136.39 115,306.12 1,186 610,076 135,909.99 116,157.54 MNE Profitability (%) -32.95 82.55 12.26 14.68 -81.19 94.86 9.10 19.64 Industry Type 0 1 0.42 .50 0 1 0.42 .50

Table 1: Descriptive Statistics

2014 2015

(31)

25 and 765 independent directors, the numbers are nearly equally distributed. As expected, the board composition from 2014 to 2015 rarely changes and most mean values remain almost unchanged. However, there is a slight increase in female directors to an average of 27.35% and of foreign directors to 31.08%. Further, Board Independence depicts an extreme value of 100% which does not affect the mean value since this change was caused by a single company and one director. CEO Duality decreases to 48 MNEs in which the CEO performs a dual role.

Correlations

The correlation matrix is presented in table 2 (2014) and in Appendix E (2015) below. In the 2014 data set, the independent variables Gender, Tenure, and Nationality, and the control variable Board Size correlate significantly with CSR Performance. However, it is noticeable that the correlations with Tenure and Nationality are not significant in the 2015 data, but CEO

Duality becomes significant. Further, Board Independence appears to significantly correlate

with CEO Duality and Board Size. This observation supports prior theoretical approaches which use CEO Duality as a proxy for Board Independence. Further, Board Independence, in fact, seems to significantly correlate with all independent variables (except Tenure 2014) which is a positive indication of its potential moderation effect on the independent variables. Moreover,

Gender correlates significantly with Tenure and Nationality in the 2014 sample. However, since

(32)

26

Table 2: Correlation Matrix 2014

CSR Performance Gender Age Nationality Tenure

Board

Independence CEO Duality Board Size MNE Size

MNE

Profitability Industry Type

(33)

27 4.2 Regression Results

The results of the OLS regression for 2014 are presented in table 3. First, the effects of the control variables on CSR Performance are tested in model 1. Board Size is the only control variable which has significant explanatory power on the dependent variable in this sample (p≤.05). However, MNE Size basically has no effect (B=-.000000277; p=.95) and appears to create unnecessary noise in the model. Hence, this variable is dropped for the further analysis. In model 2, the independent variable Gender is separately added to the control variables to test hypothesis 1. Looking at the results, Gender depicts a positive effect in model 2 (B=.161) as well as in the general regression model 6 (B=.134) at a significant level (p≤.01). Both resulting models support hypothesis 1 which can hence be accepted.

Next, hypothesis 2 and the impact of Age on CSR Performance is assessed in model 3. However, neither the regression model nor the weak effect of the variable itself is significant (B=.083). The non-significant effect is further confirmed in the main regression analysis in model 6 (B=-.013) which even depicts a negative effect. Thus, Age diversity rarely does not seem to impact CSR Performance and hypothesis 2 finds no support.

To test the assumption of hypothesis 3, the variable Nationality is added in model 4. Contrary to the hypothesized effect, Nationality seems to have a negative, even though the small and slightly significant effect on CSR performance (B=-.041; p=.074). However, this outcome is not confirmed in the main regression model 6 which shows a negative but insignificant effect (B=-.018; p=.444). Since the effect has low statistical significance, the variable might simply lose its effect when adding more predictors. Due to the weakness and instability of the effect, hypothesis 3 is rather not supported.

(34)

28

Table 3: Regression Analysis 2014

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10

Independent Variables Gender .161 (.044)*** .134 (.051)*** .126 (.053)** .112 (.1054)** .144 (.050)*** .136 (.084) Age .083 (.141) .004 (.138) .028 (.145) .069 (.151) -.036 (.138) .058 (.194) Nationality -.041 (.023)* -.018 (.023) -.018 (.024) -.032 (.027) -.013 (.023) -.005 (.037) Tenure .051 (.025)** .021 (.026) .022 (.026) .021(.085) .027 (.026) .047 (.032) Moderating Variables Board Independence .014 (.024) -.057 (.134) CEO Duality 1.977 (1.030)* 1.997 (1.050)* Interactions

Gender X Board Independence -.003 (.002)*

Age X Board Independence .005 (.006)

Nationality X Board Independence .001 (.001)

Tenure X Board Independence .001 (.002)

Gender X CEO Duality .083 (.110)

Age X CEO Duality -.231 (.282)

Nationality X CEO Duality -.026 (.049)

Tenure X CEO Duality -.085 (.057)

Control Variables

Board Size .189 (.086)** .240 (.079)*** .183 (.083)** .141 (.086) .152 (.083)* .196 (.089)** .224 (.101)** .183 (.110)* .129 (.095) .147 (.098)

MNE Size -2.770E-07 (.00) excl. excl. excl. excl. excl. excl. excl. excl. excl.

MNE Profitability .013 (.033) .011 (.031) .015 (.033) .025 (.033) .019 (.032) .020 (.032) .018 (.032) .013 (.033) .032 (.032) .033 (0.033) Industry .306 (.966) .753 (.912) .311 (.963) .704 (.975) .275 (.945) .840 (.941) .903 (.951) 1.055 (.963) .884 (.928) 1.032 (.981) Constant 9.294 (1.479)*** 8.288 (1.401)*** 9.319 (1.465)*** 9.689 (1.460)*** 9.752 (1.455)*** 8.821 (1.494)*** 8.396 (1.666)*** 9.227 (1.790***) 8.675 (1.475)*** 8.401 (1.553)*** R Square .054 .170 .059 .085 .093 .181 .184 .226 .213 .242 Adjust. R Square .014 .135 .019 .047 .055 .119 .112 .119 .144 .137 F-stat. 1.347 4.869*** 1.479 2.210* 2.468* 2.907*** 2.568** 2.119** 3.078*** 2.314** CSR Performance

(35)

29 Model 7 shows that the moderating variable Board Independence rarely has a direct effect on the dependent variable. Through adding the interaction terms in model 8, it becomes obvious that the variable itself does not only rarely have an effect (B=-.057; p=.134), it also rarely and insignificantly interacts with the independent variables against prior expectations due to the correlations of the variables. Only the interaction with Gender shows a marginally significant but negative effect (B=-.003; p=.095). Further, the interactions with Age (B=.005; p=.384),

Nationality (B=.001; p=.403) and Tenure (B=.001; p=.670) are very small and insignificant.

Hence, Board Independence does not seem to positively moderate the diversity effects on CSR

Performance and hypotheses 5a-d need to be rejected.

Model 9 shows that if CEO Duality is present, MNEs seem to have a significantly higher

CSR Performance as opposed to MNEs where this is not given (B=1.977; p=0.058). Adding

the interaction terms in model 10, the results in fact indicate negative relationships on Age (B=-.231; p=.414), Nationality (B=-0.026; p=0.600) and Tenure (B=-0.085; p=0.141) while the effect on Gender is positive (B=0.083; p=0.456). However, since all p-values are insignificant,

CEO Duality does not appear to negatively moderate the relationship between the board

diversity effects and CSR Performance. Thus, hypotheses 6a-d are rejected. 4.3 Robustness Test

Robustness Test 1: 2015

The 2015 analysis (Appendix F) generally seems to be consistent with the 2014 analysis. Yet, the models depict slight differences. Model 1 confirms the decision to exclude MNE Size from the analysis. Models 2 (B=.114; p=.011) and 6 (B=.093; p=.041) support the significant positive effect of gender diversity, thus confirming hypothesis 1. The effect of Age remains insignificant (model 3: B=.139; p=.351; model 6: B=.025; p=.874) rejecting hypothesis 2. In contrast to 2014, and complementing the argumentation to reject hypothesis 3, the impact of Nationality is insignificant in both models (model 4: B=-.034; p=.127; model 6: B=-.023; p=.327) which does not support hypothesis 3. Further, Tenure also does not show a significant effect separately anymore (model 5: B=0.025; p=0.346; model 6: B=0.013; p=0.427). The interactions with

Board Independence continue to be insignificant including the interaction with Gender thus

(36)

30 a negative effect (B=-0.139; p=0.02). This result supports hypothesis 6d while hypotheses 6a-c are still reje6a-cted. All other intera6a-ctions stay insignifi6a-cant. Hen6a-ce, apart from hypothesis 6d, the 2015 analysis supports the results of 2014.

Robustness Test 2 & 3: Blau Index

The robustness tests using the Blau’s diversity index for Gender, Age, Nationality, and Tenure (Appendix G for 2014; Appendix H for 2015) validate most of the previous outcomes although the coefficients are noticeably changed. Gender diversity continues to impact CSR Performance significantly in both years separately tested and in the regression model at a p≤.01 level thus confirming hypothesis 1. Further, hypothesis 2 and 3 remain not supported as neither Age nor

Nationality depicts significant values in any of the conducted regressions. The separately

regressed impact of Tenure on CSR Performance, however, is significant for both years. In the 2014 model, model 5 creates a moderately significant coefficient of 6.365 (p=.086) and of 8.701 (p=.032) in 2015. However, the main regression in model 6 in the 2015 analysis also depicts a significant effect (B=7.224; p=.051). These results would confirm a partial support hypothesis 4. The inconsistent findings among the analyses depending on the year and calculation of the variable are further evaluated in the section ‘5. Discussion’.

Board Independence interacts negatively with all independent variables except with Age. Most effects remain insignificant consistent with the previous results. However, the

significant interaction with Gender in the previous 2014 analysis cannot be confirmed and the new 2015 analysis depicts a significant negative interaction with Tenure (B=-.549; p=.012) which makes the variable Tenure significant as well (B=8.445; p=.022). Nonetheless, hypotheses 5a-d are still rejected since the effects are either not significant or are negative in contrast to the prediction. Similarly, the robustness tests of the moderating effect of CEO

Duality delivers insignificant results and do not support hypotheses 6a-d. All interactions apart

from Tenure in the 2014 analysis deliver a negative result while all interactions are negative in the 2015 analysis. In contrast to the prior 2015 analysis without the Blau index values, Tenure and CEO Duality show no significant interaction and the variable itself remains insignificant.

(37)

31

5. Discussion

The key objective this study is to contribute to the field of research on the influence of board diversity on CSR performance by examining the effects of gender, age, nationality and tenure diversity in boards of Europe’s 100 largest MNEs. By testing whether the structural characteristics board independence and CEO duality have a moderating role on the individual diversity effects, this thesis expands the knowledge on how different types of diversity interact and gives new insights these relationships. Generally, the findings of the study are consistent with existing research which has presented mixed results in the effects of board diversity. While gender diversity may be an important predictor of a firm’s CSR performance, age, nationality and tenure diversity do not seem to have a meaningful impact. Besides, the interactions with board independence and CEO duality do not depict explanatory power. Therefore, although diversity in boards may be important, it is necessary to differentiate between different types of diversity.

Consistent with prior studies (e.g. Bear et al., 2010; Byron & Post, 2016; Webb, 2004) and with public debates about female representation on corporate boards, gender diversity has a significant positive effect on CSR performance in this study and is the variable with the highest explanatory power. This result holds true independent of which measurement method and which year’s data set is used. This finding could confirm the argumentation that female directors generally bring in more sensitivity for CSR issues and foster the firm’s CSR performance (Boulouta, 2013; Hafsi & Turgut, 2013). However, although reverse causality is addressed through the time lag in the study (Carter et al., 2010), a lag of one and two years may not entirely eliminate it. An MNE which is known for CSR engagement might be a more attractive workplace for women because this may reflect their personal orientation for social responsibility better or the MNE may be more committed to integrating women in leadership positions. Furthermore, the social dimension of the Asset4 data includes scores concerning female employees (e.g. percentage of female employees and managers, equal opportunities, maternity leave, family friendliness etc.). Thus, female directors on the board may either be attracted by a firm performing well in these categories or may be responsible for an improved score on these dimensions. Therefore, it would be interesting to investigate the cause-outcome relationship of this phenomenon further with data sets over longer periods.

(38)

32 Since the analysis does not provide meaningful implications for this variable, age diversity may be an inadequate predictor of CSR performance due to the following reasons. First, the sample shows that there generally is a low age variation in the examined boards. Most of the directors are in the age group between 50 and 69 years while the others are underrepresented. Hence, the expected differences among the boards must be small. Second, since other age groups are rarely included in boards, this may affect team dynamics and effectiveness of the board’s decision-making process. Younger directors may not express or push their views to their older counterparts due to seniority effects or respect for the older director’s experience. Simultaneously, older directors may not listen to their younger colleagues due to perceived lower expertise. This would hinder a positive effect of combining different levels of experience, generational opinions, and values and would support social identity theory of creating distinct social groups (Tajfel, 1978; Turner et al., 1987). Finally, while there are arguments which support that older directors are more likely to support CSR efforts since moral reasoning develops over time and the outlook for societal environment increases (Forte, 2004; Kets de Vries & Miller, 1984; Ruegger & King, 1992), studies also show that younger directors may be more concerned about environmental issues and societal well-being (Ali et al., 2014; Gardyn, 2003). Both groups show indices to promote CSR performance and the actual age diversity may therefore not be of importance. Team dynamic, individual personality and characteristics, as well as the firm-specific context, may be the more conclusive indicators.

(39)

33 to speak up for foreign directors who are usually a minority on boards or lead to an unwillingness to listen for domestic directors. Research on minorities in groups suggests that such social differences lower the consideration of minorities in decision-making processes (Westphal and Milton, 2000). Further, unfamiliarity with stakeholders in the MNE’s home country may be an obstacle for foreign directors and may lead to wrong assumptions about their interests (Masulius et al., 2012). Second, this study focused on large MNEs operating in various institutional environments. Although it may be helpful to have foreign directors on the board for MNEs to have a broader insight into international relations, a higher percentage of foreign directors could indicate a higher internationalization degree of the firm including increased operations in countries with lower social and environmental standards (e.g. regarding working hours, minimum wages, safety, pollution, etc.) which may lower the overall CSR performance (Strike, Gao & Bansal, 2006; Surroca, Tribó & Zahra, 2013). Hence, it would be interesting to investigate this relationship more in detail by looking at the internationalization degree of the assessed firms.

(40)

34 integrated in the strategic decision-making of the MNE. At the same time, due to their familiarity with the firm’s management, longer-tenured directors may not discuss topics which are not within the firm’s former perspectives or may have a restricted view on possibilities. The relationship between tenure diversity and CSR performance should be further examined with larger samples to examine how the significance of the effect changes to receive more meaningful results.

Board Composition and CSR: The Influence of Board Diversity on CSR Performance of European MNEs and the Moderating Effects of Board Independence and CEO Duality