BOARD DIVERSITY AND CORPORATE SOCIAL PERFORMANCE

(1)

BOARD DIVERSITY AND CORPORATE SOCIAL

PERFORMANCE

Joost Blokhuis University of Groningen Faculty of Economics and Business

Supervisor Dr. B. van Oostveen June 2017

ABSTRACT

This study examines the effect of various board diversity aspects on corporate social performance. Moreover, I circumvent endogeneity issues by excluding corporate governance from corporate social responsibility ratings. The results of a sample of 381 European firms, covering the period 2002 to 2015, show that there is no effect of gender, age and board tenure diversity on corporate social performance. Expertise diversity and overall board diversity positively influence corporate social performance. The study provides an interesting remark on prior empirical research.

Keywords: Board of directors, gender, expertise, age, tenure, diversity, corporate social responsibility

(2)

I. INTRODUCTION

Popular theory states that increased board diversity positively affects firm performance. Extending board heterogeneity brings along directors with different ideas, backgrounds and experience which enhances decision making and ultimately increases firm performance. Considerable literature is available on the effect of board diversity on financial performance, whereas the effect of board diversity on corporate social performance (CSP) remains underexposed. Existing studies mainly focus on the US market and provide mixed results. This study adds to the available research by using a unique database based on European companies. Furthermore, the paper stresses the importance of carefully examining the pillars on which corporate social responsibility ratings are based.

Quoted from Deutsche Bank (2012): “In the early 2000s there emerged a renewed interest and desire for a more concrete definition of SRI to include corporate governance, in addition to financial, social and environmental factors.” Academics and investors increasingly emphasize good corporate governance in relation to a company’s risk and reward. Figure I shows that the various SRI strategies are growing rapidly. Sustainability themed investing increased from approximately €59 billion in 2013 to roughly €145 billion in 2015. Thus, it shows the growing importance of corporate social performance as a whole instead of just financial performance.

Figure 1: The evolvement of SRI strategies in Europe, 2013 compared to 2015

(3)

Various studies have examined the effect of board characteristics on CSP, however relatively less studies have examined board diversity. One of the first to examine the relationship is the paper of Siciliano (1996) in which expertise, gender and age diversity are examined. The results show a significant positive relationship of expertise and gender diversity to CSP, whereas there is no effect for age diversity. A variety of papers find a significant positive relationship between female board representation and CSP (see, e.g. Bear, Rahman and Post, 2010; Deschênes, Rojas, Boubacar, Prud’homme and Ouedraogo, 2015; Galbreath, 2016; Harjoto, Laksmana and Lee, 2015; Post, Rahman and Rubow, 2011). The research on age diversity and its effect on CSP is limited and inconclusive, Hafsi and Turgut (2013) find a significant negative relationship, contrary to their expectations. Additionally, Harjoto et al. (2015) find no relationship. Furthermore, Harjoto et al. (2015) and Hafsi and Turgut (2013) examine the effect of board tenure diversity on CSP, the former paper shows significant positive results, whereas the latter finds no significant relationship.

The contribution of this study to existing research is twofold. Firstly, CSR ratings are based on four pillars: social, environmental, governance and economic. All studies, with the exception of Harjoto et al. (2015), ignore the fact that board diversity is endogenous to the governance pillar of CSR ratings, as shown in Table A.I (see Appendix). This study shows the possible drawbacks that ignoring this endogeneity issue may pose by examining the effect of board diversity when the corporate governance score is excluded from CSP. Additionally, the study addresses the endogeneity problem in the form of omitted variables by incorporating firm-fixed and year-fixed effects. Secondly, existing papers mostly focus on U.S. firms and yield inconclusive results for some of the board diversity aspects. By using a unique European dataset I examine whether the results are similar for European firms.

(4)

II. LITERATURE REVIEW

This section consists of three parts. First, a brief history about CSR is given as an introduction to CSR. Secondly, three essential theories that provide rationale for the relationship between board diversity and CSP are presented. Lastly, prior research on the relationship between various aspects of board diversity and CSP will be discussed.

History of CSR

Corporate social responsibility may be defined in many ways. Dahlsrud (2008) examined the various definitions of CSR and states that the most used definition is the following: “CSR is a concept whereby companies integrate social and environmental concerns in their business operations and in their interaction with their stakeholders on a voluntary basis” (Commission of the European Communities, 2001). The CSR movement gained ground in the 1960s and focused mainly on environmentally issues such as carbon dioxide emissions and deforestation (Katsoulakos, Koutsodimou, Matraga and Williams, 2004). Several national environmental protection agencies and the United Nations Environment Programme were established (UNEP). The scope of CSR kept growing and throughout the ‘70s and ‘80s social issues were highlighted, such as poverty, social inequity and terms of trade. Several countries mandated large companies to report on their social and environmental performance. Companies started to introduce quality and environmental management systems. Moreover, institutions like the Council on Economic Priorities (USA) and others began to publicly rate companies. The ‘90s saw the CSR movement growing even larger, fostered by increased transparency of business activities through the modern information and communication technologies. From the 2000s and onwards CSR practices and policies are common throughout the world and nearly every large companies issues a yearly sustainability report.

Stakeholder theory

(5)

Heterogeneous boards provide different perspectives which address a wider range of stakeholders than homogeneous boards, leading to improved CSP.

Resource dependence theory

Pfeffer and Salancik (1978) argue that a firm is dependent on external resources, which are scarce and critical to their survival. Organizations seek to establish relationships with others in order to obtain these valuable resources. Furthermore, firms aim to acquire control over these resources to minimize their dependence on other organizations and maximize the dependence other corporations have on themselves. In the context of board diversity, directors are valuable resources themselves or they provide them through environmental relations. Furthermore, a well-diversified board provides legitimacy to the outside world, reflecting the norms and values a company holds, which in turn is beneficial to a firm’s reputation.

Human capital theory

Becker (1964) popularized the term ‘human capital’, it addresses the role of an individual’s unique level of education, skills and experience that can be used to improve firm performance. Terjesen, Singh and Vinnicombe (2009) investigate the view that new male directors provide more human capital than their counterparts. However, their results suggest that female directors provide fairly similar human capital and sometimes even additional human capital to their male peers. Kesner (1988) argues that each director brings a unique set of human capital resources to the board. Companies seek board members that provide complementary human capital resources to the existing board. A diversified board should provide a unique mix of human capital that results in higher CSP.

Prior research

There is extensive research available on the relationship between board characteristics and CSP, whereas relatively little research focuses on board diversity. The research on the former does provide interesting insights, therefore they are included in this literature review.

Gender diversity

(6)

more likely to come from non-business careers as opposed to their male counterparts. Women have been found to have particular skills in supporting and maintaining relationships, they also readily focus on the needs of others rather than on their own needs (Galbreath, 2016). Therefore, women may be more sensitive than men, to not only shareholders, but to all stakeholders of a firm and in turn enhance the company’s CSP. Multiple studies examine the effect of female representation on CSP. Galbreath (2016) finds a significant effect by using data of 295 Australian Securities Exchange 300 firms, covering the period 2005 to 2009. The results are in line with several studies (Bear et al., 2010; Deschênes et al., 2015; Post et al., 2011; Zhang, Zhu and Ding, 2013). Fewer studies examine the effect of gender diversity on CSP. Harjoto et al. (2015) examine 1,489 U.S. firms from 1999 to 2011 and find a significant positive effect. This is in line with the results Hafsi and Turgut (2013) in a study of 100 randomly selected S&P 500 firms in 2005.

Age diversity

The age of an individual correlates with work experience. An older individual is expected to have more work experience than a younger individual. Younger generations of directors are often seen as more sensitive to environmental and ethical issues as a matter of logic and principle (Hafsi and Turgut, 2013). Combining the perspectives of both young and old directors may lead to more balanced decision-making and increased CSP.

The research on this relationship is limited and inconclusive. The study of Siciliano (1996), with data from 240 YMCA organizations for the year 1989, does not find a relationship of age diversity to social performance. However, organizations with increased age diversity show higher levels of philanthropy, which implies more concern for the welfare of others. Hafsi and Turgut (2013) expected to find a positive relationship, however their results show a significant, negative effect of age diversity on CSP. The authors argue that age diversity possibly leads to polarization, a sort of generation conflict. There may be a certain level of age diversity, resulting in an undesirable large generation gap, which hinders decision-making processes. Furthermore, Harjoto et al. (2015) do not find a relationship between age diversity and CSP.

Expertise diversity

(7)

their study suggest that firms benefit from expertise diversity, only when it consists of a subset of backgrounds (lawyers, consultants, accountants, bankers and outside CEOs on the board). Worthy and Neuschel (1984) investigate “underlying characteristics of companies and their chief executive officers which contribute to the shaping of board organizations and behaviour”. The authors stress that companies should attract directors who can bring new insights and perspectives to the board. A variety of backgrounds, experience and skills provides greater idea stimulation and avoid excessive emphasis on any one business aspect. Harjoto et al. (2015) examined the effect of expertise diversity and CSP. The results show a significant positive relationship. Furthermore, the results of Siciliano (1996) reveal increased social performance when board members had greater occupational diversity.

Board tenure diversity

Board tenure is related to a board member’s knowledge of the firm. Experienced board members are more familiar with firm’s strategic issues and management team practices (Hafsi and Turgut, 2013). However, long term directors may have stale ideas that hinder creative and innovative thinking. Berberich and Niu (2011) investigate 12,258 directors of S&P 1500 firms in 2009 and find that a long term relationship between directors and executives shifts the directors’ allegiance from shareholders to the executives, resulting in lower CSP. Therefore, a balanced board with varying board tenures may result in better decision-making and improved CSP. Research studies that examine this relationship yield inconclusive results. Harjoto et al. (2015) find a significant positive effect, whereas Hafsi and Turgut (2013) find an insignificant positive relationship.

III. METHODOLOGY

To determine the effect of board diversity on CSP the following equation is estimated:

CSP_𝑖,𝑡 = 𝛼₀+𝛼₁DIVERSITY_𝑖,𝑡+𝛼₂CONTROLS_𝑖,𝑡+𝑐_𝑖+𝑣_𝑡+∈_𝑖,𝑡 (1)

where, i is an index for the firm, t is a time index (year), CSP is the CSR rating, 𝛼₀ is a constant, 𝛼1 and 𝛼2 are coefficient vectors for DIVERSITY and CONTROLS, the board

(8)

First, model (1) is estimated without firm-fixed and year-fixed effects as a benchmark model. Next, model (1) is estimated with firm-fixed and year-fixed effects1 to correct for omitted variables. This approach avoids problems resulting from residual dependency of the covariates, caused by a firm-fixed effect. Moreover, it circumvents problems arising from conditions that affect the performance of all companies in a certain year, a year-fixed effect. Furthermore, the model is estimated with robust standard errors to correct for autocorrelation2 and heteroscedasticity3. Finally, the results of CSR scores with the corporate governance pillar are compared to the results without the corporate governance pillar, to point out the importance of carefully examining the potential endogeneity problems Asset4 ratings contain.

IV. DATA

The initial sample for this study comprises the 1081 constituent firms of the Asset4 Europe database. The database provides CSR ratings, so-called Asset4 ratings, that are built upon four pillars: Environmental, Social, Governance and Economic. The scores are based on more than 250 ESG key performance indicators and over 750 data points covering all the pillars. Figure A.I (see Appendix) provides an overview of the categories on which the Asset4 scores are based. This study uses the available data covering the years 2002 to 2015. Furthermore, data concerning gender and expertise characteristics are retrieved from the Asset4 database. The remaining board characteristics, age and board tenure, are retrieved from the BoardEx database. BoardEx provides in-depth board information concerning a wide range of

characteristics. Finally, the firms’ financial characteristics that form our control variables are collected from Datastream. The following table shows an overview of the definitions and sources of the variables used in this study.

1_{To discriminate between random and fixed effects, the Hausman test is performed. The random effects model is} preferred under the null hypothesis because it has higher efficiency, whereas under the alternative hypothesis the fixed effects is preferred since it is at least consistent.

2_{The Wooldridge test detects the presence of autocorrelation. The null hypothesis indicates that the covariance} between the error terms over time is zero. The alternative hypothesis states that the errors terms are correlated over time.

(9)

Table I

Definitions and sources of the firm and board characteristics for 1081 European companies from 2002 to 2015

Definitions Source

Corporate social performance

Asset4 rating Overall CSR score ranging from 0 to 100 Asset4 Environmental rating CSR score for the Environmental pillar ranging from 0

to 100

Asset4

Social rating CSR score for the Social pillar ranging from 0 to 100 Asset4 Governance rating CSR score for the Governance pillar ranging from 0 to

100

Asset4

Economic rating CSR score for the Economic pillar ranging from 0 to 100

Asset4

Board characteristics

Gender Categorization of board member as female or male Asset4

Age Age of individual board members BoardEx

Expertise Board members who do or do not have either an industry specific background or a strong financial background

Asset4

Board tenure Board tenure of individual board members BoardEx

Firm characteristics

Total assets (ln) Natural logarithm of total assets at the end of the year Datastream Total sales (ln) Natural logarithm of total sales at the end of the year Datastream Return on assets (ln) Natural logarithm of return on assets at the end of the

year

Datastream

Firm age (ln) Natural logarithm of the number of years since incorporation

Datastream

Board size Number of board members Datastream

CEO duality Dichotomous variable that equals 1 if the CEO is also chairman of the board and 0 otherwise

Asset4

Corporate social performance

The dependent variable, CSP, will be measured in multiple ways. First, the regular Asset4 score serves as the dependent variable. Secondly, a CSP score excluding Governance, is constructed to address the endogenous relationship of board structure to the Governance pillar (as seen in Figure A.I4). The CSP score excluding Governance is the average of the scores of the remaining pillars (Environmental, Social and Economic) and will be referred to as

‘Asset3’ rating in the remainder of this study. Finally, the individual pillars will serve as CSP performance measures to see in which ways board diversity affects the separate pillars on

4

(10)

which the Asset4 scores are based. An overview of the evolvement of the CSP measures from 2002 to 2015 is shown in Table II. Overall CSP increases with 16.1% from 56.47 to 65.58 in 13 years. However, when governance is excluded the CSP increase is smaller, from 59.24 in 2002 to 64.48 in 2015 (8.8%).

Table II

Average CSR ratings in the period of 2002 to 2015

Asset4 indicates total corporate social performance. Asset3 is the average of Environmental, Social and Economic. Environmental, Social, Governance and Economic indicate environmental-, social-, corporate

governance- and economic performance respectively.

Asset4 Asset3 Environmental Social Governance Economic

2002 56.47 59.24 60.15 61.74 38.38 56.06 2003 55.58 58.98 61.44 61.55 36.61 53.99 2004 57.23 57.83 58.39 61.63 46.85 53,57 2005 57.20 56.85 57.13 61.52 49.89 52,09 2006 57.24 56.99 57.24 61.00 50.93 52.90 2007 59.62 59.23 59.77 61.27 52.91 56.75 2008 60.83 60.68 62.01 64.06 50.59 56.06 2009 64.34 63.16 64.39 65.53 55.67 59.67 2010 67.31 65.21 65.62 66.69 58.86 63.41 2011 66.01 63.81 64.52 65.63 60.29 61.24 2012 66.72 64.62 64.88 65.52 59.97 63.49 2013 66.32 64.23 65.38 66.13 58.92 61.20 2014 65.90 63.77 64.81 66.21 58.44 60.32 2015 65.58 64.48 65.28 66.02 56.68 61.98 Board diversity

To measure the effect of board diversity on CSP, we have to construct diversity indices for the four independent variables. Board diversity is estimated using Blau’s index of heterogeneity, which is calculated as follows.

𝐵 = 1 − ∑ 𝑃𝑖2 𝑘 𝑖=1

(2)

(11)

years), 2, 3, 4, 5 and 6 (more than 15 years). Blau’s index has a maximum that increases by the number of categories, which can be calculated by the following formula:

𝐵_𝑚𝑎𝑥= 1 − 1/𝐾 (3)

where, Bmax is the theoretical maximum of Blau’s index and K is the number of categories. and diversity thus has a theoretical maximum of 0.5 with 2 categories, whereas age diversity has a maximum of 0.83 with 6 categories. To improve comparability of the diversity measures, the Blau index is converted by the formula of Agresti and Agresti (1978):

𝐵_𝑎𝑑𝑗 = 𝐵 𝑥 𝐾/(1 − 𝐾) (4)

where, Badj is the adjusted form of Blau’s index with a maximum of 1 and K is the number of categories.

Table III shows the evolvement of board diversity throughout the years. Gender diversity strongly increases by roughly 293.8% from 0.16 in 2002 to 0.63 in 2015. Age diversity is almost static, whereas expertise diversity increases from 0.47 to 0.78 (66.0%) over thirteen years. Board tenure diversity increases slightly from 0.66 to 0.72 (9.1%). Combined, total board diversity has increased significantly over the sample period, from 0.49 in 2002 to 0.73 in 2015 (49.0%).

Table III

Average board diversity in the period of 2002 to 2015

Board diversity is measured by using Blau’s index of heterogeneity.

Overall Gender Age Expertise Board tenure

(12)

Control variables

There may be various factors, beyond board diversity, that drive a firm’s CSP. To increase the accuracy of the models, this study controls for firm size, financial performance, firm age, board size and CEO duality. Firm size is measured by the natural logarithm of total assets and the natural logarithm of total sales. Udayasankar (2008) argues that larger firms tend to be more visible and therefore more socially responsive. Smaller firms have less resources which makes it less likely and unviable to engage in CSR activity. Thus, a positive relationship between firm size and CSP is expected. The results of studies by Galbreath (2016) and Harjoto et al. (2015) are in line with the expectation. Financial performance is measured by the natural logarithm of the return on assets5. Waddock and Graves (1997) find a positive relationship between financial performance and corporate social performance. Popular reasoning behind it is that firms with slack resources, generated by strong financial performance, have more freedom in investing in CSR activities. Additionally, the results of Bear et al. (2010) and Harjoto et al. (2015) show a positive relationship between return on assets and CSP. Furthermore, this study controls for firm age, measured by the natural logarithm of the years since incorporation. Badulescu, Badulescu, Saveanu and Hatos (2016) find evidence that young firms are less involved in CSR activities. Jawahar and McLaughlin (2001) argue that managements have to adjust their strategies to varying stakeholders in each stage of the organizational life cycle. The results of Harjoto et al. (2015) show no relationship between firm age and CSP, however Badulescu et al. (2016) point out that the relationship between firm age and CSR activity does not seem to be linear. Board size is measured as the number of directors on a firm’s board. Guest (2009) find evidence that large boards experience problems of poor communication and decision-making and therefore worse performance. According to Post et al. (2011) large boards potentially have worse decision-making processes than small boards, however large boards also have a greater array of knowledge and resources. The results of their study show an insignificant positive relationship between board size and CSP, which is in line with the estimates of Hafsi and Turgut (2013). Finally, we control for CEO duality, which is a dichotomous variable that equals 1 if the CEO is also chairman of the board and 0 otherwise. Rechner and Dalton (1991) investigate the

5_{As ROA can be both positive and negative, the log-modulus transformation by John and Draper (1980) is} performed which is estimated as follows:

(13)

relationship between CEO duality and firm performance and conclude that firms with independent leadership consistently outperformed those with CEO duality. Thus, a negative relationship between CEO duality and CSP is expected, however the results of Deschênes et al. (2015), Galbreath (2016) and Post et al. (2011) are inconclusive. An overview of the control variables (and methodology) of prior empirical research can be found in Table A.I (see Appendix).

Descriptive statistics

(14)

Table IV

Descriptive statistics of the corporate social performance, board- and firm characteristics

Mean Std. dev. Min Max Obs.

Corporate social performance

Asset4 rating 62.971 29.728 2.510 98.330 10,416 Asset3 rating 62.006 25.638 4.780 98.060 10,306 Environmental rating 62.805 29.526 8.570 97.50 10,416 Social rating 64.397 28.558 3.450 99.000 10,416 Governance rating 54.451 27.618 1.250 97.670 10,416 Economic rating 58.873 29.796 1.080 99.170 10,306 Board diversity Overall 0.676 0.137 0.000 0.949 4,369 Gender 0.393 0.326 0.000 1.000 10,379 Age 0.764 0.132 0.000 0.992 5,417 Expertise 0.759 0.290 0.000 1.000 8,416 Board tenure 0.713 0.238 0.000 1.000 5,424 Firm characteristics Total assets (ln) 15.788 2.253 0.000 23.119 10,403 Total sales (ln) 15.051 2.154 0.000 22.437 10,392 Return on assets (ln) 0.060 0.081 -0.772 1.306 10,216 Firm age (ln) 4.026 0.9206 0.000 6.299 7,487 Board size 11.049 4.242 1.000 33.000 10,391 CEO duality 0.222 0.416 0.000 1.000 10,293

(15)

(16)

V. RESULTS

This section presents the results of the statistical models used in this study. Firstly, I discuss the results of ordinary-least squares (OLS) regressions without firm-fixed and year-fixed effect. Secondly, the results of OLS regressions including firm-fixed and year-fixed effects are discussed.

Table VI presents the results of OLS regressions of overall board diversity on the various measures of CSP. In Column 1, the Asset4 CSR score is used as a performance measure. Column 2 uses the Asset3 CSR score as a performance, excluding Governance from the Asset4 rating in order to correct for possible endogeneity. Column 3, 4, 5 and 6 use each of the individual pillars of the Asset4 rating as a performance measure, respectively Environmental, Social, Governance and Economic. Robust standard errors are used in order to correct for heteroscedascity6 and autocorrelation7. The model is checked for robustness8. The results presented in Table VI suggest that board diversity has a significant positive effect on all the measures of CSP, at the 1% significance level. An increase of board diversity by 1 unit, which is the same as transitioning from a fully homogeneous to a fully heterogeneous board of directors, leads to an increase of the Asset4 score by 37.39 points.

Table VII presents the results of OLS regressions with gender, age, expertise and board tenure diversity as the independent variables, instead of overall board diversity. We see that gender diversity has a significant positive (p < 0.01) effect on all measures of CSP. Furthermore, Columns 1, 2, 4 and 6 show that age diversity has a positive effect on the Asset4, Asset3, Social and Economic ratings, at the 1% significance level. The effect of age diversity on the Environmental rating is positive as well, although only significant at the 10% level. Additionally, age diversity has a relatively small insignificant positive effect (coefficient of 0.363) on the Governance rating which makes sense as the age of board members is not taken into account when rating the board structure and thus corporate governance. Moreover, the results of Table VII indicate that expertise diversity has a significant, positive effect on the Asset4, Asset3, Social, Governance and Economic ratings at the 1%, 5%, 5%, 1% and 5% levels respectively. Expertise diversity appears to have an insignificant positive effect on the

6

The results of the Breusch-Pagan and Cook-Weisberg tests reject the null hypothesis of homoscedascity at the 1% signficance level.

7_{Using the Wooldridge test, the null hypothesis of no autocorrelation is rejected at the 1% significance level.} 8

(17)

Environmental rating. Finally, the results in Columns 1, 2, 3, 4 and 6 suggest that board tenure diversity has a significant positive effect on Asset4, Asset3, Environmental, Social and Economic scores at a 1%, 1%, 10%, 5% and 1% level respectively. Additionally, board tenure diversity has an insignificant positive effect on the Governance score. The result is not unsurprising as board tenure diversity is not taken into consideration for the Governance score, only the maximum number of years a member is allowed to be on the board is.

Table VI

Corporate social performance and board diversity in European firms, 2002-2015

This table presents the results of ordinary least-squares regressions excluding firm-fixed and year-fixed effects, using a sample of 393 European firms and covering the period 2002 to 2015. Board diversity data are retrieved from the Asset4 subset of Datastream and BoardEx, data regarding firm characteristics are retrieved from Datastream. The columns show the results where the dependent variables are Asset4, Asset3, Environmental,

Social, Governance and Economic respectively. Asset4 is the total CSR score. Asset3 is the CSR score when

governance is excluded. Environmental is the CSR score for the environmental pillar. Social is the CSR score for the social pillar. Governance is the CSR score for the governance pillar. Economic is the CSR score for the economic pillar. Board diversity is the average of Blau indices for gender, age, expertise and board tenure. Total

assets(ln) is the natural logarithm of total assets. Total sales (ln) is the natural logarithm of total sales. Return on assets (ln) is the natural logarithm of return on assets. Firm age (ln) is the natural logarithm of number of years

since incorporation. Board size is the number of directors on a board. CEO duality is a dichotomous variable that equals 1 if the CEO is also chairman of the board and 0 otherwise. All variables are measured at the end of the year. ***, **, * represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are displayed in parentheses below the coefficients.

(18)

Table VII

This table presents the results of ordinary least-squares regressions excluding firm-fixed and year-fixed effects, using a sample of 393 European firms and covering the period 2002 to 2015. Board diversity data are retrieved from the Asset4 subset of Datastream and BoardEx, data regarding firm characteristics are retrieved from Datastream. The columns show the results where the dependent variables are Asset4, Asset3, Environmental,

governance is excluded. Environmental is the CSR score for the environmental pillar. Social is the CSR score for the social pillar. Governance is the CSR score for the governance pillar. Economic is the CSR score for the economic pillar. Gender diversity, Age diversity, Expertise diversity, Board tenure diversity are the Blau indices of gender, age, expertise and board tenure, respectively. Total assets(ln) is the natural logarithm of total assets.

Total sales (ln) is the natural logarithm of total sales. Return on assets (ln) is the natural logarithm of return on

assets. Firm age (ln) is the natural logarithm of number of years since incorporation. Board size is the number of directors on a board. CEO duality is a dichotomous variable that equals 1 if the CEO is also chairman of the board and 0 otherwise. All variables are measured at the end of the year. ***, **, * represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are displayed in parentheses below the coefficients.

[1] [2] [3] [4] [5] [6] Gender diversity 11.376*** (1.433) 7.597*** (1.198) 8.597*** (1.451) 7.529*** (1.383) 14.586*** (1.343) 6.665*** (1.428) Age diversity 11.817*** (3.628) 11.747*** (2.996) 6.970* (3.633) 13.652*** (3.278) 0.363 (3.488) 14.618*** (3.631) Expertise diversity 6.680*** (1.629) 3.037** (1.362) 1.879 (1.716) 3.733** (1.535) 11.565*** (1.481) 3.500** (1.571) Board tenure diversity 8.443***

(19)

Regular OLS regressions may suffer from bias or inconsistency caused by omitted variables, therefore this part of the research corrects for the potential endogeneity issue by incorporating firm-fixed and year-fixed effects9. Table VIII and IX present the results with inclusion of firm-fixed and year-fixed effects, the tables are organized identical to Table VI and VII.

Table VIII presents the results for overall board diversity. Contrary to the results of Table VI, we find that board diversity has no significant effect on the Environmental and Economic score. The effect of board diversity on the Asset4 and Governance scores remain significantly positive at the 1% level. Whereas, the effects on Asset3 and Social scores are now less significant (p < 0.05), yet still positive.

The results for the separate diversity indices (age, gender, expertise and board tenure diversity) are shown in Table IX. We see that gender diversity has a significant positive effect on the Asset4 score, at the 10% level. Furthermore, gender diversity has no effect on the Environmental and Social scores. The effect on Governance and Economic scores is positive and significant at the 1% and 10% level respectively. Once the governance pillar is excluded from Asset4, i.e. Asset3 score, there is no effect of gender diversity on CSP, contrary to the results of previous studies on board diversity and CSP (Bear et al., 2010; Deschênes et al., 2015; Galbreath, 2016; Hafsi & Turgut, 2013; Harjoto et al., 2015; Post et al., 2011; Zhang et al., 2013). However, none of these studies account for the endogenous effect of board structure on the Governance pillar. Age diversity appears to have no effect on any of the performance measures. Hafsi and Turgut (2013) argue that age diversity possibly leads to polarization. There may be a certain level of age diversity, which results in an undesirable large generation gap that hinders decision-making processes and offsets the positive effects of age diversity. The effects of expertise diversity are significantly positive on Asset4, Asset3, Environmental, Social and Governance scores at the 1%, 5%, 10%, 1% and 1% levels respectively. Thus, expertise diversity positively affects CSP, which supports the results of Harjoto et al. (2015) and Siciliano (1996). Surprisingly, we find no effect of expertise diversity on the Economic rating as it is expected to positively influence financial performance, however other factors may significantly influence the rating as well (see Figure I in the Appendix). Finally, the results show that board tenure diversity has an insignificant negative effect on Asset4, Asset3, Environmental, Governance and Economic scores. The

(20)

regression of board tenure diversity on the Social score returns a significant negative effect, at the 5% significance level. These results are similar to the ones in the research by Hafsi and Turgut (2013).

These findings indicate that is important to look at the drivers of CSP and possible endogeneity problems. Once firm-fixed and year-fixed effects are included, the effects become less significant in general. Additionally, once Governance is excluded from the Asset4 scores, only board diversity and expertise diversity appear to significantly increase CSP.

Table VIII

This table presents the results of ordinary least-squares regressions including firm-fixed and year-fixed effects, using a sample of 393 European firms and covering the period 2002 to 2015. Board diversity data are retrieved from the Asset4 subset of Datastream and BoardEx, data regarding firm characteristics are retrieved from Datastream. The columns show the results where the dependent variables are Asset4, Asset3, Environmental,

governance is excluded. Environmental is the CSR score for the environmental pillar. Social is the CSR score for the social pillar. Governance is the CSR score for the governance pillar. Economic is the CSR score for the economic pillar. Board diversity is the average of Blau indices for gender, age, expertise and board tenure. Total

assets(ln) is the natural logarithm of total assets. Total sales (ln) is the natural logarithm of total sales. Return on assets (ln) is the natural logarithm of return on assets. Firm age (ln) is the natural logarithm of number of years

since incorporation. Board size is the number of directors on a board. CEO duality is a dichotomous variable that equals 1 if the CEO is also chairman of the board and 0 otherwise. All variables are measured at the end of the year. ***, **, * represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are displayed in parentheses below the coefficients.

(21)

Table IX

This table presents the results of ordinary least-squares regressions including firm-fixed and year-fixed effects, using a sample of 393 European firms and covering the period 2002 to 2015. Board diversity data are retrieved from the Asset4 subset of Datastream and BoardEx, data regarding firm characteristics are retrieved from Datastream. The columns show the results where the dependent variables are Asset4, Asset3, Environmental,

governance is excluded. Environmental is the CSR score for the environmental pillar. Social is the CSR score for the social pillar. Governance is the CSR score for the governance pillar. Economic is the CSR score for the economic pillar. Gender diversity, Age diversity, Expertise diversity, Board tenure diversity are the Blau indices of gender, age, expertise and board tenure, respectively. Total assets(ln) is the natural logarithm of total assets.

Total sales (ln) is the natural logarithm of total sales. Return on assets (ln) is the natural logarithm of return on

assets. Firm age (ln) is the natural logarithm of number of years since incorporation. Board size is the number of directors on a board. CEO duality is a dichotomous variable that equals 1 if the CEO is also chairman of the board and 0 otherwise. All variables are measured at the end of the year. ***, **, * represent significance at the 1%, 5% and 10% levels respectively. Robust standard errors are displayed in parentheses below the coefficients.

[1] [2] [3] [4] [5] [6] Gender diversity 3.894* (1.941) 1.958 (1.598) 1.113 (2.083) 0.654 (1.962) 7.762*** (1.818) 4.107* (2.175) Age diversity 0.572 (2.838) 0.039 (2.340) -1.596 (2.933) 3.781 (2.897) 1.675 (2.785) -2.069 (3.615) Expertise diversity 3.850*** (1.123) 2.030** (0.940) 2.257* (1.193) 3.127*** (1.067) 4.358*** (1.273) 0.705 (1.437) Board tenure diversity -0.797

(22)

VI. CONCLUSION

This study investigates the relationship between board diversity and corporate social performance. Agency theory, resource dependence theory and human capital theory articulate that board diversity positively affects CSP. Various studies examine the effect of gender diversity on CSP, whereas only a few studies investigate the effect of other diversity measures such as age, expertise and board tenure diversity. The majority of existing studies ignore endogeneity issues. Board structure (diversity) is endogenous to corporate governance, to account for this endogeneity issue the corporate governance part of the CSP score is excluded. Additionally, this study corrects for the omitted variable bias by incorporating firm-fixed and year-fixed effects.

The study uses a unique dataset which comprises 380 European firms and a total of 3140 observations, covering the period 2002 to 2015. The results of the OLS regression without fixed effects suggest that all the diversity measures positively affect Asset4 and Asset3 (excluding Governance) ratings. However, once the firm-fixed and year-fixed effects are included in the model the regression estimates show different results. The effect of overall board diversity remains positive, however the results for the other diversity measures differ. Age diversity appears to have no effect on CSP. Furthermore, board tenure has no influence on CSP except for a negative effect on the Social rating. Expertise diversity has a significant positive effect on CSP, even when corporate governance is excluded. Gender diversity positively affects the Asset4 rating, at the 10% level, however once corporate governance is excluded the effect of gender diversity becomes insignificant.

(23)

research shows that the existing theoretical framework needs to be improved and elaborated upon. Finally, this paper captures multiple board diversity aspects whereas it may be easier and better to thoroughly explore just one of them.

REFERENCES

Agresti A., Agresti B. F. (1978) Statistical analysis of qualitative variation. Sociological Methodology 9, 204–237.

Badulescu, A., Badulescu, D., Saveanu, T., Hatos, R. SMEs and social responsibility: does firm age matter? Opportunities and Risks in the Contemporary Business Environment, 963-975.

Barney, J., 1991. Firm resources and sustained competitive advantage. Journal of Management 17, 99-120.

Bear, S., Rahman, N., Post, C., 2010. The impact of board diversity and gender composition on corporate social responsibility and firm reputation. Journal of Business Ethics 97(2): 207-221.

Becker, G, 1964. Human capital theory. New York. Available at SSRN 1496221.

Berberich, G., Niu, F., 2011. Director busyness, director tenure and the likelihood of encountering corporate governance problems. CAAA Annual Conference 2011.

Commission of the European Communities, 2002. Corporate social responsibility: a business

contribution to sustainable development. Available at:

http://trade.ec.europa.eu/doclib/docs/2006/february/tradoc_127374.pdf

Dahlsrud, A., 2008. How corporate social responsibility is defined: an analysis of 37 definitions. Corporate Social Responsibility and Environmental Management 15(1), 1-13.

Deschênes, S., Rojas, M., Boubacar, H., Prud’homme, B., Ouedraogo, A., 2015. The impact of board traits on the social performance of Canadian firms. Corporate Governance 15: 293-305.

Deutsche Bank, 2012. Sustainable investing. Available at:

(24)

Eurosif, 2016. European SRI study 2016. Available at: http://www.eurosif.org/wp-content/uploads/2016/11/SRI-study-2016-HR.pdf

Freeman, R., 2010. Strategic management: a stakeholder approach. Cambridge University Press.

Galbreath, J., 2016. When do boards and management resources complement each other? A study of effects on corporate social responsibility. Journal of Business Ethics 136, 281-292.

Gray, S., Nowland, J., 2014. Professional expertise and board diversity. Available at SSRN 2289689.

Hafsi, T., Turgut, G., 2013. Boardroom diversity and its effect on social performance: conceptualization and empirical evidence. Journal of Business Ethics 112, 463-479.

Harjoto, M., Laksmana, I., Lee, R., 2015. Board diversity and corporate social responsibility. Journal of Business Ethics 132, 641-660.

Hillman, A., Cannella, A., Harris, I., 2002. Women and racial minorities in the boardroom: how do directors differ? Journal of Management 28, 747–763.

Jawahar, I., McLaughlin, G., 2001. Toward a descriptive stakeholder theory: an organizational life cycle approach. Academy of Management Review 26(3), 397-414.

John, J., Draper, N., 1980. An alternative family of transformations. Applied Statistics, 190-197.

Katsoulakos, P., Koutsodimou, M., Matraga, A., Williams, L., 2004. A historic perspective of

the CSR movement. CSRQuest Sustainability Framework. Available at

http://www.csrquest.net/uploadfiles/1D.pdf

Kesner, I., 1988. Directors' characteristics and committee membership: an investigation of type, occupation, tenure, and gender. Academy of Management Journal 31, 66-84.

Pfeffer, J., Salancik, G., 1978. The external control of organisations. New York 175.

(25)

Rechner, P., Dalton, D., 1991. CEO duality and organizational performance: a longitudinal analysis. Strategic Management Journal 12(2), 155-160.

Reuters, 2010. Asset4 ESG scores on Creditviews. Available at:

https://customers.reuters.com/community/fixedincome/material/asset4esgscores.pdf

Setó-Pamies, D., 2015. The relationship between women directors and corporate social responsibility. Corporate Social Responsibility and Environmental Management 22, 334-345.

Siciliano, J., 1996. The relationship of board member diversity to organizational performance. Journal of Business Ethics 15, 1313-1320.

Singh, V., Terjesen, S., Vinnicombe, S., 2008. Newly appointed directors in the boardroom: how do women and men differ?. European Management Journal 26(1), 48-58.

Udayasankar, K., 2008. Corporate social responsibility and firm size. Journal of Business Ethics 83(2), 167-175.

Waddock, S., Graves, S., 1997. The corporate social performance-financial performance link. Strategic Management Journal, 303-319.

(26)

APPENDIX

Figure A.I

Overview of the categories on which the Asset4 performance scores are based

(27)

Table A.I

Variables and methodology used in previous studies Bear, Rahman and Post (2010) Deschênes, Rojas, Boubacar, Prud’homme and Ouedraogo (2015)

Galbreath (2016) Hafsi and

Turgut (2013) Harjoto, Laksmana and Lee (2015) Post, Rahman and Rubow (2011)

Siciliano (1996) Zhang, Zhu

and Ding (2012) Gender x x x x x x x x Age x x x x Expertise x x Tenure x x x Assets x Sales x x ROA x x Firm age x Board size x x x x CEO duality x x x

Firm fixed effects x

Time fixed effects x

CSP x x x x x x x x

Acknowledged governance endogeneity

x

Methodology OLS using

lagged variables

OLS OLS OLS OLS, 2SLS OLS Partial

correlation Logistic regression Sample 689 Fortune's 2009 Most Admired List firms S&P/TSX 60 Index (60 Canadian firms) 295 Australian Securities Exchange 300 firms 100 S&P 500 firms