The influence of board characteristics on environmental disclosure quality and external assurance on environmental disclosures

The influence of board characteristics on environmental disclosure quality

and external assurance on environmental disclosures

Combined Master Thesis for Accountancy and Controlling

Lysanne Johanna Maria Pieper S2545039

University of Groningen Faculty of Economics and Business

Supervisor: Dr. N. Hussain Date: 24-06-2019 Word count: 9.440

Acknowledgment: First of all, I want to thank my supervisor Dr. Hussain for helping me with

this thesis and providing me with feedback. Further, I would like to thank KPMG for providing me the opportunity to write my thesis during an internship. Through this, I received additional support for writing my thesis and became familiar with KPMG. Lastly, I would like to thank my friends and family for the mental support.

2

ABSTRACT

This study investigates the association between board characteristics on environmental disclosure quality and the probability to buy external assurance on environmental reports. We investigated this by examining 6.840 firm-year observations on an international level from 2009 to 2017. We find that the proportion of women on board have a negative impact on the quality of environmental reporting and on the probability to buy external assurance. A possible explanation could be that women directors are only placed on the board for legitimacy reasons, resulting in selection of directors based on gender instead of the capabilities of that director. CEO duality have a positive impact on the environmental reporting quality and the probability to buy external assurance. A possible explanation for these positive relations could be that a powerful CEO wants to legitimate themselves by providing high-quality environmental disclosure and buying external assurance on these reports. Lastly, we address that external assurance on environmental reports will lower the quality of environmental disclosures. All our findings are inconsistent with our predictions and prior research, and therefore, our research suggests revisiting the effect of CEO duality and women on board on CSR and environmental issues. In summary, this research provides new insights into the determinants of environmental disclosure quality and environmental external assurance.

Keywords: Environmental disclosure quality; Environmental external assurance; The proportion

3

TABLE OF CONTENT

2.3.1 Proportion of women on the board & Environmental disclosure quality ... 12

2.3.2 Proportion of women on the board & probability to buy external assurance ... 13

2.3.3 External assurance & Environmental disclosure quality ... 14

2.3.4 CEO duality & Environmental disclosure quality ... 15

2.3.5 Moderating role of CEO duality ... 16

2.3.6 CEO duality & Probability to buy external assurance ... 16

2.4 CONCEPTUALMODEL ... 18 3 RESEARCH METHOD ... 19 3.1 SAMPLEDESIGN ... 19 3.2 RESEARCHDESIGN ... 20 3.2.1 Dependent variables... 20 3.2.2 Independent variables ... 21 3.2.3 Control Variable ... 21 3.3 MODELDEVELOPMENT... 24 4 RESULTS ... 25 4.1 DESCRIPTIVESTATISTICS... 25 4.2 CORRELATIONS ... 26

4.3 RANDOM-EFFECTPANELREGRESSION ... 28

4.4 PROBITREGRESSION ... 29

4.5 ADDITIONALANALYSES... 31

5 DISCUSSION AND CONCLUSIONS ... 33

5.1 SUMMARY,DISCUSSIONANDCONCLUSIONS ... 33

5.2 LIMITATIONSANDFUTURERESEARCH ... 35

6 REFERENCES ... 36

4

1 INTRODUCTION

Since the adoption of the climate agreement at the 21ste United Nations Climate Conference (COP21) in Paris in December 2015, taking action to scale down the impact on the environment is becoming more and more important (Clemencon, 2016). Companies are one of the key actors in achieving the goals set out in the COP211_{. Accordingly, companies communicate}

their environmental impact through specific reports (Aerts and Cormier, 2009; Gray and Bibbington, 2001). These reports enable society/stakeholders to evaluate the environmental performance of the companies (Aerts and Cormier, 2009). However, the quality of these reports is inconsistent due to the voluntary character and lack of regulation (Brammer and Pavelin, 2008). As a result of these low-quality reports, there have been multiple emissions scandals over the years2. As a consequence, stakeholders are paying more and more attention to the actual quality of environmental reports (KPMG, 2013; Coram, Monroe and Woodliff, 2009). In line with this shift in stakeholders’ attention, companies are focusing more and more on the quality of their reports (KPMG, 2013). Aligning to the stakeholders’ expectations could lead to a better reputation, more customer satisfaction and a positive corporate image which, in turn, leads to an increase in sales and revenue (Pfau et al., 2008; Weber, 2008; Bear, Rahman and Post, 2010).

A recommended tool to achieve high-quality environmental reports is external assurance. External assurance are the methods and processes employed by an external assurance provider to evaluate the organization's disclosures, as well as the underlying systems (Fonseca, 2010). It is commonly perceived as a valuable tool to ensure and improve the quality, credibility, and reliability of disclosures (Dando and Swift, 2003; Jones and Solomon, 2010). Some stakeholders specifically ask for external assurance on environmental disclosure (Moroney, Windsor and Aw 2012; Lash and Willington, 2007). Previous research on external assurance is mainly focused on financial reports and Corporate Social Responsibility (CSR) reports (Perego and Kolk, 2012; Skouloudis, Evangelinos and Moraitis, 2011; Simnett, Vanstraelen, and Chua, 2009). As such,

1 _{United Nations Climate Change: Business is key driver of Global Climate Action. Retrieved from:}

https://unfccc.int/news/business-is-key-driver-of-global-climate-action

2 _{Cultbizztech: 7 Corporate scandals we cannot forget. Retrieved from:}

5 current research on environmental assurance is limited. Nevertheless, it is very relevant to study because of the increased demand for high-quality reports by stakeholders.

Initially, the decisions of whether, what and how much to disclose is strongly influenced by the board of directors (Millstein, 1991). Previous research has shown the importance of board characteristics such as board composition, size, the existence of Chief Executive Officer (CEO) duality and independence on the functioning of the board (Agrawal and Knoeber, 1996; Yermack 1996; Laksmana, 2008). Correspondingly, board characteristics have an impact on reporting decisions, including the decision of providing high-quality environmental disclosure or buying external assurance on environmental reports or not (Barako and Brown, 2008; Brammer et al., 2008; Akbas, 2016). For example, in research conducted by Cheng and Courtenay (2006), they found that the proportion of independent non-executive directors influences voluntary disclosure quality. This study will focus on the board characteristics of gender diversity and CEO duality. Gender diversity is a ‘trendy topic’ in literature but underdeveloped in combination with environmental disclosure quality and environmental external assurance. The research on corporate governance emphasizes that CEO factors have an important influence on the functioning of the board (Laksmana, 2008). Moreover, Guldiken et al. (2018) suggest that the CEO can have a substantial impact on gender diversity on the board since the CEO lobbies for board candidates. The combination of gender diversity and CEO duality is interesting because from prior research on gender diversity, there is mostly a positive board outcome found, while for CEO duality is mostly a negative board outcome found. Combining these variables will probably have a different impact on environmental disclosure quality and the likelihood to buy environmental external assurance.

Nowadays, gender diversity receives increasingly more attention from companies. However, the board is in general still dominated by men and only a small number of women have a seat on the board, despite the positive effects of a more diverse workforce found in research (Merono-Gomez, Lafuente, and Vaillant, 2018; Seto-Pamies, 2015; Kakabadse et al., 2015; Fodio and Oba, 2012). For example, Bear et al. (2010) found that women more actively address stakeholders’ needs than men. Furthermore, Kurger (2009) and Braun (2010) suggest that companies with a higher proportion of women on board have a stronger social and environmental attitude. This suggests that women may be more engaged in social and environmental issues than men. However, the research of Roderiquez-Dominquez, Gallego-Avarez and Garcia-Sanchez

6 (2009) do not find evidence that women directors are more engaged in social and environmental issues. According to the literature on reporting, there is a positive relationship found between the proportion of women on the board and the quality of financial reporting and CSR disclosure (Krisnan and Parsons, 2008 Barako et al., 2008; Nekheli et al., 2017). However, Akbas (2016) do not find an association between the proportion of women on board and the environmental disclosure. According to the literature on external assurance, Liao, Lin and Zhang (2018) found that the proportion of female directors have a positive influence on the probability to buy external assurance on CSR disclosure.

As mentioned earlier, this study will also examine the effect of CEO duality on the quality of environmental disclosure and on the likelihood of buying environmental external assurance. CEO duality refers to a board structure in which the CEO of the firm is also the chairman of the board of directors (Rechner et al., 1991). The research on CEO duality is fragmented. On the one hand, the majority of literature stated that the presence of CEO duality reduces the board's ability to fulfill its monitoring function and lowers the effectiveness of the board (Rechner et al., 1991; Muttakin, Khan and Mihret, 2018). CEO duality will promote that the CEO-chairman will act more in her/his interest instead of the interests of the stakeholders (Jizi et al., 2014). Correspondingly, research has shown that CEO duality has a negative impact on the quality of financial and CSR disclosures (Abdullah, Ismail and Jumaluddin, 2008; Muttakin et al., 2018). On the other hand, some literature stated that the CEO-chairman has the tendency to legitimate themselves, which can be achieved by voluntary disclosure and/or buying external assurance. Correspondingly, Jizi, (2014) finds a positive relation between CEO duality and CSR disclosure.

In short, the key objective of this study is to investigate the effect of the proportion of women on the board and CEO duality on the quality of environmental reporting and the likelihood of buying external assurance on environmental reports. Here, the quality of the environmental disclosure can be seen as the decoupling of the environmental performance and performance disclosed in the environmental report. The quality of the environmental report is perceived as high when the disclosed environmental performance is similar to the actual environmental performance. In addition, we investigate the moderating role of CEO duality on the relationship proportion of women on the board and environmental disclosure quality. Therefore, the following research question will guide this study:

7

To what extent does the proportion of women on the board of directors (%) and CEO duality influence the quality of environmental disclosure and the likelihood of buying external

assurance on environmental reports?

The data used in this research is retrieved from Bloomberg database and Thomas Reuters ASSET4 database. These databases contain adequate information on board characteristics and environmental- performance and disclosure. The final sample includes 6.840 firm-year observations, which consist only companies who disclose their environmental performance. The data covers a period from 2009-2017, and the observations are divided over 54 countries. In our study, we find that the proportion of women on board have a negative influence on the quality of the environmental disclosure and the probability to buy external assurance on environmental reports, which is contradicting to our expectations. A possible explanation could be that women directors are only placed on the board for legitimacy reasons, resulting in selection of directors based on gender instead of the capabilities of that director. Further, we find that CEO duality has a positive influence on the quality of the environmental disclosure and the probability to buy external assurance, which is again contradicting to our expectations. A possible explanation for the positive influence of CEO duality could be that a powerful CEO wants to legitimate themselves by providing high-quality environmental disclosure and buying external assurance on these reports. Lastly, we address that the external assurance on environmental reports increases the gap between the environmental performance disclosed and the actual environmental performance, which is again contradicting to our hypothesis. This may also be a signal that external assurance is only used for legitimacy/signal purposes.

This study contributes to the literature in several ways. First of all, our study finds results which are contradicting with prior research and therefore, it provides new insight in the literature by revisiting the current claims in literature. Secondly, environmental disclosure in combination with board characteristics is not well researched yet. Previous studies are mainly focused on firm and industry levels (Patten, 2002; Berthelot, Cormier and Magnan, 2003; Hackston and Milne, 1996). This study adds to the existing research by examining the effect of the proportion of women on the board and CEO duality. Thirdly, this research uses the difference between the actual environmental performance and the performance stated in the environmental report as a proxy for the environmental disclosure quality, whereas other studies only focus on the environmental score (Eng and Mak, 2003; Akbas, 2016). This measure provides a more realistic view of the true

8 environmental disclosure quality. Fourth, as revealed in the study of Cohen and Simett (2015), the environmental external assurance remains an underdeveloped research area where potentially valuable results are expected. Prior research on environmental assurance is mainly focused on the effect and value of assurance (O’Dwyer, Owen and Unerman, 2011), but there is limited research found combining external assurance with board characteristics. Finally, research conducted on the proportion of women on the board, environmental disclosure quality and external assurance on environmental reports is mainly focused on one or a small number of countries (Nekhili et al., 2017; Barako et al., 2008; Liao, Luo and Tang, 2015). The consequence of this is that it is complicated to generalize these results. This study, therefore, extends the literature by using data from 54 countries to ensure that the results can be generalized.

The remainder of this paper is structured as follows. The following section describes the relevant prior literature and hypothesis development. Section 3 introduces the research methodology, and the results are presented in section 4. Section 5 contains the conclusions, discussion and gives some suggestions for future research.

9

2 THEORETICAL BACKGROUND

This section presents an overview of the existing literature on the proportion of women on the board and CEO duality in connection to the quality of environmental disclosures and the likelihood of buying external assurance on environmental disclosures. This study builds on two well-established theories – the stakeholder-agency framework and the legitimacy theory. Next, these theories will be linked to the objects of research – the proportion of women on the board, CEO duality, the quality of environmental reporting and the likelihood of buying external assurance. Finally, this section presents hypothesizes based on these theories.

2.1 STAKEHOLDER-AGENCY FRAMEWORK

The stakeholder-agency framework combines the features of the stakeholder theory and agency theory. The stakeholder-agency framework links management to its stakeholders (Hill and Jones, 1992). A stakeholder can be defined as ‘any group or individual who can affect or is affected by the achievement of the organization’s objectives’ (Freeman, 1984, p. 46). In this view, the firm has a responsibility that reaches beyond the shareholders’ interest and takes the interest of all stakeholders into account (Bowen, 1953). There are two groups of stakeholders; primary stakeholders, who are crucial to the firm’s survival such as investors, suppliers, employees, customers and government, and secondary stakeholders, who are not directly involved in the firm’s transactions. However, certain secondary stakeholders, such as media, activist groups, and NGOs, can have a substantial impact on the firm's reputation (Clarkson, 1995). Hence, a firm can be seen as ‘a nexus of contracts’ with implicit and explicit contractual relationships between all the stakeholders (Hill et al., 1992). The managers are unique in their position at the center of the nexus of contracts, and they can be seen as the agents of other stakeholders (Hill et al., 1992). However, the stakeholder-agency framework stated that ‘all individual action is driven by self-interest and that individuals will act opportunistically to increase their wealth' (Deegan and Samkin, 2009, p.71). A central assumption in this framework is that the interests between the managers and stakeholders are not aligned (Jensen and Meckling, 1979). Moreover, due to the separation of ownership and management, there is information asymmetry between management and all other stakeholders (An, Davey and Eggleton, 2011). Information asymmetry makes it difficult for stakeholders to monitor the actions executed by management (Hooghiemstra, Hermes and

10 Emanuels, 2015). The role of the board of directors is to monitor and supervise the management and make sure that the management act on behalf of the stakeholders (Jensen and Meckling, 1979). Further, the board of directors plays an essential role in the communication towards the stakeholders. For example, they make the decisions of whether, what and how much to disclose (Millstein, 1991). The board of directors is perceived as effective if it fulfills its monitoring, supervisory and communication role well (Liao et al., 2018; An, et al., 2011; Brammer et al., 2008). Nowadays, the monitoring and the application of stakeholder-management contracts adapt to new situations, such as the demand for a higher quality of environmental disclosures (Moroney et al., 2012). Therefore, an effective and well-governed board should be in place to ensure high-quality monitoring, and voluntary communication of environmental information (Liao et al., 2018; An, et al., 2011; Brammer et al., 2008).

2.2 LEGITIMACY THEORY

The legitimacy theory can help us to explain the rational of firms to undertake voluntary actions relating to environmental disclosures and buying external assurance (Campbell, Craven and Shrives, 2003; O’Dwyer, 2002; Peters and Romi, 2015). The legitimacy theory suggests that there is a ‘social contract', which explains the ongoing relationship between the society and the organization, concerning the state of organizational legitimacy (Deegan, 2006; Deegan, Ranking and Tobin, 2002). Legitimacy can be defined as ‘a generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions’ (Suchman, 1995, p. 574). By fulfilling society and stakeholders' expectations, which are in the ‘social contract', a firm can enhance the legitimacy of its operations by various actions including voluntary disclosure of environmental information and buying external assurance on environmental reports (O’Dwyer, 2002; Suchman 1995). The firm may benefit from enhanced credibility, reputation and corporate branding, which can lead to increased sales, revenues and attracting new investors (Pfau et al., 2008; Weber, 2008; Bear et al., 2010; Gray, 2006; Ashforth and Gibbs, 1990). Therefore, the status of legitimacy is considered to be crucial for the survival of the organization (An et al., 2011). According to the stakeholder-agency framework, the board of directors is responsible to make sure that management actions are aligned with the expectations of society/stakeholders (Jensen, 2002). Nowadays, stakeholders are asking for more credible and transparent environmental information (Gray et al., 2001; Dando et

11 al., 2003; Moroney et al., 2012). Moreover, some stakeholders specifically ask for external assurance on environmental disclosure (Moroney et al., 2012). In order to comply with these expectations, the board has to act. Following the legitimacy theory, providing high-quality environmental disclosure and buying external assurance on environmental disclosure can be seen as a way to legitimate themselves by society/stakeholders.

Study name Results/contribution Dependent variable Data Theory applied Country

Rodriguez-Dominguez et al. (2009)

Women on board (0) Ethical issues DEUS Database and corporate websites for European companies

Legitimacy theory/Stakeholder theory

Spain, Italy and the UK Liao et al. (2015) Women on board (+) GHG disclosure CDP FTSE350 reports on

CDP website

Legitimacy theory/ Stakeholder theory

UK Lone et al. (2016) Women on board (+) CSR disclosure Annual and sustainability

reports of companies on KSE

No specific theory Pakistan Ben-Amar, Chang

and Mclkenny (2017)

Women on board (+) Voluntary climate change disclosure

CDP-Canada annual survey and Stock-Guide database

Socialization theory Canada Nalikka et al. (2009) Women on board (+) Voluntary disclosure in

annual reports

Annual reports and Thomson Financial World scope database

No specific Theory Finland

Lakhal (2015) Presence at least three women on board (-)

Earnings management Annual reports and Thomson One Banker database

Agency

Theory/Critical mass theory

French

Post et al. (2011) Women on board (0) Environmental corporate responsibility

KLD Agency theory US

Giannarakis (2014) Women on board (0), CEO duality (0)

CSR disclosure Bloomberg data source Stakeholder theory US Akbas et al. (2016) Women on board (0) Environmental disclosure Annual reports Agency theory Turkish

Bear et al. (2010) Women on board (+) CSR Fortune Signaling theory World’s Most

Admired companies Gul et al. (2011) Women on board (+) Stock price

informativeness

Thomas Reuters Database No specific theory US Cheng et al. (2006) CEO-duality (0) Voluntary disclosure Annual reports and Corporate

Governance and Intellectual Capital database

Agency theory Singapore

Jizi (2017) Women on board (+), CEO duality (0)

CSR disclosure Bloomberg database Legitimacy theory UK

Jizi et al. (2014) CEO duality (+) CSR disclosure Thomson One banker database

Agency theory US Liao et al. (2018) Women on board (+),

CEO duality (+)

CSR assurance Chinese Stock Market Trading Database

Stakeholder-agency framework

China Mallin et al. (2013) CEO duality (+) Social performance Annual reports and KLD’s

SOCRATES database

Agency theory US Muttakin et al. (2018) CEO duality (-) CSR disclosure Annual reports Resource dependency

theory/agency theory

Bangladesh Chau and Gray

(2010)

CEO duality (-) Voluntary disclosure Annual reports and Perfect Analysis Database

No specific theory Hongkong Gul and Leung

(2004)

CEO duality (-) Voluntary disclosure Annual reports, Company Research Database

Agency theory Hongkong Donnely and

Mulcahy (2008)

CEO duality (-) Voluntary disclosure Annual reports Stakeholder-agency framework

Ireland Table 1: Theoretical background and review of prior research, whereas: +, positive; 0, insignificant; -, negative.

12

2.3 HYPOTHESIS DEVELOPMENT

2.3.1 Proportion of women on the board & Environmental disclosure quality

There is a significant amount of literature emphasizing the importance of women on the board. Previous research has shown that a board functions differently when the composition of the board, the ratio of women to men, changes (Adams and Ferreira, 2009). Furthermore, Adams et al. (2009) argue that women are more independent and objective in their decision-making compared to men because women are not part of the ‘old boys' network and they are, by nature, less self-interest orientated. Moreover, women have a higher board meeting attendance, which suggests that they invest more time and energy in executing their board member function (Adams et al., 2009). Additionally, women directors are more focused on the long-term, and the feminine leadership style is characterized by trust and cooperation (Adams and Funk, 2012; Klenke, 2003). Developing this trust and cooperation requires greater information exchange (Niederle and Vesterlund, 2007; Luckerath-Rovers, 2010). This independence, higher board meeting attendance, and feminine leadership style will improve the supervisory and monitoring role of the board as well as the quality of the board discussions (Gul, Srinidhi and Ng, 2011). This is in line with the research conducted by Liao et al. (2018) who find that more gender-diverse boards are more effective in monitoring management. The improved monitoring will reduce the information asymmetry and helps aligning managers interest with that of the stakeholders (Gul et al., 2011; Lakhal et al., 2015). As discussed in the stakeholder-agency framework, the improved supervisory and monitoring role improves the effectiveness of the board. This is in line with the research of Adam et al. (2009) who noticed that gender diversity is an indicator of board effectiveness. Furthermore, an increase in the number of women on the board may further increase the effectiveness of the board (Adams et al., 2009; Bear et al., 2010). A more effective board results in a more transparent organization which will lead to more and better information exchange. In consequence, it is expected that this more and better information leads to more and higher quality environmental disclosures. However, a few studies do not find the aforementioned differences in the board when there is more gender diversity on the board (Rodrigues-Dominguez et al., 2009; Post, Rahman and Rubow, 2011). They suggest that women may adopt conventional behavior and even adopt a male role in order to execute their board role. This will eliminate the positive effects, which are expected that women will bring on board. Despite some studies do not find the positive

13 effects of women on board, we claim that a higher proportion of women on the board has a positive effect on the quality of environmental disclosure.

H1: There is a positive association between the proportion of women on the board and the quality of the environmental disclosure

2.3.2 Proportion of women on the board & probability to buy external assurance

As mentioned in the stakeholder-agency framework and legitimacy theory, organizations interact with a broad set of stakeholders and to achieve legitimacy they need to fulfill stakeholders needs. Drivers for companies to buy voluntary external assurance could be reputation enhancement and meeting the demands of stakeholders (Moroney et al., 2012; Dando et al., 2003). Adopting the legitimacy perspective, buying external assurance can be seen as a way to legitimize the firm’s operations. Ciorcirlan and Petterson (2012) found that women improve the legitimacy of an organization. Correspondingly, women will pay more attention to all stakeholder’s needs and they are more interested in all economic, social and environmental aspects than men (Konrad and Kramer, 2006; Bear et al., 2010). In addition, Bear et al. (2010) argue that women are more ethical and dedicated to giving more attention to social and environmental responsibility and philanthropy. For example, Zhang, Zhu and Ding (2013) found that women directors would increase the number of charities, which is in contrast to men who are more interested in the shareholders instead of other stakeholders. This is in line with research conducted by Williams (2003), who found that companies with more women directors on board have a higher level of charitable giving. Furthermore, women are more likely to have an expert background outside the business, and they evaluate themselves more on non-financial performance indicators than men(Stephenson, 2004; Hillman et al., 2002). This broader view of women provides a more diverse perspective on the board and encourages conversations that may enable the board to more effectively address environmental issues and stakeholders’ needs (Bear et al., 2010; Liao et al., 2018). This is in line with the stakeholder-agency framework, which emphasizes the importance to consider a broad view on stakeholders. Further, Hillman, Cannella and Haris (2002) found that women on the board are more likely to support external specialists, such as auditors than men. More specifically, Liao et al. (2018) find that a higher proportion of women on board have a positive influence on the probability of buying external assurance on CSR reports. Given the aforementioned skills and knowledge that women bring to the board and the results of the research of Liao et al. (2018), it is

14 expected that female directors pay more attention to the need for external assurance on environmental disclosure than male directors. An increase in the number of female directors would bring more of these skills and knowledge to the board (Adams et al., 2009; Bear et al., 2010). Accordingly, we formulate the following hypothesis:

H2: There is a positive association between the proportion of women on the board and the probability to buy external assurance on environmental disclosures

2.3.3 External assurance & Environmental disclosure quality

Organizations buy external assurance because low-quality information is not useful, and it is perceived as non-credible (Deegan, 2006; Junior, Best and Cotter, 2014; Dando et al., 2003). In general, external assurance is used to enhance the credibility, reliability, transparency, and quality of the company's information by reducing information asymmetry between the managers and the stakeholders (Watts and Zimmerman, 1986; Francis, 2004). Studies on financial reports confirm the aforementioned positive effects of external assurance (Peck and Sinding, 2009; Karapetrovic and Willborn, 2000). The external assurance on environmental reports is based on a voluntary character and is not regulated in the majority of countries. This results in varying quality of assurance (Deegan, 2006). To boost the external assurance on environmental reports, the Sustainability Reporting Guidelines of the Global Reporting Initiative (GRI, 2002) provides some guidelines for assuring these disclosures. Assurance providers can use these guidelines to add more creditability to environmental reports (Moroney et al., 2002). Nowadays, these guidelines play a leading role in environmental disclosures and are used by many organizations (KPMG, 2013). Additionally, some studies already acknowledge the value of assurance on sustainability reports (i.e. environmental) (Martinez-Ferrero and Garcia-Sanchez, 2017; Coram et al., 2009). For example, Coram et al. (2009) find increased credibility on voluntary non-financial reports when they are assured. Given that the assurance on financial reports improves the quality, some studies acknowledge the value of assurance on sustainability reports and many auditors make use of the guidelines of GRI for the environmental assurance, we expect that external assurance also improves the quality of the environmental disclosure.

15 Hence, the following hypothesis is formulated:

H3: There is a positive association between the existence of external assurance and the quality of the environmental disclosure

2.3.4 CEO duality & Environmental disclosure quality

As mentioned earlier, the other corporate governance mechanism in our study is CEO duality. CEO duality could indicate managerial power since the appointment of the CEO as chairman of the board is mostly determined by their successful career records and/or controlling a significant proportion of shares (Hermalin and Weisbach, 1998; Jizi, 2017). Moreover, the CEO-chairman often has a substantial influence on the selection of the other board members (Muttakin et al., 2018; Guldiken et al., 2018). On top of that, the chairman of the board has the authority to set the agenda of the board, and he/she can decide which information is transferred to the other board members (Jizi, 2017; Liao et al., 2018). As aforementioned in the stakeholder-agency framework, the interests between the CEO and the board of directors may not be aligned; therefore, a powerful CEO may affect board decisions which may decline the effectiveness of the board (Millstein 1991; Jizi et al., 2014). Galbraith (2010) argues that companies where the CEO and the chairman are different persons obtain better governance and monitoring and are therefore more effective boards, which is in line with the previous example. A reduced monitoring role of the board leads to a less transparent organization, which will result in less and lower quality information exchange and hence results in lower quality of environmental disclosure. However, the literature reveals mixed results between the relationship CEO duality and reporting quality. Finkelstein et al. (1994) suggests that CEO duality is not always ineffective. Correspondingly, Jizi et al. (2014) and Mallin, Michelon and Raggi (2013) shows that CEO duality facilitates the voluntary reporting to legitimate themselves by society. Post et al. (2011), Giannarakis (2014) and Cheng et al. (2006) find no significant relation between CEO duality and CSR disclosure. The majority of studies finds a negative relationship between CEO duality and voluntary disclosure, which is in line with the theory (Muttakin et al., 2018; Chau et al., 2010; Gul et al., 2004; Donnely et al., 2008). Clearly, according to the theory, a negative relationship is expected, which suggest that separation of the CEO and chairman is desired.

16 Therefore, the following hypothesis is formulated:

H4: CEO duality has a negative association with the quality of environmental disclosure

2.3.5 Moderating role of CEO duality

As mentioned earlier, women on board bring positive skills and knowledge on the board, which may lead to higher quality environmental disclosure. However, there is a difference between the presence and the adoption of these skills and knowledge (Forbes and Miliken, 1999). To take advantage of these positive skills and knowledge on the board, the contribution of women in the board’s decision-making is important (Nielsen and Huse, 2010). The women’s contribution can be defined as their ‘influence or impact on decision-making’ (Westphall and Milton, 2000 p. 376). However, when there is CEO duality, the CEO-chairman has more power in the board, then the other board members (Jizi, 2017). Moreover, the CEO-chairman has the authority to exclude board members in the decision process (Jizi, 2017). This authority can reduce the contribution of the other board members in the decision-making process of the board, and therefore this may reduce the positive influence which women bring to the board. Therefore, it is expected that CEO duality negatively moderates the relationship of women on the board and the environmental disclosure quality. This leads to the following hypothesis:

H5: CEO duality weakens the relationship between the proportion of women on the board (%) and the quality of environmental disclosure

2.3.6 CEO duality & Probability to buy external assurance

According to the stakeholder-agency framework, the interests between the CEO and the board may not be aligned (Hill et al., 1992). The compensation of the CEO-chairman depends on several performance indicators, which are mostly financial in nature (Coombs and Gilley, 2005). Therefore, it is reasonable that the CEO-chairman will mainly focus on financial activities, such as profit-maximizing, compared to other non-financial activities (Hill et al., 1992). This is in line with previous research by Stanwick and Stanwick (2001), who show a negative relationship between executive compensation and social performance. This suggests that the performance indicators may not be in line with the goals of social/environmental performance/disclosure. Allocating resources to environmental purposes, such as buying external assurance on

17 environmental reports, can result in a decrease in the financial resources and therefore, may not in line with the aim of profit maximization (Friedman, 1970). Therefore, the CEO-chairman may be less likely to buy external assurance because they may view this as unfavorable to their interests (Liao et al., 2018; Peters et al., 2015). The other members of the board may accept the decisions of the CEO-chairman, which is not in favor of the stakeholders, to avoid confrontation with their CEO-chairman (Dey, 2008). This argument is in line with the findings by Shaukat, Qui and Trojanowski (2016). They found that when the CEO is also the chairman of the board, it is more likely that the board will act in line with the interest of the management/CEO. According to this, the board will converge more to the profit-maximizing interests of the CEO-chairman. However, some studies argued that environmental assurance can be seen as a legitimacy tool (Jizi, 2014; Haffina et al., 2005). Legitimacy can lead to better reputation, more customer satisfaction and a positive corporate image which, in turn, lead to an increase in sales and revenues (Pfau et al., 2008; Weber, 2008; Bear et al., 2010; Freeman, Wicks and Parmar, 2004). Hence, this can stimulate the CEO-chairman to buy external assurance on environmental reports. However, it is more likely that the CEO-chairman will focus on the short-term and therefore views environmental external assurance as costly instead of a value-creating item. Correspondingly, Abdullah et al. (2008) finds that CEO duality leads to a reduction in the effort put into buying a high-quality auditor for the financial statements. This suggests that the CEO-chairman does not view the external assurance as a value-creating tool. Therefore, we argue that CEO duality has a negative impact on the probability to buy external assurance.

H6: CEO duality has a negative association with the probability to buy environmental external assurance

18

2.4 CONCEPTUAL MODEL

According to the provided theory, the hypotheses can be summarized into one conceptual model (Figure 1). The conceptual model shows the positive relationships between the proportion of women on board on the quality of environmental disclosure and on the probability to buy external assurance (H1 & H2). Besides that, the model shows a positive relationship between external assurance and the quality of environmental reports (H3). Furthermore, the conceptual model shows the negative relation of CEO duality on the quality of environmental disclosure and the probability to buy environmental external assurance (H4 & H6). Lastly, the model shows the negative moderating role of CEO duality on the relationship between the proportion of women on board and the quality of environmental disclosure (H5).

19

3 RESEARCH METHOD

In this chapter, the research method is discussed. First, the sample design will be presented, after that the research design is discussed and lastly the analytical models will be presented, which will be used to test the hypotheses.

3.1 SAMPLE DESIGN

The information of this sample is retrieved from different databases. The data related to environmental disclosure score and external assurance of environmental reports are retrieved from the Bloomberg database, which provides a wide coverage of data. This database offers relevant and objective data and is already used in previous studies on CSR disclosure (Giannarakis, 2014; Jizi, 2017). The other variables are retrieved from the Thomas Reuters ASSET 4 ESG database. All the collected data from the different databases are matched into one file in excel.

The original sample included 64.863 firm-year observations with a period of 9 years, ranging from 2009 to 2017. The main reason for this period was to capture the recent developments of the environmental importance for society and the increasing demand for external assurance on environmental reports. The firm-year observations are starting from 2009 because of the potential issue that the financial crisis of 2007 and 2008 could bias these results. In the beginning, the results were controlled for duplicates and missing values. For calculating the environmental disclosure quality, the standardized environmental performance score and the standardized environmental disclosure score are needed. In case quality could not be calculated due to missing variables, items are deleted. This reduced the sample to 17.745 firm-year observations. Furthermore, 16 firm-years observations were removed because of the missing percentage of women on board. The data regarding CSR external assurance, which is used as a proxy for external assurance of environmental reports, consists of 6.840 firm-year observations. This lowers our firm-year observations substantially. Lastly, the control variable R&D intensity has too many missing items to delete them, and therefore the missing items are replaced by the mean. So, the final sample consists of 6.840 firm-year observations and is distributed over 54 countries, which is shown in table 2 in the Appendix. The reason for focusing on a global set of organizations is because previous research is mostly focused on one or a small number of countries (Nekhili et al., 2017;

20 Barako et al., 2008; Liao et al., 2015). The distribution over the countries in our sample is not equal. The Appendix shows that the US is overly represented with 1.238 firm-year observations, while some other large countries have relatively low observations. For example, Pakistan has only one firm-year observation.

3.2 RESEARCH DESIGN 3.2.1 Dependent variables

This research consists of two dependent variables, namely the quality of environmental disclosure and the probability to buy environmental external assurance.

Environmental disclosure quality (EQ): The quality of the environmental disclosure can be seen as the decoupling of the firm’s environmental performance and the environmental performance disclosed. Therefore, the quality of the environmental disclosure can be calculated by subtracting the standardized environmental disclosure score from the standardized environmental performance score. The value of the performance score, as well as the disclosure score, lies between 0 and 100. The variables are standardized because they are retrieved from different databases. When the calculated quality score has a value near to zero, it means that the environmental performance is similar to the disclosed environmental performance and therefore, the quality of the disclosure is perceived as high. The value of the variable EQ lies between -100 and 100. The environmental performance score and the environmental disclosure score are collected from the Bloomberg database and Thomas Reuters ASSET4 database.

Probability of buying external assurance on environmental disclosure (EXAS): External assurance on CSR reports is used as a proxy for external assurance on environmental reports since environmental reports are mostly part of the CSR reports. The (in)dependent variable is measured as a dummy variable which equals 1 when there is external assurance on the environmental disclosure and equals 0 when there is no external assurance. The data used for this variable is obtained from the Bloomberg database.

21

3.2.2 Independent variables

Both models have the same independent variables, namely the proportion of women on board and the existence of CEO duality.

The proportion of women on board (% WOB): The percentage of women on board is calculated through dividing the number of women directors by the total number of board members. Previous research shows that it is better to use a percentage instead of an absolute number of women on board because a percentage of women on board provides more information than an absolute number (Konrad, Kramer and Erkut, 2008; Liao et al., 2015). Therefore, this study makes use of the percentage of women on board. The data for this variable is collected from the Thomas Reuters ASSET4 database.

CEO Duality (CEODU): This variable is a dummy variable. Whether the CEO and chairman are the same person the variable equals 1, and whether the CEO and Chairman are different persons this variable equal 0. The data for this variable is collected from the Thomas Reuters ASSET4 database.

3.2.3 Control Variable

We added control variables to ensure that we only measure the tested effect. First, capture the impact of the profitability of the firm. Previous research shows that the disclosure of environmental information entails costs and firms with high profitability provide managers funding to support environmental disclosures (Brammer et al., 2006). In line with previous research, the profitability of the firm is equal to the Return on Total Assets (ROTAw), which is calculated by dividing the net profit by the total assets (Brammer et al., 2006). Second, the Capital intensity is included (CAPw), Cohen (2008) found that capital intensity is positively associated with the quality of information. Third, R&D intensity is included (RDw). Previous research found that the higher the R&D investments, the higher the agency costs are. Therefore, firms with high R&D investments are more likely to demand external assurance to reduce these agency costs (Godfrey and Hamilton, 2005). Moreover, R&D intensity is significantly associated with the market value of the firm (Wang, Huang and Lee 2013). Fourth, the presence of a CSR committee (CSRC) is included as a control variable. A CSR committee should improve the CSR and guarantees the proper functioning of CSR and environmental information (Fuente, Garchia-Sanchez and Lozano, 2017). This is a dummy variable, which equals 1 when a CSR committee is present and equals 0 when there is no

22 CSR committee present. Fifth, the board size (BSIZEw) is included as a control variable because prior research shows that an increase in the size of boards makes it easier to become involved with environmental issues and reporting (Fuente et al., 2017; Goodstein, Gautam and Boeker, 1994; Akbas, 2016). Sixth, to capture the effect of firm size, the natural logarithm of the total asset is included (FSIZEw), which is in line with the research of Brammer et al. (2006). Lastly, the county, the year, and the industry control variables are used to consider the country, the year and the industry differences about environmental disclosure and external assurance on these disclosures. The country dummy’s in our sample represent the three countries with the most firm-year observations. Further, the industries are divided into ten groups following the Industry Classification Benchmark. Our sample has no firm-year observations in the Wholesale trade industry, and therefore our sample consist of nine different industries. Mock, Strohm and Swartz (2007) find that the probability of buying assurance differs per industry. For example, buying assurance is very common in the electricity, mining and oil and in the manufacturing sectors. Year-effects are used to control the increasing attention of environmental disclosure quality and the external assurance on environmental reports over the years. In order to eliminate the effect of outliers the control variables, ROTAw, CAPw, RDw, BSIZEw, and FSIZEw are winsorized. These variables are winsorized one percent at the top and bottom. The other variables are not winsorized because these variables are dummies and ratio variables. Table 3 contains an overview of all the variables which are used in this study.

23

Name Role Measurement

EQ Dependent variable The difference between the standardized environmental performance and the standardized environmental disclosure

EXAS Dependent variable Dummy variable equals 1 when external assurance is exercised and equal 0 when there is no external assurance exercised

%WOB Independent variable

Women on board as a percentage of the total board WOB3 Independent

variable

Dummy variable equals 1 whether there are three or more women on board and equals 0 whether there are less than three women on board CEODU Independent

variable/ Moderator

Dummy variable equals 1 whether the CEO and Chairman are the same person and equals 0 when they are different persons

ROTAw Control variable The winsorized ratio of the firm’s net profit to total assets

CAPw Control variable The winsorized ratio of the firm’s total assets to net sales

RDw Control variable The winsorized ratio of the firm’s total R&D costs to firms’ total sales

CSRC Control variable Dummy variable equals 1 when a CSR committee is present and equals 0 when there is no CSR committee

BSIZEw Control variable The winsorized number of seats in the board FSIZEw Control variable The Natural Log of Total Assets

COUN Control variable Dummy variable: GB, JPN, US and other. INDU Control variable Eight dummy variables based on the Industry

Classification Codes

YEAR Control variable Eight dummy variables based for the nigh years in our sample

24

3.3 MODEL DEVELOPMENT

This study contains panel data, and therefore the panel data methods should be employed. For the continuous dependent variable (environmental disclosure quality), there are two possible techniques to analyze the panel data, namely Fixed-effects or Random-effects regression. A Durbin-Wu-Hausman test is performed to consider if the fixed-effects model or the random-effects model should be employed. After that, the Breusch pagan test is performed. Both tests conclude that the Random-effects model should be employed. To test the hypotheses presented with environmental disclosure quality as a dependent variable, I run panel regressions with Random- effects to reduce endogeneity issues. Further, a probit analysis is performed for the hypotheses with the dichotomous dependent variable (environmental external assurance). The following regressions models are used:

EQ = β0 + β1*%WOB + β2*EXAS + β3*CEODU + β4* CEODU * %WOB + β5*ROTAw + β6 * CAPw+ β7 * RDw + β8 * CSRC + β9 * BSIZEw + β10 * FSIZEw + β11 * COUN + β12 * INDU + β13 * YEAR + ε

EXAS= β0 + β1*%WOB + β2 * CEODU+ β3*ROTAw + β4 * CAPw+ β5 * RDw + β6 * CSRC + β7 * BSIZEw + β8 * FSIZEw + β9 * COUN + β11 * INDU + β12 * YEAR + ε

25

4 RESULTS

In this chapter, descriptive statistics are presented. After that, Pearson’s correlation is performed and discussed. Third, the Random-panel regressions is presented and discussed for the first model, and after the probit analysis is presented and discussed for the second model. Lastly, an additional test is conducted to determine whether the results are robust and consistent.

4.1 DESCRIPTIVE STATISTICS

Table 4 provides the descriptive analysis of variables that are used in this study. This study has 6.840 firm-year observations. The average score of environmental quality is 48.11. Which is a relatively low score, because the quality is perceived high when the score is close to zero. Further, 73% of our firm-year observations buy external assurance on their environmental reports. The average percentage of women on board is 14.48%, which is relatively low, and this is in line with the fact that women are still underrepresented in boards (Morono-Gomez et al., 2018). In our sample, the distribution of the variable percentage women on board is converged. Some boards have no female directors, while others have 85.71% female directors on board. Further, only 17% of the boards have three or more women, and CEO duality exists in 64% of the firm-year observations.

N Mean Min Max Std. Dev.

EQ 6840 48.11 -57.58 93.07 20.45 EXAS 6840 0.73 0 1 0.44 %WOB 6840 14.48 0 85.71 12.74 WOB3 6840 0.17 0 1 0.37 CEODU 6840 0.64 0 1 0.48 ROTAw 6840 0.74 0.02 2.94 0.58 CAPw 6840 4.06 0.34 37.47 6.73 RDw 6840 3.45 0 31.26 2.12 CSRC 6840 0.87 0 1 0.34 BSIZEw 6840 11.42 5 22 3.67 FSIZEw 6840 17.79 12.60 25.21 2.83

26

4.2 CORRELATIONS

The correlations between the variables must be inspected to control the effect of multicollinearity. Therefore, table 5 presents the Pearson correlation matrix. There is collinearity whenever the correlation between the variables is very high or very low. In this study, variables are collinear if the correlation is higher than 0.8 or lower than – 0.8, which is in line with prior research by Greene (1999). All the correlations in the table are within 0.8 and -0.8, which indicates that there are no collinearity issues. In addition, the variation inflation factor (vif) test is performed. In line with the correlation matrix, the ratio’s (highest vif: 1.43 and mean vif: 1.15) do not give any sign of multicollinearity. This matrix suggests that the proportion of women on board and the probability of buying external assurance have a significant negative correlation with environmental disclosure quality (r= 0.0546; p <.01 and r= 0.1301; p<.01). This is contradictory to our expectation in the first and third hypotheses. Further, there is a significant positive correlation found between CEO duality and environmental disclosure quality (r= -0.0526; p<.01). This is contradicting with our fourth hypothesis. The proportion of women on the board on the probability to buy external assurance is significant positive correlated (r=0.0410; p<.01), which is in line with our expectation. Lastly, CEO duality is significant positive correlated with external assurance on environmental reports (r= 0.0470; p<.01), which is contradicting to our expectation.

EQ EXAS %WOB CEOD ROTAw CAPw RDw CSRC BSIZEw FSIZEw EQ 1.0000 EXAS 0.1301*** 1.0000 %WOB 0.0546*** 0.0410*** 1.0000 CEODU -0.0526*** 0.0470*** 0.0567*** 1.0000 ROTAw 0.0526*** -0.0661*** -0.0021 -0.0216* 1.0000 CAPw 0.0143 0.0344*** 0.0569*** 0.0698*** -0.5155*** 1.0000 RDw 0.0031 0.0076 -0.0088* -0.0117*** -0.0330*** 0.0024 1.0000 CSRC 0.1982*** 0.2092*** 0.0991*** -0.0242*** 0.0021 -0.0009 -0.0275 1.0000 BSIZEw 0.0493*** 0.1282*** 0.0993** -0.0950*** -0.0884*** 0.1002*** -0.0273** 0.0706*** 1.0000 FSIZEw -0.0979*** 0.1447*** -0.2515*** -0.0207* -0.2238*** 0.2570*** -0.0292** -0.0289** 0.0752*** 1.0000

28

4.3 RANDOM-EFFECT PANEL REGRESSION

In this section, we present the results of the random-effect panel regression of the relation between the independent variables and the quality of environmental disclosure for the hypotheses 1, 3, 4 and 5, as shown in table 6.

Model 1 contains a Random-effect panel regression with only the control variables. We made the assumptions for the control variables based on prior literature. Most of our control variables are significant in our model, however the direction is different as which is expected from prior research. Further, R&D intensity is not significant in this model. Model 2 presents the relation between the proportion of women on board and the quality of the environmental disclosure. The coefficient is positive and highly significant (β = 0.107; p<.01). Therefore, we cannot accept hypothesis 1. This means that the proportion of women on board is not positively associated with the quality of the environmental disclosure.

Model 3 shows the relation between external assurance on environmental disclosures and the quality of the environmental disclosures. This model shows a significant negative relation between external assurance and the quality of the environmental reports (β= 8.391; p<.01). Based on the results, hypothesis 3 is not accepted. Model 4 presents the results for hypothesis 4, which estimates a negative relation between CEO duality and the quality of the environmental reports. The variable CEO duality is positive significant on the quality of the environmental reports with a confidence interval of 1% (β= -2.229; p<.01). This is contradicting with the expectation of hypothesis 4. Therefore, we again cannot accept hypothesis 4.

Model 5 measure the moderation effect of CEO duality on the relation between the proportion of women on board and the quality of the environmental disclosure. Model 5 tests the hypothesis 5 of our study. We predicted that CEO duality negatively moderates the relationship between the proportion of women on board and the quality of environmental reports. The interaction variable is positive significant, with a confidence interval of 1% (β= -0.0866; p<.01). Therefore hypothesis 5 cannot be accepted. Model 6 presents the results of all determinants, which are separately presented in model 1 to 5. The results are similar and the coefficient have the same direction. The adjusted R2 _{is increased from 0.088 (model 1) to 0.108 (model 6) after adding our determinants,}

29

4.4 PROBIT REGRESSION

This section presents the results of the probit regressions of the relation between the independent variables and the probability to buy external assurance on environmental disclosures for hypotheses 2 and 6, as shown in table 7.

Model 7 contains a probit analysis with only the control variables. This model shows that most of the control variables have a significant impact on environment external assurance, which is in line with prior research. However, the Return on Total Assets and the R&D intensity do not have significant effects on the probability to buy environmental external assurance. Model 8 is used to address the impact of the proportion of women on board and the probability to buy external assurance on environmental disclosure. According to the second hypothesis, the probability of buying environmental external assurance is higher (lower) whether the proportion of women on board is higher (lower), so a positive relationship is expected between these two variables.

(1) (2) (3) (4) (5) (6)

Hypothesis H1 H3 H4 H5

VARIABLES Expected relation

%WOB + 0.107*** 0.113*** 0.0541** EXAS + 8.391*** 9.116*** CEODU - -2.229*** -2.381*** CEODU * %WOB - -0.0866*** -0.0116* ROTAw 2.713*** 2.660*** 2.578*** 2.794*** 2.665*** 2.679*** CAPw 0.199*** 0.187** 0.212*** 0.203*** 0.188** 0.223*** RDw 0.108 0.118 0.122 0.108 0.120 0.118 CSRC 16.07*** 15.83*** 14.48*** 16.01*** 15.79*** 14.40*** BSIZEw 0.660*** 0.621*** 0.548*** 0.639*** 0.605*** 0.537*** FSIZEw -1.420*** -1.367*** -1.574*** -1.434*** -1.371*** -1.628***

Country dummy Yes Yes Yes Yes Yes Yes

Industry dummy Yes Yes Yes Yes Yes Yes

Year dummy Yes Yes Yes Yes Yes Yes

Constant 50.07*** 48.05*** 48.32*** 52.23*** 48.37*** 51.44***

Observations 6,840 6,840 6,840 6,840 6,840 6,840

Adjusted R2 _{0.088 0.089. 0.105 0.090. 0.090 0.108}

Table 6: Results of the model are related to environmental disclosure quality, whereas ***, **, and * are significant at 1, 5 and 10 percent, respectively.

30 However, this model shows a significant negative relation (β = -0.00312; p<.05). Therefore, hypothesis 2 cannot be accepted.

Model 9 presents the results for hypothesis 6, which estimates the relation between CEO duality and the probability to buy environmental external assurance. CEO duality has a significant relation with the probability to buy external assurance on environmental reports (β= 0.0468; p<.1). Contrary to our expectations, CEO duality shows a significant positive relationship with the probability to buy environmental external assurance. Since CEO duality shows contradictory results, we reject hypothesis 6. Model 10 presents the probit regression results of all determinants, which are separately presented in the models 7 to 9. This model shows similar significant results, and the coefficients have the same direction as in the individual models. The adjusted R2 _{in this}

model increases from 0.164 (model 7) to 0.165 (model 10) after adding our determinants. This increase is relatively low.

(7) (8) (9) (10) VARIABLES Hypothesis H2 H6 Expected relation %WOB + -0.00312** -0.00316** CEODU - 0.0468* 0.0488* ROTAw -0.0589 -0.0563 -0.0606 -0.0581 CAPw -0.0125*** -0.0120*** -0.0126*** -0.0121*** RDw 0.0105 0.0103 0.0107 0.0104 CSRC 0.601*** 0.610*** 0.603*** 0.613*** BSIZEw 0.0543*** 0.0555*** 0.0549*** 0.0561*** FSIZEw 0.0640*** 0.0616*** 0.0642*** 0.0618***

Country dummy Yes Yes Yes Yes

Industry dummy Yes Yes Yes Yes

Year dummy Yes Yes Yes Yes

Constant -0.952*** -0.864*** -1.000*** -0.913***

Observations 6,840 6,840 6,840 6,840

Adjusted R2 _{0.164 0.164 0.164 0.165}

Table 7: Results of the model are related to environmental external assurance, whereas ***, **, and * are significant at 1, 5 and 10 percent, respectively.

31

4.5 ADDITIONAL ANALYSES

In order to test the validity of the results mentioned above, we performed an additional test. This test determines whether a minimum of three women directors is needed to affect the corporate board’s outcomes.

The results of the proportion of women on board on the quality of environmental disclosure and probability to buy external assurance on environmental reports are negatively significant, which is contradicting to our expectations. According to the token status of women directors, a lonely woman director in the board is considered as a minor ‘token' by various stakeholders, and therefore her impact on board decisions is limited (Liu, Wei and Xie, 2014). When a group is faced with a consistent opinion from multiple minorities, it is more likely to acknowledge the minority voice (Asch, 1955). Consequently, the critical mass theory on board gender diversity suggests that if there are three or more women on board, women have a voice (Liu et al., 2014). Therefore, this test determines whether there are three or more women on board needed to influence the environmental disclosure quality and the probability to buy external assurance on environmental reports. We replace the proportion of women on board for a dummy variable with at least three female directors on board (WOB3). The WOB3 variable equals 1 when there are three or more women on board and equals 0 when there are less than three women on board. The coefficient of WOB3 on the quality of environmental disclosure in the Random-effect panel regression is 0.832 instead of 0.107 and is significant, which is shown in table 8. So, even when there are three or more women on the board, there is a negative significant effect on the quality of environmental disclosure. Moreover, the negative relation becomes stronger when there are 3 or more women on board. The coefficient in the probit analysis on the probability to buy external assurance is -0.0130 instead of -0.00312 and is significant, which is shown in table 9. This suggests that whether there are three or more women on board, the board is still less likely to buy environmental external assurance. Therefore, this additional test confirms the previous results of our analyses.

32 (1) (2) (3) (4) VARIABLES Hypothesis H2 H6 Expected relation WOB3 + -0.0130* -0.0129* CEODU - 0.0468 0.0468 ROTAw -0.0589 -0.0589 -0.0606 -0.0606 CAPw -0.0125*** -0.0124*** -0.0126*** -0.0125*** RDiw 0.0105 0.0105 0.0107 0.0107 CSRC 0.601*** 0.601*** 0.603*** 0.604*** BSIZEw 0.0543*** 0.0548*** 0.0549*** 0.0554*** FSIZEw 0.0640*** 0.0639*** 0.0642*** 0.0641***

Country dummy Yes Yes Yes Yes

Industry dummy Yes Yes Yes Yes

Year dummy Yes Yes Yes Yes

Constant -0.952*** -0.952*** -1.000*** -1.000***

Observations 6,840 6,840 6,840 6,840

Adjusted R2 _{0.164 0.164 0.164 0.164}

Table 9: Probit regression on environmental external assurance in combination with WOB3, whereas, ***, **, and * are significant at 1, 5 and 10 percent, respectively.

(1) (2) (3) (4) (5) (6) VARIABLES Hypothesis H1 H3 H4 H5 Expected relation WOB3 + 0.832* 0.894* 1.025* EXAS + 8.391*** 8.412*** CEODU - -2.229*** -2.258*** CEODU * WOB3 -0.0754*** -0.00032 ROTAw 2.713*** 2.702*** 2.578*** 2.794*** 2.707*** 2.644*** CAPw 0.199*** 0.197*** 0.212*** 0.203*** 0.198*** 0.213*** RDw 0.108 0.108 0.122 0.108 0.109 0.121 CSRC 16.07*** 16.06*** 14.48*** 16.01*** 16.04*** 14.40*** BSIZEw 0.660*** 0.633*** 0.548*** 0.639*** 0.619*** 0.492*** FSIZEw -1.420*** -1.415*** -1.574*** -1.434*** -1.420*** -1.581***

Country dummies Yes Yes Yes Yes Yes Yes

Industry dummies Yes Yes Yes Yes Yes Yes

Year dummies Yes Yes Yes Yes Yes Yes

Constant 50.07*** 50.09*** 48.32*** 52.23*** 50.45*** 50.51***

Observations 6,840 6,840 6,840 6,840 6,840 6,840

Adjusted R2 _{0.088 0.089 0.105 0.090 0.089 0.107}

Table 8: Random-effect panel regression on environmental disclosure quality in combination with WOB3, whereas, ***, **, and * is significant at 1, 5 and 10 percent, respectively.

33

5 DISCUSSION AND CONCLUSIONS

The last section of this research provides a summary of this study, a discussion of the results, and provides the conclusions of this study. After that, the limitations of the study and the opportunities for future research will be presented.

5.1 SUMMARY, DISCUSSION AND CONCLUSIONS

The main objective of this study is to analyze the level of the firm’s environmental disclosure quality under two most research internal corporate governance mechanisms i.e. gender diversity and CEO duality. These two mechanisms are selected because of their prominence in the governance-CSR literature and their expected opposite influence on corporate board outcomes. The major determinants of environmental disclosure quality are the level of decoupling of environmental disclosure from environmental performance and the availability of external assurance on environmental reports. Therefore, this study examines the relationship between the proportion of women on board and CEO duality on the quality of environmental disclosure and the probability of buying external assurance on environmental reports. Furthermore, it studies the moderating effect of CEO duality on the relationship of women on board and environmental reporting quality. The international data of our sample is collected on the Bloomberg database and Thomas Reuters ASSET4 database. A Random-effect panel regression is performed for the determinant quality of environmental disclosure, and a probit regression is performed for the determinant environmental external assurance.

Despite the application of the state-of-the-art methodological rigor, I am unable to support the hypothesized relationships. I stated that the proportion of female directors on board have a positive effect on environmental disclosure quality. However, in my model I find a significant negative relationship between the proportion of women on board and the quality of environmental disclosure. This finding is in contrast with the research of Fernandez-Feijoo, Romero and Ruiz (2014), Lone et al. (2016), and Jizi (2017) which shows that a higher proportion of women on board leads to higher quality CSR disclosure. Therefore, our finding could not confirm the argumentation that women improves the effectiveness of the board, which will result in a more transparent organization with better environmental disclosure quality. Some studies find no association between women on board and CSR disclosure quality. They provided the argument that women may adopt a male role to execute their board role (Rodrigues-Dominguez et al., 2009;