Corporate Governance, Ownership Structure and Voluntary
Carbon Information Disclosure: Evidence from China
Name: Xiaolu Xiong Student number: 11577991
Thesis supervisor: dr. Alexandros Sikalidis Date: June 25, 2018
Word count: 12,210
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
Statement of Originality
This document is written by studentXiaolu Xiong who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
To investigate what factors could influence the voluntary carbon information disclosure behaviour of companies and the extent to which the carbon information is disclosed, thus to help the general development of carbon trading and information disclosure transparency, I conducted a study to look at environmental information disclosure situations in developing countries versus developed countries. China has been on the road to the improvement of environmental protection actions. Following the global trend, a few enterprises in China started to disclose carbon information. Using a sample of Chinese listed firms in Shanghai Stock Exchange during the period 2014-2016, I examined both corporate governance specifically regarding board characteristic and ownership structure, and I find that fewer female directors would increase the likelihood for firms to disclose carbon information. My study also shows potential that a younger board would be more likely to make carbon information disclosure. However, the results show that a less age diverse board would increase the propensity to make carbon information disclosure, while non government-owned firms intend to disclose more greenhouse gas emissions information. I also found no evidence that suggests a less age diverse board would have higher disclosure quality, nor that a government-owned enterprise makes clear and complete carbon information disclosure. In sum, firms which are owned by the government are less willing to make carbon information disclosure, but the content of their disclosure does not significantly differ from those of private business. Finally, significant coefficients show that a board with fewer female members and of higher average age could provide a higher level of voluntary carbon information disclosure. Although, interestingly, the findings are mostly inconsistent with my expectations. This inconsistency either reflects an inappropriate theory foundation or is possibly due to a poor sample choice and an inexperienced act while assigning value for a dependent variable that required personal subjective judgement. Even so, it remains a brave attempt at contributing to the academic research fields of environmental information disclosure.
Keywords: corporate governance; board diversity; gender diversity; age diversity; ownership structure; carbon information disclosure
Contents
1 Introduction...5
2 Literature review...11
2.1 Research about CID in diverse setting...11
2.2 Research about determinants of CID...11
2.3 Theory regarding voluntary environmental disclosure...13
3 Hypothesis development...15
3.1 Corporate governance...15
3.1.1 board gender diversity...15
3.1.2 board age diversity... 16
3.2 Ownership structure...17
4 Data and method... 18
4.1 Sample selection... 18 4.2 Research design... 19 4.2.1 Dependent variables... 20 4.2.2 Independent variables...23 4.2.3 Control variables... 24 4.3 Regression model...25 5 Empirical results... 26 5.1 Descriptive analysis...26 5.2 Corporate governance...27 5.3 Ownership structure...29
6 Conclusion and limitation... 32
References...35
1 Introduction
In the past decades, the emission of the greenhouse gases has created a severe global climate problem. For many years, climate change and green-house effect have appeared in many aspects of the international society, especially in political and business fields. Under the global background, most nations have taken global warming (which is mainly caused by the emission of carbon dioxide and other greenhouse gases (GHG)) as an increasing issue to have waited for mitigation in governmental and corporate establishments. To effectively and efficiently decrease carbon emission and to take the steps towards a low-carbon economy has lately been the consensus of various countries in the world. In order to reach the goal of carbon emission mitigation, the first step is to control the emission of carbon dioxide. Among all the aspect of the control management, a significant part is a need for corporations to measure and disclose data to eventually reduce their level of carbon gas emission. Since the emergence of the carbon trading market, the related accounting issues have also received the attention of the academic community while the concepts of “carbon accounting” and “carbon information disclosure” rising as the times require. The importance of carbon disclosure is that the corporations have to be aware of the level of the emission initially, then make environmentally friendly policies and exercise activities that match the policies. (Bae Choi et al.,2013). Information disclosure about strategics, activities and firms’ impact on carbon emissions is vital because this information affects stakeholders’ decision. Therefore, it helps to develop a healthy carbon trading market and facilitate low-carbon economic development.
Since it is important for the market to have firms disclose carbon information, some firms have chosen to voluntarily report their carbon information about their business activities in the absence of specific public-policy requirements. Still, some firms do not act this way. Considering the different choices various firms make, it is necessary to research what determinants are affecting the disclosure of carbon information so that I may apply these results to increase motivations for disclosing carbon information.
Previous studies have researched a few determinants for social responsibility information disclosure such as company size, financial risk, society factors with broad concept. Seldom do they look at the more characteristic factors inside firms, for example, corporate governance and ownership structure. However, a wide variety of studies have attempted to
identify corporate governance as the determinants for certain economic issues such as executive pay, firm performance, earning and tax management and so on (Core et al. 1999; Firth et al. 2007; Bhagat & Bolton, 2008; Xie et al. 2003; Cornett et al. 2009). Hence, the purpose of this essay is to examine the relationship between corporate governance and voluntary carbon information disclosure, as well as the relationship between ownership structure and voluntary carbon information disclosure. More specifically, this paper will investigate the relationship between government-owned and non government-owned enterprise, the propensity and the level of voluntary carbon disclosure, the relationship between some corporate governance characteristics and voluntary carbon disclosure, and attempt to answer the following research question:
RQ: Do corporate governance and ownership structure affect the propensity and the level of voluntary carbon information disclosure in China?
I restricted my samples to Chinese firms because, although the disclosure of carbon emission is quite a new notion, along with the wider subject of environmental reporting like Corporate Social Reports having been studied in diverse settings previously, this determinants type of prior research is mostly based on a western sample of firms. In the US, Stanny(2013) examines voluntary disclosures about greenhouse gas emissions by the US S&P 500 firms which answered to the Carbon Disclosure Project (CDP). There is also a Canadian sample used in the Ben-Amar et al.(2017) paper investigating the effect of female representation in the board of directors on corporate response to stakeholders’ demands. More comprehensively, Luo et al.,(2014) examined whether voluntary carbon disclosure reflects firms’ true carbon performance based on a sample of US, UK and Australian firms. Other European settings were tested as well (Cormier et al. 2005; Larrinaga et al. 2002; Brammer et al. 2006). It is vital to look at what the situation is in other lesser developed and developing countries so that the whole global economy can work together to mitigate the universal climate problem.
From the above-mentioned literature, either the determinants tested in western contexts were overlooked in developing countries, or the focuses were on the propensity of carbon information disclosure; this has inspired me to investigate on corporate governance and
distinguished cultural and social setting and whether there are potential factors such as the typical ownership structure (government-owned firms) that could affect the tendency and the quality of the disclosure of carbon information are open for further research. Moreover, carbon disclosure in China and other developing countries is basically in its initial stage. There is an increasing number of firms in China disclosing carbon information since China joined CDP in 2010 and apparently the disclosure is still at voluntary level. Until 2014, 45 Chinese companies, within 100 firms invited by the CDP organization, have replied to the CDP questionnaires, increased from 32 corporations in 2013. These companies are showing their efforts to contribute to the mitigation of climate change risk. However, compared to the rate of replied companies in Europe (91%), this is still a relatively low rate. Therefore, with China being the largest greenhouse gas emission country, it is relevant to research voluntary carbon information disclosure propensity and quality, and its relationship with firm-level characteristic, with the purpose of stimulating the awareness of environmentally friendly development in business and providing input for systematical carbon information disclosure. The results should be of relevance to stakeholders such as shareholders and government as they would better understand how to manage companies, for example, hiring board members and delivering high-quality carbon information reports.
Previous studies mostly focus on a specific industry ( Kalu et al.2016; Li et al. 2017) which could be a limitation as understanding the results in the whole background. In my thesis, industry boundary would be loosened and tempt to take a look in not only large enterprise (Stanny 2013; Brammer et al. 2006; Cormier et al. 2005) but small- and medium-sized enterprises. Furthermore, few studies have looked at the level of the disclosed carbon information, I attempt to follow a scoring method used in measuring carbon information quality presented in Peng et al.(2015), combined with data in a website providing Chinese listed companies’ annual Corporate Social Responsibility (CSR hereafter) reports ranking every year as measurement for the carbon information disclosure level. Also, most studies only use a single period for research, which is insufficient for the credibility of the research findings. Hence, I also investigate the research question in multiple periods. With the new data, I add new knowledge to the current literature.
I argue that firms with more female directors in board may have no effect on voluntary decision to disclose carbon information, and no effect on the level of voluntary carbon information disclosure. Although evidence has shown that females on the board helped the firm focus more on non-financial aspects (Huse & Solberg, 2006; Ben-Amar & McIlkenny, 2017; Kathyayini et al. 2012), some studies based on Chinese sample show that usually females have no impact on decision making due to the masculine corporate culture in most eastern countries(e.g. Hu & Zhou, 2016). The situation is doubtful in terms of female directors’ roles in the board. In addition, regarding the age situation in board composition, it is a respective that no research has tested yet for environmental information disclosure. I therefore expect that young blood on the board could make some change to the company concerning a more sustainable future, because firms with a younger composition appear to be more likely to focus more on delivering an environmental and sustainable image to the public so that the quality of the carbon information reports would logically be better than those firms comprising of more senior board member. In addition, due to a historical reason, China has a history of having many government-owned companies which leaves a strong impression of China economic characteristic to western countries. Although after 1992, with the Chinese reform of shareholding system, a part of firms which used to be owned by government now transferred to private-owned enterprises. Out of curiosity, I intend to investigate whether or not a firm’s ownership structure would influence its carbon-disclosure behaviour. I argue that government-owned firms are more likely to voluntarily disclose carbon information and also have a higher level of voluntary carbon information disclosure (Wan-Hussin, 2009; Peng et al., 2015). It seems that government-owned firms would be under regulatory pressure and would have sufficient resource to participate in carbon trading market, thus they are willing to disclose their effort to show concern about climate change and sustainability environment and to show it in a high-quality format.
To realize the research design, I use female percentage of the board, mean and standard deviation of ages of the board members to proxy for corporate governance concepts respectively-- gender diversity and age diversity. For ownership structure variable, it is a dummy variable in which 1 stands for the government-owned, 0 for otherwise. When defining whether a company is government-owned, it is decided by capital structure. I identified government-owned companies when the state-owned shares take up equal to or
board size, board meeting, board independence, firm size, leverage and return on assets, are explained in Table 2 in a later chapter.
To measure the decision of whether a firm discloses carbon information, I identify a firm as committed to carbon information disclosure (i.e. CID hereafter from time to time) if either the firm has responded to the CDP China request for public disclosure of climate change strategies and greenhouse gas emissions, or the firm has issued a complete systematic CDP report, or the firm has issued a CSR report with carbon information included. Thus, it is a dummy variable that equals one when meets the above conditions, and zero otherwise. In order to measure the quality of corporate voluntary carbon disclosure, I follow a scoring index method to extract the CID-related information from CSR reports which would deliver a large amount of hand-collected work, and score the collected information pursuant to ISO14064-1.
Using a sample of Shanghai Stock Exchange listed firms in China from the year 2014 to 2016, I collected all the data through CSMAR database and reviewed hundreds of CSR reports. Surprisingly, the findings are generally inconsistent with my expectations. I found that having fewer female directors would increase the likelihood for firms to disclose carbon information. Also, there is no evidence that a younger board would be more likely to make a carbon information disclosure. However, the results show that a less age diverse board would increase the propensity to make CID, while non-government-owned firms are more likely to disclose greenhouse gas emissions information. When it comes to the level of content with firms who commit to carbon information disclosure, the results went in a different direction. I found no evidence that suggests a less age diverse board would have higher disclosure quality, nor that a government-owned enterprise makes clear and complete carbon information disclosures. However, significant coefficients show that a board with fewer female members and of higher average age could provide higher level of voluntary carbon information disclosure. These outcomes are slightly unexpected. The most possible reason for that could be a rough selection of sample or errors appeared in the variables’ value assignment. In spite of this, this thesis can still be seen as a bold attempt to contribute to the academic research fields of environmental disclosure.
As has been elaborated above, this archival study will contribute to the improvement on the practices of voluntary carbon information disclosure in the context of developing countries, and contributes to testifying whether existing determinants in developed countries affect differently in a different background and to examining other potential determinants of voluntary carbon disclosure. What’s more, this research would not only provide the empirical evidence filling the gap in the literature between developed countries and developing countries but also complement the corporate governance and ownership structure studies in the way measuring the quality/level of voluntary carbon information disclosure in multiple periods. Finally, business executives can benefit from the evidence regarding the extent to which the characteristics of governance can influence carbon information transparency.
The rest of this paper is organized as follows. Section 2 presents a literature review and Section 3 develops my hypotheses. Section 4 presents my research design. The main results are discussed in Section 5, and I provide a summary of my results and conclusions in Section 6.
2 Literature review
2.1 Research about CID in diverse setting
In the UK, Brammer et al.(2006) examine the disclosure patterns in voluntary environmental disclosures based on a sample of large UK companies. Similarly, Cormier et al.(2005) examined the environmental reporting quality of large German companies, while studies from a Spanish setting examined environmental disclosure standards in Larrinaga et al.,(2002) paper. There is a Canadian sample used in the Ben-Amar et al.(2017) paper investigating the effect of female representation on the board of directors on corporate response to stakeholders’ demands. As aforementioned examples, previous studies of voluntary carbon disclosure have typically focused on western contexts naturally because of the leading roles which western countries usually act in this controversial environmental issue, while very few have addressed carbon information disclosed by firms in developing countries, particularly those in Asian countries. For instance, Rashid et al. (2008) conducted research on the influence of ownership structures and board practices on corporate social disclosures in Bangladesh, Lee et al. (2015) investigates market responses to firms’ voluntary carbon information disclosure using Korean data, and Kalu et al. (2016) found out the determinants influencing carbon disclosure in real estate companies in Malaysia, also a newly published paper by Li et al. (2017) tested the association between media reporting, carbon information disclosure, and the cost of equity financing by using listed companies in heavy polluting industry in China as research objects.
2.2 Research about determinants of CID
Many studies have investigated the influencing factors of carbon information disclosure and the economic effect following carbon information disclosure. Note that carbon information disclosure is not yet made mandatory, in this case, I am motivated to discover what drives a company to voluntarily disclosing private carbon information. Most of these researches are conducted based on the CDP (Carbon Disclosure Project) report relating to world top 500 companies. For instance, Gonzalez-Gonzalez and Ramirez (2016) studied the drivers of carbon information disclosure using the 2012 CDP report data in which Spanish companies answered to the questionnaires. Also, Ben-Amar and McIlkenny (2017) found a positive
association between board female percentage and the decision to respond to the CDP questionnaires as well as the quality of carbon information disclosure. This paper applied CDP information as the evidence of corporate response to sustainability demands. Similarly, Lee et al. (2015) examine the market reaction to the voluntary carbon information disclosure by firms in Korea involving CDP Korea sample from 2008 to 2009. However, there is a limit in using the CDP report data. Due to the low response to the CDP questionnaires, research on some countries like China could be difficult to carry out especially since China, as a major greenhouse gas emitter, is of great importance for encouraging companies to disclosure carbon information. Moreover, it is hard to generalize the above findings if only focusing on western settings where they have comparatively more data to research on.
The thesis is related to the literature about determinants of voluntary carbon information disclosure. Previous researchers mainly looked at the external economic factors and internal factors concerning the influencing factors of carbon information disclosure. Gonzalez-Gonzalez and Ramirez (2016) found that the tendency of voluntarily give a carbon footprint disclosure is explained by the influence of society pressure, markets threat, shareholders demands, and international interactions. In the Spanish case, the factors with a stronger influence are company size, financial risk, their listing in the IBEX35 and FT500 indexes, and ownership concentration. In Kalu et al.(2016) paper, they testified social and financial market as critical determinant factors for carbon disclosure, meanwhile significant effect on voluntary carbon disclosure was not detected with the economic and institutional factors. Peng et al.(2015) found that companies operating in high-emission sectors are more likely to make CID and tend to disclose more information especially firms which have better performance are more willing to make the CID. Luo et al. (2012) discovered that the willingness to disclose carbon information is related to the degree of a country’s financial development. Other drivers like female representation on the board (Ben-Amar et al. 2017), ownership structure (Rashid et al. 2008) and gender diversity, board independence, and the presence of environmental committee (Liao et al. 2014) were discussed. Nevertheless, among all the studies mentioned above, I can see that there are still some spaces waiting for investigation. Other potential determinants may be concerned with the disclosure such as age diversity on board members and government ownership structure characterized in the Chinese setting.
2.3 Theory regarding voluntary environmental disclosure
One theory that has been widely used in the field is legitimacy theory, arguing that companies tend to legitimize their activities, in order to change the view of the public, through discretionary disclosure (Cho & Patten, 2007). In terms of environmental disclosure, management often claims that their operations do not cause harm to the environment, although this is not necessarily the truth. However, Deegan & Rankin (1996) expressed that the legitimacy point of view may only provide explanation partially, especially when companies confront legitimacy issues. Another frequently admitted theoretical perspective is agency theory, indicating that managers would increase disclosure to mitigate information asymmetry between companies’ insiders and external shareholders (Healy & Palepu, 2001; Jensen, 1988). The traditional agency theory as the shareholder priority model focuses purely on economic functions, ignoring complex issues associated with corporate inefficiencies and multiple incentive issues (Sternberg, 2000). Thus, a broader means was encouraged and Donaldson and Preston (1995) proposed stakeholder theory that extends the traditional perspective to a wide range of legitimate individuals or groups that are impacting, or being affected by the firm. They recognized that these stakeholders have the right to obtain benefits and information, although the priorities of all stakeholders are not always the same and multiple corporate stakeholders may have compatible- or competing- interests (Collier, 2008), and these interests may interact with other aspects to affect the corporate activities change (Adams & Whelan, 2009). Stakeholder theory emphasizes the strength of conflicting stakeholder needs and preferences, and provides a potential explanation for carbon information disclosure for the firm’s response to climate change, its strategic positioning of social and environmental responsibility and the trade-off between economic and ecological objectives (Macve & Chen, 2010).
In my context of environmental carbon dioxide disclosure, stakeholders theory appears especially significant. Emissions of carbon dioxide is everywhere global and exist persistently. To respond to the situation, climate change legislation has been proposed and may either directly or indirectly, favorably or unfavorably, affect the companies, and result in either beneficial or harmful consequences, and sometimes even both. Therefore, companies must formulate strategic decision which would have far-reaching implications for their future development. This kind of decision would ultimately affect numerous interest groups in different ways, thus inevitably be supported by some stakeholders and at the same time be
opposed by others. Hence, information about a firm's carbon strategy and its influence are important on the firm's carbon information report or social responsibility report which has to clearly explain how the suggesting carbon activities would benefit the firm and its stakeholders including those without financial interests, under this cost. This means that stakeholders require carbon information to evaluate the suitability and sustainability of the firm's climate policy(Gray, Kouhy, & Lavers, 1995). In this connection, adopting a broad concept of the management’s multi-accountability to multiple stakeholders with divergent interests, stakeholder theory can probably provide a better explanation for the voluntary carbon information disclosure behaviour studied in this paper.
Last but not least, there are studies involving the effect of voluntary disclosure on firm value. Dhaliwal et al.(2011) have found that firm value is dependent on firm’s carbon information disclosure. Then, Luo and Tang (2014) further disclosed that between carbon disclosure and performance there exists a significant positive association. From a different direction, Kolk et al. (2008) pointed out that the comparability and reliability of carbon information disclosure are of still insufficient power although enterprises which participated carbon information disclosure survey are increasing gradually. Matsumura et al. (2014) used 2006 -2008 CDP report data with S&P 500 companies and found that the more carbon emission the firm emitted, the lower the firm value is. Firm median value of firms that did not disclose their carbon emissions is an average of $2.3 billion lower than that of the disclosing firms. It is noted that firm value is affected by the carbon emission while the firm does its business and the behaviour of voluntary carbon footprint disclosure. Hence, firms which could not disclose carbon information timely and highly effectively would receive a negative market evaluation. As a result, it is supported that voluntary carbon disclosure is beneficial for corporations as well as being important for investigating the potential drivers to facilitate carbon information disclosure.
3 Hypothesis development
3.1 Corporate governance
A wide variety of studies have attempted to identify corporate governance as the determinants of social/environmental disclosure. For example, the structure and characteristics of the board are associated with environmental decisions. Board independence (Eng & Mak, 2003), female directors and institutional investors (Kathyayini et al. 2012), board size, chief executive officer (CEO)/chair duality and non-executive directors (Gul & Leung, 2004), and the presence of a CSR/environmental committee (Michelon & Parbonetti, 2012) are the characteristics which are applied often. In general, the prior literature shows a significant correlation which influenced by corporate governance factors.
This paper selects two interested firm-level factors about board diversity. It is arguable that the diversity of board members expands the possibility of various knowledge, opinions and ideas while board members making decision (Post et al. 2011). They believe diversity in a group where individuals have diverse ages, culture and education backgrounds may lead to information diversity and value diversity. In other words, members in such a community carry different knowledge bases and share various networks, also for example have difference on the altitude towards CSR report. In this sense, I intend to test board gender diversity and board age diversity.
3.1.1 board gender diversity
One of the standout characters is gender diversity. Female proportion of board members is an important aspect of corporate governance, because it is historically accepted that male and female have differences in cultural and society roles. There are studies showing that women are different from men in many dimensions such as personality, communication style, educational background, career experience and professional knowledge. As many researchers think female can make an important advice contributing to the board development, Huse and Solberg (2006) found that females are more devoted and diligent so that eventually created a fairly nice atmosphere on board. At the same time, female directors are thought of being less self-beneficial oriented because of their mother character in nature. Therefore, the participation of female on the board may have positive effect on the social responsibility behaviour of the corporations. Ben-Amar and McIlkenny (2017) have found a positive
association between board female percentage and the decision to respond to the CDP questionnaires as well as the quality of carbon information disclosure. However, there is also evidence shown female directors have no significant effect on corporate performance in the non state-owned companies in China(e.g. Hu & Zhou 2016). It means that female does not help or has no power to influence the decision-making process. Due to the rare percentage of female members on the board in China and its long effected ideas of men dominance, in this thesis, I expect the below hypothesis:
H1a: Firms with more female directors in board have no effect on voluntary carbon information disclosure.
H1b: Firms with more female directors in board have no effect on the level of voluntary carbon information disclosure.
3.1.2 board age diversity
Potential characteristic of the board -- age characteristic of the board members is overlooked by the prior studies examining the determinants of carbon information disclosure. As traditionally acknowledged, most board tables are composed of middle-aged and aged members after retirement, usually either who have worked as CEOs of other companies or who served as esteemed scholars in universities. However, things change, and studies show that companies from the consumer services and product industry sector tend to appoint directors in a more diverse age range (Kang et al. 2007). Although aged group can offer experience and wisdom, the younger group is energetic and has the motivation to succeed so that younger generation focuses on the future, in this way board members with lower average age could care more about sustainability and the whole society development. It is possible to believe younger board member pay more attention to social/environmental disclosure to public rather than focus on financial performance of the enterprise. While young directors composition of the board is a aspect to represent age diversity, the distribution of different age groups could be another way to look at age diversity. Multiple age groups- thirties, forties, fifties, sixties - are merging different values they hold alone. The conflict of values may influence the final decision to information disclosure. This kind of respect is seldom focused previously. There is an uncertain effect considering the real situation of board make-up and abstractly assumption of potential positive correlation with voluntary carbon information disclosure. Hence, this brings forward the hypothesis:
H2a: The lower the average age of board members is and the higher the standard deviation of board members’ ages is, the firms are more likely to voluntarily disclose carbon information.
H2b: The lower the average age of board members is and the higher the standard deviation of board members’ ages is, the firms have higher level of voluntary carbon information disclosure.
3.2 Ownership structure
Some of the prior studies explain the effect of ownership structure on corporate transparency remains an unsettled area of research interest (eg. Wan-Hussin 2009). Ownership structure is a commonly debated subject in China as a specific character among the world. Therefore, enterprises are divided into the government-owned companies and non-government-owned companies. In those corporations dominated by government shareholders, they have more resources and can more effectively carry out national policies because of the privilege they have. Moreover, due to the government is the controlling shareholders, it would have a more accurate control and supervision on the enterprise. Many listed firms still lack the independence and professional ability. At present, China is in a transition period of low-carbon economy. The government has been emphasizing the effective disclosure of corporate social responsibility especially of carbon information. Therefore, the controlling shareholders of government-owned companies will certainly play an exemplary role in these areas, thus leading other corporations to join the carbon information disclosure troops. What’s more, empirical evidence has investigated that firms which are sensitive to regulatory pressure based on socio-political theory are often acting as pioneers in implementing CID (Patten, 2002; Huang & Kung, 2010; Zeng et al. 2011; Peng et al. 2015). This brings forward the second hypothesis of this thesis:
H3a: Government-owned firms are more likely to voluntarily disclose carbon information.
H3b: Government-owned firms have higher level of voluntary carbon information disclosure.
4 Data and method
4.1 Sample selection
My initial sample includes all listed firms with the available corporate governance data in the China Shanghai Stock Exchange for the year 2014-2016. China Shanghai Stock Exchange is the one of the largest financial markets in China. It concludes large amount of influential firms which disclose adequate information to the investors, thereby I have more usable data. At first, I compared the listed firms at Shanghai Stock Exchange with the 100 firms covered in the CDP China annual survey for the period 2014-2016. These 100 firms were asked to voluntarily complete a standard questionnaire on climate change-related risks and opportunities, greenhouse gas emission accounting, carbon performance, and governance. I found out that the firms covered in the annual CDP survey are a mix of firms from China Shanghai Stock Exchange and China Shenzhen Stock Exchange. Due to unavailable access to the CSR reports from firms listed in the Shenzhen Stock Exchange, I investigates my research using listed firms in China Shanghai Stock Exchange. I also collect voluntary carbon disclosure data from CDP reports in China and other corporate social responsibility and sustainable development reports. I hand-extracted specific carbon information from the CSR reports available through certain website and also from the annual reports available through the Shanghai Stock Exchange website because listed companies are required to release their annual reports every year. My initial sample comprises the firms listed in the Shanghai Stock Exchange, 1467 firms each year in total. Of these 1467 firms, in 2014, 463 firm were excluded due to missing data for control variable. In 2015, 397 firms were also excluded due to missing data for control variable. In 2016, 251 firms were again excluded because of missing data for control variable. In the next step, a few firms were excluded due to missing continuous data during the three year period. Thus the final sample A includes 2976 firm-year observations. In order to investigate the effect that the hypothesized determinants have on firms which release carbon information, the subsample - sample B - includes 1110 firm-year observations. Table 1 describes my sample selection and composition. The idea that the sample period starts in 2014 is because from this year on, there are almost half companies in the invited 100 companies participate the CDP China 100 survey, showing an increasing number of corporations willing to voluntarily disclose carbon information, therefore it supposes to provide sufficient observations as many as possible. As can be affirmed by Table
TABLE 1
Sample Selection and Description Panel A: Sample selection
2014 2015 2016
Initial sample of all firms 1467 1467 1467
Less
Missing variables data 463 397 251
Number of firms 1004 1070 1216
Panel B: Sample description
Year Initial Sample A
(ALL Firms)
Sample A (ALL Firms)
Initial Sample B (Firms with CID)
Sample B
(Firms with CID)
2014 1004 992 400 370
2015 1070 992 417 370
2016 1216 992 460 370
Total 3290 2976 1277 1110
4.2 Research design
The Libby boxes presented in the Appendices to this thesis show how the conceptual relation examined in this thesis will be operationalized in the research design. Table 2 provides a summary of my dependent, independent, control variables and explanatory note. In the latter content, dependent variables, independent variables and control variables are discussed in detail.
TABLE 2
Summary of Variables with Explanatory Dependent Variables:
CID A dummy variable which equals to 1 when the firm publicly disclose carbon information in a official way and 0 otherwise.
VCIDL A score rated by me according to the ISO14064-1 standard. The higher the score, the higher the quality of the disclosure of carbon information.
Independent Variables:
FEMALE A percentage which illustrates how many females make up the board at the end of each year.
AGEMEAN The mean of the age of all board members. The lower the number, the younger the board is.
AGE The standard deviation of the age of all board members. The higher the number, the more complex the age diversity in a board is.
OWNERSHIP A dummy variable which equals to 1 when the firm is government-owned and 0 otherwise.
Control Variables:
BOARDMEETING The number of board meeting held in a year.
BOARDSIZE The number of members in a board at the end of each year.
INDEPENDENCE The proportion of independent directors on the total number of directors at the end of each year.
FIRMSIZE It is the natural logarithm of the total book value of the assets of the company at the end of each year.
LEVERAGE It is the ratio that indicates what proportion of debt a company has relative to its assets at the end of each year.
ROA Return on assets, an indicator of how profitable a company is relative to its total assets at the end of each year.
4.2.1 Dependent variables
Two dependent variables are the propensity of carbon information disclosure and the level of carbon information disclosure. The propensity of carbon information disclosure (CID) is defined as a dummy variable that equals one either the firm has responded to the CDP China request for public disclosure of climate change strategies and greenhouse gas emissions or the firm has issued a complete systematic CDP report or the firm has issued a CSR report with carbon information included, and zero otherwise. For the voluntary carbon information
report contains information in the certain category to the standard (basically the CDP requirements), then some points would be given. In this way, I gather the observations of the level of the carbon information disclosure.
To investigate the factors that drove Chinese listed companies to disclose carbon information, I use the score method in Peng et al.(2015) for reference by identifying corporate disclosed information about carbon emissions from eight items (I1to I8). These eight items are closely
referred to CDP’s questionnaire. What follows are the detailed explanatory of the eight items: I1: Targets and results of emission reduction;
I2: Method of measuring carbon emission;
I3: Scope 1 emission data: direct greenhouse gas emissions;
I4: Scope 2 emission data: indirect greenhouse gas emissions of energy;
I5: Scope 3 emission data: other indirect greenhouse gas emissions;
I6: Energy consumption of total operation in the reporting year;
I7: Emissions trading and
I8: Other carbon-related information.
Items I1 to I8 are detailed information about carbon emission. I score these items in
accordance with ISO14064-1.(ISO14064-1 specifies with guidance at the organisation level for quantification and reporting of greenhouse gas emissions and removals.) The scores range from 0 to 2 points. For I1 to I6, no information provided would be given a score of 0, and 1
point is given for general non-quantitative information. For the different items, a score of 2 represents different things. For I1, 2 is assigned for detailed quantitative information
including the time, quantitative goals, completeness of targets; for I2, if there is detailed
measuring process including the method used, the formula used and the parameter applied, a score of 2 is given; for I3, 2 is used for detailed quantitative information including the
boundaries used for Scope 1 greenhouse gas inventory and emissions figures in metric tons of CO2, same for I4 and I5to Scope 2 and 3; for I6, 2 is for detailed quantitative information
including fuel consumption data measured in tons and per value. For I7 , the scoring method
that no information appeared, 1 is given for general non-monetary information like a firm says it joins the carbon trading market, 2 awarded for monetary information about how much a company trades carbon dioxide (Wiseman, 1982). And for I8, 0 is the same, for no
information, 1 would be indirect information includes such information as the goals or plans of the company for climate change and may just a mention of the firms’ concern for climate change or environmental pollution. 2 is fulfilled when the information related to carbon is showed in monetary but not about carbon trading.
For example, I scored the level of the other indirect greenhouse gas emissions in the reporting year, that is I5, as follows: the 2016 annual CSR report of China Eastern Airlines says that ‘In
2016, China Southern Airlines consumed aviation oil 7.310.600t compared to 6.767.500t in 2015, this number is to be equivalent to 23.028.300t of CO2’. Based on the scoring method,
China Southern Airlines gets two points in this item I5. In contrast, the 2015 annual CSR of
Nanjing Iron and Steel says that ‘ the project is effective in improving the gas quality and realizing extremely low particulate matter emission.’. For this kind of disclosure, the company gets one point for I5. Another example would be scoring for I8, other carbon-related
information. The 2014 annual CSR report of Hua Xia Bank says that ‘ To encourage the development of green credit, by the end of 2014, the loan to steel, cement and other three serious overcapacity industries reduce 5,321 billion RMB compared to the beginning of the year’. Based on the scoring method, Hua Xia Bank gets two points in this item I8.
Each company is assigned to a score of CID level based on the equation below:
)
(
)
(
8 1 j i j iScore
I
CID
Score
where Score(CID)i is the total score of CID for firm i; and Score (Iij ) is the score of the jth
item for firm i, in which j = 1, 2, ... , 8. As I have set up a dummy variable equal to 1 for firms with CID, if the Score(CID)iis 0, it has the same effect as on the firms that do not meet
the conditions for CID. Then I restrict my attention to firms that actually have carbon information disclosed (i.e., Score(CID)iis greater than 0) and use the value of Score(CID)ias
4.2.2 Independent variables
The independent variables: female percentage on the board (FEMALE) stands for the gender diversity, the mean and the standard deviation of ages of board members (AGEMEAN and AGE) presents the age diversity, and ownership structure (OWNERSHIP) is also a dummy variable in which 1 stands for the government-owned, 0 for otherwise. Proxies for corporate governance (i.e. female percentage and, means and standard deviation of board members’ ages), and ownership structure data are available through CSMAR database. The China Stock Market & Accounting Research (CSMAR) Database offers data on the China stock markets and the financial statements of China’s listed companies. Therefore, my independent variables can all be calculated using the data in CSMAR database by managing the number metrics.
(i) Proxies for corporate governance
For board gender diversity, I use the percentage of female directors, calculated as the number of women directors divided by the total number of directors. Lots of papers have used this proxy for gender diversity (e.g. Ali et al., 2014; Ben-Amar et al., 2017; Liao et al. 2014; Kang et al. 2007; Kathyayini et al., 2012). For board age diversity, I propose two proxy for it, which are the mean of the board members’ ages and the standard deviation of board members’ ages. Larger standard deviation(larger age differences between board members) and lower mean age (higher representation of young board members) would generate higher age diversity values (Ali et al. 2014; Kang et al. 2007). By implementing two proxies for age diversity, it should provide different angles to explain the results and eventually come to a unified conclusion.
(ii)Proxies for ownership structure
For ownership structure (OWNERSHIP) is a dummy variable in which 1 stands for the government-owned, 0 for otherwise. Though some studies investigate ownership structure in China, it is unclear how they measure for this variable (e.g. Zhamg, 2013; Chen et al., 2006; Xiao & Yuan, 2007). Theoretically, when defining whether a company is government-owned, it is decided by capital structure. I identified government-owned companies as the state-owned shares take up equal to or greater than fifty percent of total share capital in the company. These initial data is also can be accessed to the CSMAR database.
4.2.3 Control variables
As has showed in the summary Table 2, I chose six control variables for the model. BOARDMEETING, BOARDSIZE, INDEPENDENCE are three control variables related to the corporate governance, while FIRMSIZE, LEVERAGE, ROA are control variables related to firm performance. For the feasibility of the research, I selected those most common but also influential control variables on the basis of previous studies on voluntary disclosure. For example, Allegrini & Greco(2013) testified that diligent monitoring activity is associated with greater transparency meaning if the number of board meeting held in a year is high, it could impact the carbon information disclosure. They also confirmed that board size has significant positive correlations with the level of voluntary disclosure with Italian sample. Though it was not supported in the results, board independence was investigated in the hypothesis. Ahmed and Courtis (1999) conducted a meta-analysis of 29 studies and confirmed significant and positive associations between the levels of disclosure and firm size and leverage. Firm size (FIRMSIZE) has been affirmed that it is significantly and positively correlated with disclosure level in a number of studies, indicating that larger enterprises disclose more information, either mandatory or voluntary, than smaller companies (Cooke, 1989a,b; Meek et al. 1995; Hossain et al. 1995; Camfferman & Cooke 2002). I measured firm size (FIRMSIZE) with the natural logarithm of the total book value of the assets of the company at the end of each year. Those companies which are with a high level of debt would try to reduce monitoring costs by disclosing more information (Jensen & Meckling 1976). I presented leverage (LEVERAGE) as the ratio that indicates what proportion of debt a company has relative to its assets at the end of each year (Depoers, 2000; Eng and Mak, 2003; Peng et al.,2015), assuming a positive relationship with voluntary disclosure. Profitability ratios usually exist in empirical research on voluntary disclosure (Meek et al., 1995; Ahmed & Courtis, 1999; Ho & Wong, 2001; Camfferman & Cooke, 2002). Companies with good performance tend to voluntarily disclose more information (Meek et al., 1995). I measured profitability with the ratio of return on assets, an indicator of how profitable a company is relative to its total assets at the end of each year. It should be a positive relationship with voluntary disclosure. I control for these six variables in my model.
4.3 Regression model
As aforementioned, two related issues are tested in my research, which are (i) the decision of whether to disclose; and (ii) the content of disclosure. The choice of firms to disclose carbon emissions information in model as a function of various firm specific features and is tested in a logistic model as follows:
i
Con
OWNERSHIP
AGE
AGEMEAN
FEMALE
CID
1
2
3
4
5
Con means control variables. To be specific, it would be the model (1) below.
CID = α + β
1FEMALE+β
2AGEMEAN+β
3AGE +β
4OWNERSHIP
+β
5BOARDSIZE+β
6BOARDMEETING+β
7INDEPENDENCE
+β
8FIRMSIZE+β
9LEVERAGE+β
10ROA+ Ɛ
i(1)
As for the content quality of the disclosure, a linear OLS regression analysis is applied:
i Con OWNERSHIP AGE AGEMEAN FEMALE VCIDL
1
2
3
4
5
Again, Con means control variables. To put it more clear, it would be explained by model (2) below.VCIDL = α + β
1FEMALE+β
2AGEMEAN+β
3AGE +β
4OWNERSHIP
+β
5BOARDSIZE+β
6BOARDMEETING+β
7INDEPENDENCE
5 Empirical results
5.1 Descriptive analysis
Table 3 shows the summary statistic of variables. TABLE 3
Descriptive Statistics of Variables
Variable Obs Mean Std.Dev. Min Max
CID 2,976 0.411 0.492 0 1 VCIDL 1,110 1.895 2.399 1 15 FEMALE 2,976 0.163 0.100 0 0.455 AGEMEAN 2,976 50.16 3.114 34.53 61.36 AGE 2,976 7.451 2.189 1.684 32.57 OWNERSHIP 2,976 0.0262 0.160 0 1 BOARDSIZE 2,976 9.028 2.006 3 19 BOARDMEETING 2,976 10.26 4.773 2 56 INDEPENDENCE 2,976 0.373 0.0539 0.231 0.800 FIRMSIZE 2,976 22.80 1.739 17.28 30.81 LEVERAGE 2,976 0.514 0.216 0.0845 0.979 ROA 2,976 0.0259 0.0570 -0.228 0.174
In order to decrease the effect generated from extremes, I winsorized variable FEMALE, LEVERAGE and ROA. Showed by the standard deviation, each variable can be seen as a normal distribution. Among all 2976 firm-year observations, 41.1% have disclosed carbon information. Female averagely take up 16% of the board positions, which is not very high. Average age of board directors is closely to age fifty, suggesting an elderly board composition throughout 2976 observations. One surprising notice should be the low proportion of the government-owned companies in the sample, only 2.62% firms have exceeding 50 percent shares owned by the government. Based on Sample B, namely sample with CID, the general quality/level of disclosed carbon information is very low, getting less
Table 4 presents Pearson correlation coefficients and shows 0.1 significance level with a star. TABLE 4
Pearson Correlation of Variables
CID FEMALE AGEMEAN AGE OWNERSHIP BOARDMEETING BOARDSIZE INDEPENDENCE FIRMSIZE LEVERAGE ROA
CID 1 FEMALE -0.1323* 1 AGEMEAN 0.2061* -0.2409* 1 AGE -0.1374* 0.1595* -0.1642* 1 OWNERSHIP -0.0045 -0.0396 0.0502 -0.0753* 1 BOARDMEET ING 0.0566 0.0433 -0.1105* 0.0351 -0.0229 1 BOARDSIZE 0.2170* -0.1487* 0.2568* -0.0937* 0.0135 0.0113 1 INDEPENDEN CE 0.0122 0.00670 0.0359 0.0272 -0.0245 0.0569 -0.3816* 1 FIRMSIZE 0.4101* -0.2379* 0.4295* -0.1529* 0.0526 0.1724* 0.4087* 0.0432 1 LEVERAGE 0.1333* -0.1047* 0.1185* -0.1302* 0.0182 0.2216* 0.2095* 0.0216 0.4376* 1 ROA 0.0647* 0.0647* 0.0712* 0.0920* 0.0285 -0.0768* -0.00780 -0.0373 0.0448 -0.3 1 Note: * Significant at 10% level.
Surprisingly, CID and FEMALE are negatively related. From the coefficients standing for board age diversity (i.e. AGEMEAN and AGE), I saw higher mean and lower age deviation lead to choice to disclose carbon emissions. Note that AGEMEAN and AGE are negatively related. Consistent with what has been documented by the prior research (e.g. Cowen et al., 1987; Deegan & Gordon, 1996; Stanny and Ely, 2008), firm size (FIRMSIZE) is positively correlated with the possibility to make environmental disclosure, and firms probably make this disclosure to keep obtaining legitimacy. FEMALE and BOARDSIZE as well as FIRMSIZE, are negatively related, suggesting that larger board and larger firms dilute female positions.
5.2 Corporate governance
Table 5 reports the results of H1a, H1b, H2a and H2b, including the logistic model and OLS model. It shows the results of two proposed driven factors in the scope of corporate governance for Chinese companies to disclose the carbon information and to influence the
content of the disclosure of carbon information. As Table 5 shows, the coefficient on my first variable of interest, FEMALE, is negative but insignificant in the logistic model, suggesting that boards with higher percentage of female directors are reluctant to disclose carbon information. However, in the OLS model, the strongly significant negative coefficient for FEMALE rejects my hypothesis H2b, indicating female not only have no voice in the decision to disclose greenhouse gas emission but worse, their opinions would very likely to be against. It could be resulted from the cultural background in long-history China that views of a woman would be treated as shortsighted or worthless views not to be taken seriously, thus what against those views would be a right option. Another variable set representing board age diversity to test the hypotheses is AGEMEAN and AGE. They are both the opposite to predicted in both models. The coefficient for AGEMEAN in the logistic model is positive but insignificant, but in the OLS model is strongly positive significant. This means that the elder the board directors are, the more likely they seem to be to disclose carbon information, and the high quality of the carbon information disclosure is. From another respect, the coefficient for AGE in the logistic model is negative and strongly significant, but in the OLS model is negative but insignificant. The indication of the results shown by AGE is similar to the AGEMEAN variable. A diverse board with regards to age, which can be explained by a board containing multiple age groups such as board members aging from thirties to sixties instead of board members all aged around forties and fifties, is less likely to voluntarily disclose carbon information. This could be rationalized by an argument that a diverse community has difficulties in unifying decision due to conflict of various value in individuals. In summary, the results generally rejected all the hypothesis for corporate governance. In the logistic model only one coefficient of interest is significant. A reason for that could be the limitation of the sample. The sample overall was relatively small due to shortage of time and resource. As for the OLS model, the dependent variable relies a lot on researcher’s subjective judgement, which resulting in biased observations caused by some extent of uncontrollable factors. I admitted that while scoring the CSR reports, though I maintained the consistency throughout the whole process when reviewing hundreds of reports, I hesitated at first, for instance, for whether to put two points in one item (e.g. I4) or put each
two points in two items (e.g. I4 and I5) when the information is combined. Therefore, this
method I believe is more suitable for experienced researcher who has more intuition on the content and the category classification. However, looking at the control variables, the results are correspondent to the results in Peng et al. (2015), where it shows that increase in firm size
facilitates the likelihood to make CID, arguing by the authors that larger firms have higher greenhouse gas emissions.
TABLE 5
Corporate Governance Effect on Disclosure Carbon Emissions and Disclosure Quality
Logistic Model OLS Model
VARIABLES Predicted Sign CID VCIDL
FEMALE +/- -0.338 -2.325*** (0.439) (0.778) AGEMEAN - 0.00614 0.163*** (0.0206) (0.0386) AGE + -0.0993*** -0.0514 (0.255) (0.460) BOARDMEETING -0.00115 -0.00857 (0.00920) (0.0149) BOARDSIZE 0.101*** 0.0675** (0.0264) (0.0343) INDEPENDENCE 1.081 -0.579 (0.878) (1.318) FIRMSIZE 0.612*** 0.250*** (0.0406) (0.0576) LEVERAGE -0.723*** -0.126 (0.261) (0.460) ROA 1.798** 1.202 (0.908) (1.423) Constant -15.16*** -11.69*** (1.045) (1.423) Observations 2,976 1,110 R-squared 0.160
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
5.3 Ownership structure
Table 6 reports the results of H3a and H3b, including the logistic model and OLS model. It shows the results of the third proposed driven factor, ownership structure, for Chinese companies to disclose the carbon information and to influence the content of the disclosure of carbon information. Similar to the findings reported in Table 5, the results of the variable of
interest rejected the hypothesis with moderate significant in logistic model. H3a states that government-owned firms may be more likely to voluntarily disclose carbon information. This hypothesis is refused by the significantly negative coefficient of OWNERSHIP, as Table 6 reports -0.551. In this regard, I related my research to Peng et al.(2015). They had a similar research generating a result that the coefficient for GOV (which indicates the same as OWNERSHIP in my research) is positive but insignificant, concluding that the decision to disclose carbon information has no difference from that of private firms in China. This contradicted finding suggests that the effect of government-owned companies has on voluntary disclosure of carbon information may still in vague. I could argue for my result that the negative relationship comes from the confidential obstacle. Another possible reason could be the error in sample selection, as Table 2 shows, only 2.62% firms in the sample owned by the government. Others such as Peng et al.(2015) could argue that under the regulatory pressure, government-owned companies should disclose carbon information more probable than those not owned by government. Also, government-owned companies have more resources and can more effectively carry out national policies on disclosure transparency. This area of whether ownership structure is a driven factor for decision to voluntarily disclose carbon information remains unsettled. I strongly suggest subsequent researchers choose a more suitable sample, and maybe try to carefully define how is a company owned by government, because if a company is government-owned, it could be the government capital shares are more than 50 percent, but also could be government has significant control through significant minority ownership. In latter case, the proxy for OWNERSHIP which equals to 1 when government capital shares are more than 50 percent of the total capital shares could be missing some other firms that under significant control by significant minority government ownership. Thus, if time and resources permitted, researches should assign dummy variable based on the annual reports or other sources which clearly show whether a entity is under government significant control, to decide the true value for the firm.
As for the OLS model, testing the content quality of carbon information disclosure, namely the level of carbon information disclosure, the coefficient is 0.716 which is positive but insignificant. As I found no evidence to support H3b, I may conclude that the carbon-disclosure content quality is indiscriminate between government-owned companies and private enterprise such as family business. This result, however, consistent with the findings
the dependent variable itself contain many interference beyond control because it subjects to personal idea and subjective judgement. Hence, I would probably conclude that no clear evidence is found for OWNERSHIP in OLS model. In sum, the findings refused H3a, and I found no evidence supporting for H3b.
TABLE 6
Ownership Structure Effect on Disclosure Carbon Emissions and Disclosure Quality
Logistic Model OLS Model
VARIABLES Predicted Sign CID VCIDL
OWNERSHIP + -0.551** 0.716 (0.255) (0.460) BOARDMEETING -0.00115 -0.00857 (0.00920) (0.0149) BOARDSIZE 0.101*** 0.0675** (0.0264) (0.0343) INDEPENDENCE 1.081 -0.579 (0.878) (1.318) FIRMSIZE 0.612*** 0.250*** (0.0406) (0.0576) LEVERAGE -0.723*** -0.126 (0.261) (0.460) ROA 1.798** 1.202 (0.908) (1.423) Constant -15.16*** -11.69*** (1.045) (1.423) Observations 2,976 1,110 R-squared 0.160
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
6 Conclusion and limitation
As environmental pollution continues to increase, enterprises have been encouraged and demanded by their stakeholders and at the meantime by regulators to improve and disclose their carbon emissions management. In recent years, China has also been taking actions in this field. China’s listed companies now in their annual CSR reports disclose carbon emission information, directly or indirectly, qualitatively or quantitatively, and document how they deal with issues associated to climate change and environmental pollution.
In this thesis paper, I use the data both from the CSMAR database and the data collected manually from the annual CSR reports of listed companies in China Shanghai Stock Exchange during 2014-2016 fiscal year period to investigate two issues: (i) what kind of factors lead the companies to decide whether or not to make carbon information disclosure; and (ii) what forces influence the degree of content, carbon emission information, disclosed. In this paper, I examined both corporate governance specifically regarding to board characteristic and ownership structure, and I found that fewer female directors would increase the likelihood for firms to disclose carbon information. My study also shows no significant evidence but potential that a younger board would be more likely to make carbon information disclosure. However, the results show that a less age diverse board would increase the propensity to make CID, while non-government-owned firms intend to disclose more greenhouse gas emissions information. When it comes to the level of content with firms which commit to carbon information disclosure, the results went to a different direction. I also found no evidence that suggests a less age diverse board would have higher disclosure quality, nor that a government-owned enterprise makes clear and complete carbon information disclosure. In sum, firms which are owned by government are less willing to make CID, but the content of their disclosure does not significantly differ from those of private business. Finally, significant coefficients show that a board with fewer female members and of higher average age could provide higher level of voluntary carbon information disclosure.
My research contributes to the literature in several aspects. First, relying on a new dataset, I contributes to the improvement on the practices of voluntary carbon information disclosure in
determinants in developed countries affect differently in a different background (i.e. Chinese firms) and to examining other potential determinants of voluntary carbon disclosure. This could both be essential and be of interest as the carbon emission problems are generally serious in developing countries in which fossil fuels are still widely used. What’s more, this research would not only provide the empirical evidence filling the gap in the literature between developed countries and developing countries but also complement the corporate governance and ownership structure studies in the way measuring the quality/level of voluntary carbon information disclosure in multiple periods. In addition, my research may help the growing number of researches that investigate the degree of participation of firms in the environment protection actions. Finally, business executives can benefit from the evidence regarding the extent to which the characteristics of governance can influence carbon information transparency.
Nevertheless, my research remains a few limitations. Firstly, the sample is relatively small. As aforementioned, firms responded to annual CDP China survey are a mix of firms from China Shanghai Stock Exchange and China Shenzhen Stock Exchange. For more creditable and comprehensive results, I suggest subsequent researchers include a larger sample. In addition, I think the proxy for board age diversity in the thesis is statistically general, as I suppose age diversity should be presented as detailed groups categories. Also, the observations are not strictly credible to some extent. For example, for ownership structure variable, if a company is government-owned, it could be the government capital shares are more than 50 percent, but also could be government has significant control through significant minority ownership. As discussed in the results chapter, in latter case, the proxy for OWNERSHIP which equals to 1 when government capital shares are more than 50 percent of the total capital shares could be missing some other firms that under significant control by significant minority government ownership. My research may suffer from potential biases of incomplete information. This could be verified by the abnormal low percentage of government-owned firms in the sample (2.62%) while each observation year at least has almost 33 percent of firms make CID. And for the VCIDL variable which measures the quality of the disclosure of carbon information is scored involving a sense of personal subjective judgement. While reviewing the CSR reports and the carbon information disclosed, I struggled to give points to certain statement. For instance, some firms would say how much they invested in a project that is equivalent of an amount of carbon dioxide reduction. I would
waver in this example to assign two point in I6. If another researcher is make to review the
same statement, he/she might assign one point in I8. Therefore, the final regression results
may be entirely different. This is a limitation concerned with the method. Finally, the restricted amount of control variables could be a limitation as well.
Recently, China has begun to energetically take up in the global group for controlling carbon dioxide emissions, but the outcomes do not seem to reach expectation. China remains high growth rate of economy along with the environmental problems becoming increasingly severe. Although my research paper attempts to investigate the factors that affect China’s listed companies’ management of carbon information including the propensity and the level of the disclosure, many problems still have not been answered. Previous studies regarding to environmental disclosure often document a lack of significance(e.g. Patten 2002; Eng & Mak, 2003; Peng et al. 2015), My study, though in which the findings are not ideal, complements a small part of the big blue print. There exist further improved questions that need to be answered in the process of rapid development in China, which bring us future research directions. For example, while I was scoring the CSR reports, I noticed changes between years, and this phenomenon is widespread. What causes the change between years? Would it be a certain policy or an event that leads to the change, for instance making firms accountable to their stakeholders? In this direction, social psychology and other noneconomic theories could supplement stakeholder views in most studies.
References
Adams, C.A., and Whelan, G. (2009). Conceptualising future change in corporate sustainability reporting. Accounting, Auditing & Accountability Journal, 22(1), pp. 118-143.
Ahmed, K., and Courtis, J.K. (1999). Associations between corporate characteristics and disclosure levels in annual reports: A meta-analysis. The British Accounting Review, 31, pp. 35-61.
Ali, M., Ng, Y.L. and Kulik, C.T. (2014). Board age and gender diversity: a test of competing linear and curvilinear predictions. Journal of Business Ethics Vol.125(3), pp. 125- 497. Allegrini, M. and Greco, G. (2013). Corporate boards, audit committees and voluntary
disclosure: evidence from Italian Listed Companies. Journal of Management and
Governance vol.17(1), pp. 187-216.
Ben-Amar, W., Chang, M., and McIlkenny, P. (2017). Board gender diversity and corporate response to sustainability initiatives: evidence from the Carbon Disclosure Project,
Journal of Business Ethics, 142(2), pp. 369-383.
Bhagat, S. and Bolton, B., (2008). Corporate governance and firm performance. Journal of
corporate finance, 14(3), pp.257-273.
Brammer, S. and Pavelin, S. (2006). Voluntary environmental disclosures by large UK companies. Journal of Business Finance and Accounting, 33, pp. 1168-1188.
Camfferman, K., and Cooke, T. (2002). An analysis of disclosure in the annual reports of U.K. and Dutch companies. Journal of International Accounting Research, 1, pp. 3-30.
Chen, G., Firth, M., Gao, D.N. and Rui, O.M., (2006). Ownership structure, corporate governance, and fraud: Evidence from China. Journal of Corporate Finance, 12(3), pp.424-448.
Cho, C., and Patten, D. (2007). The role of environmental disclosures as tools of legitimacy: a research note. Accounting, Organizations and Society, 32(7–8), pp. 639-647.
