Faculty of Economics and Business
MSc. Accountancy & Control
Thesis:
The relation between sustainability performance and financial
performance: comparing more emission-intensive industry and less
emission-intensive industry
Student name: Xiaoqing Chen
Student number: 10824103
Supervisor: Dr. Georgios Georgakopoulos
Version date: 22 June, 2015
Word count: 16,825
Statement of Originality
This document is written by student [Xiaoqing Chen] 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.
Table of Contents Abstract ... 4 1. Introduction ... 5 1.1 Background ... 5 1.2 Research Question ... 8 1.3 Motivation ... 8 1.4 Structure ... 10 2. Literature Review ... 10
2.1 System Oriented Theories ... 11
2.2 Sustainability Disclosure and Financial Performance ... 13
2.3 Carbon emission and Financial Performance ... 16
2.4 ESG and Financial Performance ... 18
3. Research design ... 21
3.1 Data and Sample Selection ... 21
3.2 Model ... 24
3.3 ESG Classification ... 25
3.4 Industry Classification ... 26
3.5 Measures and Control Variables ... 29
3.5.1 Stock Market based Approach Control Variables ... 29
3.5.2 Accounting based Approach Control Variables ... 30
3.6 Correlation analysis and Regression analysis ... 32
3.7 Validity and Reliability ... 34
4. Result ... 35
4.1 Correlation Analysis ... 35
4.1.1 Full Sample Analysis ... 35
4.1.2 More emission-intensive industries ... 41
4.1.3 Less emission-intensive industries ... 44
4.1.4 Comparing more emission-intensive and less emission-intensive industries ... 46
4.1.5 Correlation analysis on ESG ... 47
4.2 Regression Analysis ... 51 5. Discussion ... 54 6. Conclusion ... 54 References ... 58 Appendix 1 ... 62 Appendix 2 ... 63 Appendix 3 ... 64 Page 3 of 64
Abstract
This study investigates the association between financial performance measures and sustainability factors of American corporations. Three factors of sustainability, environment, social and governance, are disentangled to separately examine the correlation with each financial measure. The research method used is based on the paper of McGuire et al. (1988) titled “Corporate Social Responsibility and Firm Financial Performance”. KLD and COMPUSTAT databases are used as data sources for compiling the needed dataset, the KLD database provides sustainability data where COMPUSTAT provides financial performance data. The obtained data set is divided into two subgroups for comparison: companies of more emission-intensive industries and companies of less emission-intensive industries. Correlation analysis and regression analysis were conducted to analyse the dataset and reach conclusions. However, no significant correlations were found between the majority of financial performance measures and sustainability factors and after applying the division into more emission-intensive and less emission-intensive industries no significant results were found either. Only one financial measure, average asset, indicates significant correlations with sustainability factors. Findings demonstrate that (1) stock based financial measures show weak correlation with sustainability; (2) average asset appears to have a significant and positive correlation with sustainability; (3) there is no significant difference in correlations between more emission-intensive industries and less emission-intensive industries; (4) the associations between financial measures and independent components of ESG initiatives need to be further investigated.
Keywords: Quantitative research, Sustainability, Carbon emission, ESG, Return on Asset, Beta, Risk measures, Correlation analysis, and Regression analysis.
1. Introduction
1.1 Background
Sustainability information draws a lot attention amongst investors as investors nowadays not only analyse the financial performance of a firm before investing, but also examine whether a firm is behaving socially responsible. This ever increasing attention leads to the number of companies engaging into issuing sustainability reports mushrooming over the past decades. According to Solomon et al. (2011), sustainability reporting stands for a response to needs for greater corporate accountability for environmental impacts and has experienced rapid growth over the last decades. Since 1989, the publication of the first standalone sustainability reports, the number of companies that has started to release information on its environmental, social or sustainability has increased substantially (Kolk, 2004). Investigations (KPMG, 2011) have demonstrated that an increasing number of companies tend to disclose non-financial information in regards to corporate social responsibility (also refer to as “CSR”). KPMG (2011) reports that the total number of reporting amongst N1001 companies increased by 11 percentage points, to 64 percent in 2011.
The reasons for releasing sustainability reports are discussed in a variety of reports and researches, the main-drivers for releasing these reports can be classified into ethical motivation and economical motivation. KPMG (2011) illustrates that a series of reputational and ethical considerations are on the top of the list of global business reporting drivers for G2502 companies. This includes considerations such as reputation and brand, ethical correctness, employee motivation, innovation and learning, risk management and risk reduction, access to capital and increasing shareholder value, etc. According to GRI (2013), sustainability reporting possibly adds value in areas such as building trust, improved processes and systems, progressing vision and strategy, reducing compliance costs and adding competitive advantages. Furthermore, the legitimacy theory of social disclosure (e.g., Lindblom, 1994; Milne& Patten, 2002) suggests that the extent of environmental disclosure in a financial report is a function of exposure to public pressure in the social environment
1 The largest 100 companies in each of 34 countries surveyed. 2 The largest 250 companies worldwide ranked by Fortune.
Page 5 of 64
in which a company operates (Cho & Patten, 2007). Deegan & Blomquist (2005) argue that companies consider environmental reporting as an obligation because they feel the pressure from the public. Also by not complying to this apparent demand for environmental and social sustainability reporting, companies fear that their legitimacy might be harmed.
A CSR report includes both voluntary and non-voluntary disclosures. The CSR reports are usually known by many different names. The most common names are “sustainability reports,” “environmental reports,” or “citizenship reports,” and represent separate compilations of information about companies’ social and environmental actions (Dilling, 2010). Most of sustainability reporting focuses on the environmental, social and governance (referred to as “ESG”) issues. Environmental issues include biodiversity, climate change, water, and carbon emissions. Social issues include human rights, child labour, social displacement, and financial inclusion. Governance issues include board structure, remuneration, bribery and corruption, and disclosure (PRI, 2013). This thesis examines the correlation on corporation financial performance and disentangled ESG components in sustainability investment (these three elements are explained further in the literature review).
The general public usually expects that companies’ policies and views on employee relations and human rights are perceived as being more important and relevant for labour-intensive industries such as manufacturing, compared to less labour-intensive enterprises such as these primarily based on service, finance and technology. Product quality and diversity is prioritised by industries like manufacturing, chemical and pharmaceutical, compared to industries like service based industries. Similarly, investors and stakeholders are likely to pay more attention to environmental impacts when looking at the CSR activities of more emission-intensive industries such as energy and transportation. Companies choose to emphasise different ESG elements in their sustainability reporting, relevant to the nature and mission of their business. The three components (E, S and G components) of ESG scores have different impact over financial performance (De and Clayman, 2010). “Climate change and carbon footprints are among the most urgent concerns facing society and are key issues of corporate responsibility” (Hrasky, 2012).
One of the most commonly heard terms in environment-related sustainability reports is “carbon”, this term refers either to the principal greenhouse gas or carbon dioxide alone, and is also commonly used as a shorthand for all greenhouse gases (Bebbington and Larrinaga-Gonza´lez, 2008p. 714, note 1). In the current society, the primary producers of carbon emissions are companies. Greenhouse gas is mainly emitted on a large scale as a result of companies’ business activities. As public interest and criticism of these emissions is substantial, this drives company stakeholders to raise awareness on disclosures about environment related information. Patten (2002) also argues that companies with comparatively poor environmental performance face greater exposures to stakeholders, and would be expected to provide more extensive environmental disclosures. Hence, it is to be expected that the investors and stakeholders would pay more attention to environmental aspects (greenhouse emission, climate change, etc. ) in the sustainability reports of more emission-intensive industries. This research intends to investigate companies’ disclosures on sustainability in more emission-intensive industries (such as manufacturing) and less emission-intensive industries (such as services or technology) separately.
Nonetheless, apart from external pressure concerning environmental and social sustainability, financial performance is still the most important metric for businesses. Solomon et al. (2011) argues that powerful stakeholders are often not interested in the company’s impact on the global climate change, however they view it as something that may impact their investment’s financial performance. This indicates that there likely is a correlation between sustainability investing and financial performance for companies. Many researchers (Jones et al., 2007; Aupperle et al., 1985; Kim et al., 2012; McGuire et al., 1988; etc.) have examined this correlation closely. For example, Jones et al. (2007) find a strong statistical relationship between sustainability disclosure and various financial performance indicators; McGuire et al. (1988) find a significant association between sustainability rankings and accounting based financial performance measures. In particular, variables found to be statistically significant in the analysis included important measurements such as operating cash-flows, cash position, working capital, asset backing, retained earnings, capital expenditure, debt servicing, capital structure and price to book value (Jones et al. 2007; Kim et al, 2012; Page 7 of 64
Galema et al, 2008; Humphrey et al., 2012). However, Kim et al. (2012) classifies sample firms into all kinds of industries, and Humphrey et al. (2012) investigates ESG factors separately with all kinds of FTSE industry groups. Barely any prior research investigation on the association between sustainability performance and financial performance in different emission levels of industries has been performed3.
1.2 Research Question
As mentioned in previous section, some financial performance indicators appear to be significantly related to sustainability disclosure and sustainability investing (Jones et al., 2007; McGuire et al., 1988; Ullmann, 1985; etc.). Companies with heavier emission are expected to disclose more sustainability information (Hrasky, 2012). This research aims to conduct study based on a number of prior researches, e.g. McGuire et al. (1988), Galema et al. (2013), Aupperle et al.(1985), Kim et al (2011), Jones et al (2007), Humphrey et al (2012), Lo & Sheu (2007), in order to determine whether this relationship is actually significant. The research will focuses on a large sample of American companies which are divided into two categories: more emission-intensive industries and less emission-emission-intensive industries. More specifically, this study will investigate the impact of disentangled ESG components (environment, social and governance) in sustainability performance to financial performance indicators including accounting-based return and market-based return in more emission-intensive industries and less emission-intensive industries.
Therefore, this studies’ main research question can be stated as:
What is the relation between sustainability performance and financial performance in regards to more emission-intensive industry and in less emission-intensive industry?
1.3 Motivation
A multitude of prior research has been performed on this topic with mixed results (Jones et al. 2007; Ullmann, 1985; McWilliams and Siegel, 2000; McGuire et al,
3 The idea was generated from Hrasky (2012), she discusses two types of industries with different emission levels: more emission-intensive industries and less emission-intensive industries behave differently in legitimation.
Page 8 of 64
1988; Aupperle et al, 1985; Lee et al., 2009; Galema et al., 2008) – these results will be discussed in the literature review section. Due to the fact that prior research was contradictory, the relation between financial performance and either CSR or sustainability performance is still unclear4. The inconsistency of results (refer to footnote 1) from previous studies on the relation between CSR and financial performance is not surprising because different models and regressions are used throughout these studies. Moreover, the existing researches (e.g. Jones et al. 2007; Ullmann, 1985; McWilliams and Siegel, 2000; McGuire et al, 1988; Aupperle et al, 1985; Lee et al., 2009; Galema et al., 2008; Kim et al, 2012) generally focus on the interrelation between CSR and financial performance across all industries in their dataset – rather than separately looking at highly emission-intensive and less intensive industries. This research is considered to contribute to prior literature in two ways: scientific contribution and societal contribution.
From scientific aspect, there are several researches that have been performed in order to investigate the relationship between sustainability reporting and financial performance; however, each single one appears to have some shortcomings (e.g. limited region, small sample size, narrow industry type, etc.). Jones et al. (2007) only focus on Australian ASX companies, Yu et al. (2009) and Clacher & Hagendorff (2012) focus on companies listed on the London stock exchange, McGuire et al (1988) examined out-of-date data from the 1980s. Moreover, none of the aforementioned work has distinguished industries by their impact on the environment, the society, and the governance5. Research conducted by Hrasky (2012) classified industries by emission level, however, the sample size (50 companies) in this research is not significant and the focus is solely on Australian companies. Humphrey et al (2012) mentioned that Brammer et al.(2006) and Waddock & Graves (1997) also investigated the association between CSR reporting and financial performance, however, these examinations were biased towards financial services, healthcare and
4 Jones et al. (2007), Ullmann (1985) find correlations between financial performance and sustainability performance; Aupperle et al. (1985), Arlow and Gannon (1982) find no convincing result between financial performance and sustainability performance.
5 MSCI’s ESG research framework generates an analysis and rating of each company’s management of its environmental, social and governance performance. The following research is based on the components in KLD that are classified in to environment, social and governance.
Page 9 of 64
technology stocks, rather than industries which are difficult to track6 – such administrative and support activities. Therefore, this study will contribute to the literature on the topic by using more components (detangled ESG sub-components) and a more diversified data source7 in order to develop a holistic approach to the research question, and meanwhile, will be part of the very few studies that to provide evidence on the impact of sustainability reporting on financial performance in this different classification of industries.
From societal aspect, as it has become a trend for companies to disclose sustainability information, this topic will interest companies and stakeholders, especially those invested in heavy emission industries. It will not only give a new insight for companies that already issue sustainability reporting voluntarily, but also may give a benchmark for companies that are still holding back on disclosing sustainability information. To be precise, it will encourage more carbon emission-intensive companies to disclose sustainability information in case of either the findings show a significant positive or neutral/weak result between ESG and financial performance. With a positive correlation, these companies are likely to benefit in financial perspective from disclosing while with a neutral correlation, these companies are still likely to excel in legitimation without influencing financial performance.
1.4 Structure
Section 2 will discuss the literature review and the different results of prior studies. Section 3 will describe the research methodology, the hypotheses and the empirical tests to be executed. Section 4 will interpret the results of the empirical test, followed with section 5 to discuss the result and whether to accept or reject the hypotheses. The concluding remarks will appear in section 6.
2. Literature Review and Hypothesises
6 Wood & Jones (1995) investigates the relationship between CSR and stakeholders, the questions include who set expectations for CSR and who evaluate the outcome of CSR. The industries with closer relationships with customers, communities, or governments are easier to track with
7 The data source includes companies that are listed on S&P 500, Domini 400, Russell 1000 and Russell 2000. Page 10 of 64
This section presents an overview of involved theories and some prior studies, followed by hypotheses development for the purpose of this study.
2.1 System Oriented Theories
Basically, firms primarily issue standalone sustainability reports in response to external scrutiny by stakeholders. This is consistent with the stakeholder perspective, and the ancillary motivations8 for firms issuing standalone CSR reports. These motivations are consistent with legitimacy and signalling perspectives (Thorne et al., 2014). The underlying theories in this matter are system oriented theories. More specifically, they are legitimacy theory (Gray et al.,1995)9, stakeholder theory (Ulmann, 1985)10 and institutional theory(Fiedler and Deegan, 2002)11. There is a considerable amount of prior literatures (Gray et al., 1995; Ulmann, 1985; Fiedler and Deegan, 2002)12 with respect to system oriented theories. The main idea of system oriented theory is that the entity and the society interact. The entities influence and are influenced by the surrounding environment. Gray et al. (1996) define this theory as “it focuses on the role of information and disclosure in the relationships between organisations, individuals and the country”.
Legitimacy is 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). Legitimacy theories were
8 Thorne et al. (2014) finds that (1) signal to stakeholders that the company is interested in social responsibility, (2) CEO/Board Commitment, (3) to communicate to stakeholders that the company has a policy of corporate transparency, (4) ease of access of having all social responsibility information in one place, (5) enhance reputation by providing truthful and robust information on tough issues, (6) enhance reputation by impressing upon stakeholders that we are good corporate citizens, are the most important motivations for companies to issue standalone CSR reports.
9 As Gray et al. (1995) indicate, the most common theoretical perspectives in the social and environmental accounting are legitimacy theory and stakeholder theory.
10 Ulmann (1985) states: “our position is that organisations survive to the extent that they are effective. Their effectiveness derives from the management of demands, particularly the demands of interest groups upon which the organisation depends”.
11 Fiedler and Deegan (2002) identified a number of theoretical perspectives which had been used or could be used to explain the formation of environmental collaborations, including stakeholder theory, legitimacy theory, and institutional theory.
12Refer to footnote 9, 10 and 11.
Page 11 of 64
explained clearly in the research of Dowling and Pfeffer (1975) and Lindblom (1994). Dowling and Pfeffer (1975) describes how organizational legitimacy is determined by many factors like the method of operation and output, and the goals or domain of activity of the organization. There are three ways for companies to conform to legitimation. First, companies can accommodate their performance, objectives and approaches of running the business to conform to a social contract. Second, companies can alter the definition of social legitimacy and change societal expectations. Third, companies can become identified with symbols, values or institutions that have strong base of legitimacy (Dowling and Pfeffer, 1975). Lindblom (1994) carried out four strategies of legitimacy that include: educating and informing relevant publics about changes in performance, changing perceptions of relevant public about performance, manipulating perceptions by deflecting attention from the issue of concern, and changing external expectations of performance. Additionally, Suchman (1995) classified legitimacy into different levels: pragmatic, moral and cognitive legitimacy.
Stakeholder theory is a theory that emphasizes on the existence of different stakeholder groups and the interactions among them. Freeman (1984) contributes to the core idea of stakeholder theory: organizations that manage their stakeholder relationships effectively will survive longer and perform better than those organizations that don't. It has similarities with legitimacy theory as stakeholders and social contracts influence companies and are influenced by companies. Deegan & Unerman (2006) describe two branches of stakeholder theory, the managerial branch and the ethical branch. The managerial branch describes various groups of stakeholders that exist in society, and how the expectations of particular stakeholder groups may have more or less impact on corporate strategies. The ethical branch holds the perspective that both the primary and the secondary stakeholders have certain minimum rights that must not be violated. It can be extended to a notion that all stakeholders also have a right to be provided with information about how the organization impacts them.
Institutional theory provides a complementary perspective to both stakeholder theory and legitimacy theory in how organizations understand and respond to
changing social and institutional pressures and expectations (Deegan & Unerman, 2006). It explains why organisations tend to take on similar characteristics and form. In particular, it brings legitimacy into companies. It consists of two dimensions, isomorphism, which refers to ‘a constraining process that forces one unit in a population to resemble with other units that face the same set of environmental conditions (DiMaggio & Powell,1983)’; and decoupling, explains that due to shareholder pressure, companies that appear to be adhering to socially expected behaviour are actually departing from it (Deegan & Unerman, 2006).
Whilst corporations are more likely to disclose their sustainability information under the influence of system oriented theories, corporations also possibly benefit from releasing such information in the aspects of reputational benefits, cost savings identification, increased efficiency, enhance business development opportunities and enhanced staff moral (ACCA, 2013).
2.2 Sustainability Disclosure and Financial Performance
A number of studies (Jones et al., 2007; Ullmann, 1985; McWilliams and Siegel, 2000; McGuire et al., 1988; Aupperle et al., 1985; Arlow and Gannon, 1982; Lee et al., 2009; Galema et al., 2008) examine the relation between overall financial performance and sustainability disclosure. Results from prior research are, however, mixed. Jones et al. (2007) find a strong relation between sustainability disclosure and some financial performance indicators that the incidence of sustainability disclosure is positively associated with firms' levels of operating cash-flows to total assets, working capital to total assets, retained earnings to total assets, asset backing per share, debt servicing capacity and capital expenditure relative to assets. The result also suggests that factors other than financial performance may be contributing to the level of sustainability reporting. It also finds control variables, including general variables, accounting-based variables and stock-based variables such as firm size, a range of financial variables (e.g. earnings and cash-flows),stock return volatility and industry background, generally did not have a major effect on these results. However, the incidence of disclosure is also negatively associated with firms’ levels of cash resources to total asset and price to book value ratios. Ullmann (1985) also suggest CSR may be linked to past firm performance (accounting-based connection) and Page 13 of 64
likewise, its financial performance may influence a firm's evolving social policy and actions.
Opponents argue that there is no evidence of a positive association between CSR reporting and financial performance. McWilliams and Siegel (2000) also find there is a neutral effect of CSR on financial performance. McGuire et al (1988) find no significant relation between stock-market-based performance measures and CSR while significant correlation between accounting-based performance and CSR. Among accounting-based measures, ROA and total assets show positive relationships. Furthermore, the operating income growth has a negative correlation. Nonetheless, Aupperleet et al. (1985) conclude that it was impossible to support the notion of a relationship with either a positive correlation or a negative correlation between firm profitability and an orientation toward CSR. The results is consistent with earlier studies such as Arlow and Gannon (1982), as they could not be able to provide convincing outcome for an association between profitability and CSR to indicate if it is beneficial or harmful to organisations.
In contrast, other scholars investigated CSR and financial performance have argued for a negative association. Lee et al. (2009) argue that their result shown the market-based tests suggest a negative association between CSR reporting and financial performance, while the accounting tests indicate no association exists between them. Researchers (Friedman, 1970; Griffin& Mahon, 1997; Waddock & Graves, 1997; McWilliams & Siegel, 2000) who have suggested a negative relation between financial performance and sustainability investing have argued that high responsibility results in additional costs and therefore, lead to an economic disadvantage to high socially responsible firms comparing to less socially responsible firms. Thus, less profitable firms may be less willing to undertake socially responsible actions.
This contradictory result is explainable because different researchers have different assumptions and methodologies. Some researchers use accounting-based methods (e.g. Ullmann, 1985) while others use market-based methods or mostly, hybrid methods (e.g. Jones et al., 2007; Galema et al., 2008; McGuire et al., 1988). McGuire (1988) argues, on the one hand, market performance was usually measured Page 14 of 64
by risk-adjusted return, or alpha, and total return while market risk is measured by the standard deviation of total return and beta which focuses on a measure of systematic risk. On the other hand, accounting-based performance involves analysis of return on asset (ROA), average asset, sales growth, asset growth, and operating income growth, while the risk is measured by the ratio of debt to assets, operating leverage, and the standard deviation of operating income.
Furthermore, on the one hand, Galema et al (2008) suggest that the inconsistency between the empirical literature results and predictions rooted in theoretical models is due to the misinterpretation of the risk-adjusted performance measures. Two main arguments were brought forward in order to re-design the regression models. The first error is that financial performance is computed by controlling for systematic risk while the systematic risk captures part of the trade-off. The second error is in regards to dimensions of sustainability (e.g. community, product, etc.; will be explained further in section 3.1 and 3.2) as the aggregate measures may confound existing relationships sustainability investment and returns in different dimensions. The result establishes sustainability investment with portfolios score positive on diversity, environment and product has a significant impact on stock returns.
On the other hand, McGuire et al (1988) suggest accounting-based indexes are subject to managerial manipulation (Branch, 1983; Briloff, 1972, 1976) and tend to be stable over time (McGuire Schneeweis, & Hill, 1986) while stock-market returns are more variable over time since they primarily respond to unexpected changes in information. They discussed that some studies on stock market based measures of return lead to a mixed result because the researchers failed to adjust for risks. However, McGuire et al (1988) itself reported a weak relationship between CSR and stock market based financial performance.
McGuire et al. (1988) also discussed that accounting based financial performance measures have a higher likelihood of generating a positive result based on the results of past studies, especially for those studies that take into consideration controlling for differences in risks. The reason is because accounting based measures are more sensitive to CSR which is firm specific and unsystematic, while stock market based reflect systematic market trends and primarily respond to unexpected changes.
Additionally, McGuire et al (1988) investigated relation between CSR and risks in case of overlooking financial measures. For example, Spicer (1978) suggests that firms with a high ranking on corporate social responsibility have lower total and systematic risks. Aupperle et al (1985) found corporate social responsibility and accounting based risks have a negative correlation while CSR and stock market based risks have a neutral correlation. As it was mentioned earlier, the additional cost in sustainability lead to economic disadvantages for companies. It is contradicting to the argument from risk perspective, which would refute by demonstrating that rather than looking at profitability, a reduced financial risk would give firms a more stable return.
The discussion above leads to the following hypotheses:
H1: The relation between sustainability performance and financial performance is insignificant based on market-based measures.
H2: The relation between sustainability performance and financial performance is significant and positive based on accounting-based measures.
2.3 Carbon emission and Financial Performance
Hrasky (2012) identifies that carbon footprints broadly relate to carbon emissions, or greenhouse gas emissions expressed in carbon equivalents. Carbon footprints are measured in terms that allow assessment of their impact on global warming and thus contribution to climate change. Global warming raises a series of risks for companies, for example, regulatory risk, supply chain risk, product and technology risk, litigation risk, reputation-related risk, and physical risk (Lash and Wellington, 2007). These impacts cover direct risks and indirect risks from a higher emission relevant cost to drought and flood and corporate regulation and reputation threats (Lash and Wellington, 2007).
Disclosure is a method that companies can try to convince stakeholders that the existence of carbon emission reduction strategy and that the companies’ operations are legitimate (Hrasky, 2012). Sprengel and Busch (2011) identify stakeholders including government, investors, customers, NGOs, suppliers and competitors as the most relevant audiences in the context of climate change and carbon emissions. Moreover, Weinhofer and Hoffmann (2010) identify customers and shareholders as
particular stakeholder groups that are pressuring companies to reduce their carbon emissions. It indicates that a legitimation response by companies must be maintained with the aim of keeping their social contracts intact. If firms are acting in ways that are inconsistent with the values underpinning the contract with their stakeholders, they will be at risk of breaching the social contract and therefore facing financial consequences. (Gray et al., 1988; Deegan, 2002).
Bouteligier (2009) asserts that companies are integrating environmental values and sustainability practices into their corporate strategies as a result of the rising awareness of environmental issues. Wal-Mart CEO Lee Scott even said that a corporation which concentrates on reducing greenhouse emission as soon as possible has a good business strategy. Because it will save money for customers, and therefore not only makes for a more efficient business model, but also helps position a corporation to compete effectively in a carbon-constrained world. The importance of carbon emission is notable because it has a significant influence on corporations and the entire society.
Hrasky (2012) suggests some activities (e.g. heavy industry manufacturing, massive resource consumption manufacturing, etc.) were potentially more environmentally sensitive tended to have higher disclosure levels than those with less sensitive operations. Patten (1992) reports that, after the Exxon Valdez oil spill in Alaska in 1989, companies in relevant industries significantly increased the amount of environmental disclosure in their annual reports. Additionally, Hrasky (2012) finds out the behaviour in regards to footprint disclosure of more emission-intensive industries (e.g. automotive, energy) and less emission-intensive industries (e.g. financial services, hospitality) are different. More emission-intensive sectors are based on behaviour management approach13 while less intensive sectors’ disclosure are trend to be symbolic management strategy14. More specifically, behaviour disclosures disclose substantive actions taken to influence carbon footprints and reflect the actual
13 Disclosure decisions focus on disseminating information about the instrumental actions taken by a company to reduce its carbon footprint (Kim et al., 2007).
14 Disclosure decisions are rhetorical statements designed to create an impression of environmental responsibility, not necessarily accompanied by relevant action (Kim et al., 2007).
Page 17 of 64
emphasis of underlying operational activities; while symbolic disclosures are reporting on what appears to be substantive action but in fact has little substance in terms of the firm’s ongoing operations and impact (Hrasky, 2012; Kim et al. 2007). Kim et al. (2007) investigate the relative efficacy of symbolic versus behavioural management in reputation building in a sample of US corporations. Their results suggest that a behavioural management approach is more effective to manage perceptions and symbolic management is insufficient to promote informed decision-making by stakeholders and less sufficient to maintain legitimacy.
Hence, in terms of comparing more emission-intensive industries and less emission-intensive industries, I would set the next hypothesis based on the relation between financial return and two classifications of industries, as below.
H3: The relationship between sustainability performance and financial performance is stronger in more emission-intensive industries.
2.4 Environmental, Social and Governance (ESG) and Financial Performance Environmental, social and governance (ESG)15 is known as a “catch-all term” for the criteria in socially responsible investing. ESG refers to the three main areas of concern that are used as main factors in measuring a company’s impact on sustainability and ethics. There is a variety of different issues in relation to each sub-component of ESG, for example, environmental issues include biodiversity, climate change, water, and so on; social issues include human rights, child labour, social displacement, financial inclusion and many more; and governance issues include board structure, remuneration, bribery and corruption, disclosure, etc. (GRI, 2007). According to Dow Jones Sustainability Indexes, a list of criteria and proportion of weightings is used to assess the opportunities and risks in governance (economic), environmental and social developments for companies (Lo & Sheu, 2007).
Table 1: DJSGI corporate sustainability assessment criteria
15 The sustainability performance in the following study will be classified into these three components: the environment, the society, and the governance.
Page 18 of 64
In recent years, there is a trend showing that ESG components are incorporated into investment analysis in analysing financial and non-financial factors, such as real-estate, corporations and fixed-income investments and valuation of equity. The reason that there has been increasingly recognition of ESG is because it is a part of mandatory requirements of stock exchanges in some countries nowadays, such as UK, US and South Africa (ACCA, 2014). Hence, the development of sustainability reporting contains ESG issues. Companies face ESG opportunities and threats that may have a significantly influence on firm performance. More specifically, focus on impact of ESG, rather than sustainability reporting in general. Filbeck et al. (2009) finds companies that invest in ESG improve their financial performance, while Lee et al. (2009) finds companies that invest in ESG damage their financial performance. Despite of the debate on the relation between ESG and financial performance, several studies (discuss in the following paragraphs) demonstrate each of E, S, G components has a different impact on company performance.
The environmental component deals with the impact of companies’ business activities on living and non-living nature systems, including biodiversity loss,
greenhouse gas (GHG) emissions, climate change, renewable energy, energy efficiency, air, water or resource depletion or pollution, waste management, changes in land use, ocean acidification and changes to the nitrogen and phosphorus cycles (PRI, 2013). Derwall et al. (2005) find that based on Innovest Strategic Value Advisors16’ corporate eco-efficiency scores, the high ranked portfolios provided significantly higher average returns than lower ranked eco-efficiency portfolios. In contract, Galema et al. (2008) find an opposite result that socially responsible investing portfolio, including environment component, lowering the book-to-market ratio and not by generating positive alphas.
The social initiatives consider the impact of the organisation’s business on the social system in which it operates, including human rights, labour standards in the supply chain, child, slave and bonded labour, workplace health and safety, freedom of association and freedom of expression, human capital management and employee relations (PRI, 2013). Several authors have cited improved employee and customer goodwill as a significant result of CSR (McGuire et al., 1988). Herremans et al. (1993) find that comparing to organizations with poorer reputations, and organizations with better reputations perform better and have lower risk. However, a contrary point was brought out by Galema et al. (2008), as no correlation between employee relations and performance was found in their study.
The governance components impacts on its stakeholders on economic perspectives and on economic systems at different levels. It is a part of economic component, includes compliance and risk management. It includes issues relating to the governance of companies and other investee entities. For example, the compliance covers board structure, size, diversity, skills and independence, executive pay, shareholder rights, stakeholder interaction, disclosure of information, business ethics, bribery and corruption. The risk management covers internal controls and general risk management (PRI, 2013). Gompers et al. (2003) find that the risk-adjusted returns of firms with strong governance outperform firms with poor governance. However, the
16 A firm The firm focuses on analysing the companies' performance on environmental, social, and strategic governance issues.
Page 20 of 64
study of Core et al. (2006) indicate that there is no evidence of a significant relation between the governance and financial performance.
Despite of the existing debates as aforementioned, a research conducted by Humphrey et al. (2012) on the independent effects of the E, S, G initiatives find that choosing stocks with good E, S or G ratings does not result in lower returns nor does it appear to breach one’s fiduciary duty to maximize profits. However, no further research on the relation between a specific component and a specific industry is conducted.
Based on the mixed arguments, I intend to test the integration of the ESG and the levels of carbon emission, more specifically, the impact of ESG components in sustainability reporting comparing more emission-intensive industries and less emission-intensive industries. The sustainability performance will be classified into E, S and G components in accordance with KLD guidance (it will be explained further in the section 3.2). The correlation between each financial performance variables and each dimension of three different components will be examined separately.
Therefore, the last hypothesis is as follow:
H4: The correlation between more emission-intensive industries and E is closer than the correlation with S and G, compared to less emission-intensive industries.
3. Research design
The study uses fully reproducible quantitative methods to examine the relation between sustainability performance and financial performance.
3.1 Data and Sample Selection
Information for corporate social performance will be extracted from the KLD database (Kinder, Lydenberg, Domini). KLD Research &Analytics, Inc. is the leading authority on social research for institutional investors and uses a combination of sources including surveys, financial statements, academic journals and press and government reports to assess social performance. This very same data source has been used in Kim et al (2012), Galema et al (2008), Waddock & Graves (1997), etc. for Page 21 of 64
conducting similar research. The data covers thirteen dimensions (26 sub-dimensions) of 164 variables such as community, corporate governance, diversity, employee relations, environment, human rights, product, alcohol, firearms, gambling, military, nuclear power, and tobacco. The results consist of strengths (exemplary social performance) and concerns (poor social performance) where each strength or concern is rated on a binary summary from 0 (no strength or concern) to 1 (significant strength or significant weakness). In the 1990s, KLD covered the S & P 500 Index and the Domini 400 Social Index17 (refer to as “DSI 400”); in 2001 the database was extended to include all constituents of the Russell 1000; in 2002 the database was extended to include the large cap social index; in 2003 the database was further extended to include all stocks from the Russell 2000 and the broad market social index as well (Table 2 as below presents the expansion of database). KLD does not have historical ratings data for non-US companies, unless these are members of the S & P 500. At this point in time, KLD database contains information from 1991 to 2013 with 40,518 firm-year observations.
However, as it was mentioned as above, KLD includes dimensions such as alcohol, gambling, firearms, military, nuclear power and tobacco. These dimensions shall be excluded from this study for three reasons. Firstly, they focus only on concerns and do not contain strength data. Since the ratings are all in binary variables (0 or 1) for both of concerns and strengths. With dimensions only including concerns this will certainly lead to biased results. Secondly, Hong and Kacperczyk (2007) found portfolios with negative ethical issues have greater book-to-market values and greater excess returns compared to normal stocks18. Evidence from corporate financing decisions19 and the performance of ‘sin stocks’ suggest that norms affect stock prices and returns of companies in certain industries such as alcohol and tobacco (Hong and Kacperczyk, 2007). Additionally, DSI 400 only lists firms that derive less 17 The Domini 400 Social Index (DSI 400) is also refer to as the iShares MSCI KLD 400 Social ETF. It seeks to track the investment results of an index composed of U.S. companies that have positive environmental, social and governance characteristics as identified by the index provider. (Retrieved from https://www.ishares.com/us/products/239667/ishares-msci-kld-400-social-etf)
18 The reason is that those sin stocks being neglected by norm-constrained investors and facing greater litigation risk heightened by social norms (Hong and Kacperczyk, 2007).
19 More detail on the corporate financial perspective can be found in Hong and Kacperczyk (2007).
Page 22 of 64
than 2% of their gross revenue from the production of military weapons, have no involvement in nuclear power, gambling, tobacco, and alcohol, and have a positive record in each of the remaining categories (McWilliams & Siegel, 2000). Again, containing companies from other database/index (S&P 500, Russell 1000, and Russell 2000) which only fall in “sin” dimensions may distort the result. These aforementioned reasons has led to the decision that this research will only cover seven dimensions out of the thirteen available in the KLD database, these are: community, corporate governance, diversity, employee relations, environment, human rights, and product.
Table 2: KLD data coverage history
Source: Getting Started: An Introduction to KLD STATS and KLD’s ratings definitions
Information for financial performance will be extracted from the COMPUSTAT database. COMPUSTAT contains daily, monthly, and quarterly updates from companies operating in the world split into four segments: North America, Global, Banking and Historical. In this thesis, COMPUSTAT North America monthly data will be used because all the companies with historical ratings from KLD are US companies. The COMPUSTAT North America monthly updates database contains the following sections: fundamentals annual, fundamentals quarterly, index fundamentals, index prices, industry specific, ratings and a lot more. Among fundamental annual, variables that are standardized to accommodate the wide variety of financial accounting practices across industries (the industry sector information is presented in Table 3) can be obtained from balance sheet, income statements, cash flow, miscellaneous items and other supplemental data.
To create a connection between KLD and COMPUSTAT data that are related, CUSIP20codes are used – doing this makes data only available after 1995 when CUSIP codes were introduced. KLD covers 37,925 firm-year observations in total from 1995 to 2013. However, according to KLD coverage history (refer to Table 2), as of 2003, KLD provides a summary of strengths and concerns for approximately 3100 companies listed on the S&P 500, Domini 400 Social Index, Russell 1000, or KLD Large Cap Social Indexes as of December 31st of each year. Therefore, in order to keep consistencies of social rating data, the time range of research is reduced from 1995-2013 to 2003-2013, which is a period of 11 years. Furthermore, observations which are not available continuously throughout the years will be excluded. For example, companies that lack of one or more years of data will be excluded. In order to ensure the completeness of the data, a filter was set up to exclude companies with incomplete information, and only 653 companies contain complete data throughout all 11 years. The data from these 653 companies consists of 7,205 firm-year observations with 14 variables based on 7 dimensions (additional detail is presented in Table 3). All of this data will be used as the sample.
3.2 Model
The following model will be used in this research:
Sustain_perf= β0 + β1 Financial_perf + Controls + ɛ
In order to examine the relation between sustainability performance and financial performance the above regression model is used. Proxies for sustainability performance are the three corporate social responsibility ratings (Environment, social, governance). For the financial performance both accounting measures and stock market measures are used. The control variables are traditional accounting and stock market risk. In next few paragraphs, the variables will explained more in depth.
20 A CUSIP is a nine-character alphanumeric code that identifies a North American financial security for the purposes of facilitating clearing and settlement of trades. (Retrieved from https://www.cusip.com/cusip/about-cgs.htm)
Page 24 of 64
3.3 ESG Classification
Based on the description in regards to the concept of ESG on KLD Research &Analytics, Inc. official website, the detangled ESG components can be classified in the way as they are presented as follow.
Figure 1: ESG classification according to KLD Research &Analytics, Inc.
With the intention of investigating the association between ESG components and financial performance, only the total number of strengths and total number of concerns are extracted from KLD database for the use of this study. There are many dimensions in each dimension, for example, community includes 12 sub-dimensions (Charitable giving, innovative giving, support for housing, support for education, non-US charitable giving, volunteer programs, community engagement, other strengths, investment controversies, community impact, tax disputes and other concerns). Because one of the aims of this study is to find out the correlation of the detangled E, S and G components and financial performance. There is no point to analyse the correlation between each of the sub-dimension and financial performance measures. By referring to the above structure chart for ESG classification, a customized classification is made for the purpose of this study, as shown in Table 3. Figure 1 shows a mix of dimensions and sub-dimensions, while table 2 is composed out of solely dimensions. Any dimensions that are shown in Figure 1 that are not Page 25 of 64
mentioned in Table 3 are sub-dimensions – for example these listed under “Environment ratings” and “Governance ratings”.
Table 3: ESG Classification for 7 dimensions
3.4 Industry Classification
To answer the posed research question, whether there is a significant difference in the relation between sustainability reporting and financial performance between more and less emission-intensive industries, all companies within the sample must be categorized in one of these two categories. According to Hrasky (2011), more emission-intensive industries include materials, industrials, energy, and utilities; less emission-intensive industries include financials, consumer discretionary, consumer staples, healthcare and telecommunication. Asselt and Biermann (2007) also identified energy-intensive industries in their paper – the emissions trading directly covers large, energy-intensive industries, including the steel, glass, pulp and paper and cement industries, as well as large combustion installations, including power generators. Environment Ratings •Environment Concerns •Environment Strengths Social Ratings •Community Concerns •Community Strengths •Diversity Concerns •Diversity Strengths •Employee Relations Concerns •Employee Relations Strengths
•Human Rights Concerns •Human Rights Strengths •Product Concerns •Product Strengths Governance Ratings •Corporate Governance Concerns •Corporate Governance Strengths Page 26 of 64
This study will use SIC21 (Standard industrial classification) code found in COMPUSTAT, to classify industries in alignment with the way Hrasky (2011) and Asselt and Biermann (2007) classified industries. The summary of the numbers of companies for each sector is as follow in Table 4.
Table 4: Overview of sample industries distribution
Industry SIC N
More Emission-intensive industries (491 firms)
Mining, Construction 100-1999 58
Food, textile, apparel 2000-2390 24 Forest, products, paper,
publishing
2391-2780 28
Chemicals, pharmaceuticals
2781-2890 59
Refining, rubber, Plastic 2891-3199 19 Containers, steel, heavy
manufactures 3200-3569 65 Computers, autos, aerospace 3570-3990 149 Transportation 3991-4731 20 Utilities 4732-4991 69
Less Emission-intensive industries (161 firms)
Wholesale, retail 4992-5999 54
Bank, financial services 6000-6999 7
Hotel, entertainment 7000-7199 1
Other Service 7200-9989 97
None-classifiable establishments
9990-9999 2
21 Standard Industrial Classification (SIC) codes are four digit numerical codes assigned by the U.S. government to business establishments to identify the primary business of the establishment.
Page 27 of 64
3.5 Measures and Control Variables
Data will first be collected from the aforementioned databases, and followed by an exclusive analysis on return on asset (refer to as ROA), average asset, operating income growth, asset growth, and sales growth, debt to asset, market return, alpha and beta. Since the aim of this study is to examine the impact on financial performance of companies engaging in issuing sustainability information, comparing more carbon-intensive industries and less-carbon-intensive industries, the two classifications of industries will be evaluated separately to obtain a result. In the analysis part, the companies’ activities will be classified into three sections: E, S and G.
Programming software R, the R Project for Statistical Computing will be used to run regressions. Similar to Stata and SPSS, R is a free software environment for statistical computing and graphics. R provides a wide variety of statistical (linear and nonlinear modelling, classical statistical tests, time-series analysis, classification, and so on) and graphical techniques, and is highly extensible22.
There are debates going on for years on whether accounting-based measure of performance or stock-based measure of performance is more reliable for conducting research on the relationship between CSR and financial performance. Accounting-based measures consider historical data and are influenced by different accounting procedures and different levels of management manipulation (McGuire et al., 1988). Although stock-based measures by using market return is less susceptible to the shortcomings of accounting-based measures, concentration on firm valuation may not be sufficient and it is based on too many assumptions (McGuire et al., 1988). However, to offset the drawbacks of these two measures, this study will use both accounting-based measure and stock-based measure to generate a result. Based on Beaver et al. (1970), the stock market based measures determine the equilibrium prices for all securities, and the accounting based measures that capture most of the important relationships include dividend pay-out, growth, leverage, liquidity, asset size, variability of earnings, and covariability of earnings.
22More information can be found on R website.
Page 28 of 64
- For stock market returns, to follow the method used by McGuire et al (1998), total return and alpha (risk adjusted measurement) will be used as main measures.
- For accounting-based measures, McGuire (1988), Aupperleet al (1985) pay particular attention to ROA and firm profitability. Therefore, in this study, ROA, average asset, operating income growths, asset growth, and sales growth will be taken in to consideration as independent accounting based financial performance measures.
Additionally, previous study focusing on firm profitability and marketed based stock return has overlooked (McGuire et al. 1988). In case of overlooking returns, risk measures are also included in the analysis, which consists of market return based risks and accounting based risks. The correlation between these risk measures and sustainability factors will also be examined.
- For stock market based risk measures, Amihud and Lev (1981) assumed that a two-index market model was the appropriate return-generating process for bank stock returns, it includes total return risk, market return risk (beta), non-systematic risk and interest rate risk. However, to be consistent with McGuire et al. (1988), only beta and standard deviation of total return are used.
- For accounting based risk measures, Dechow et al. (2013) suggest that percentage of total assets that were financed by creditors predict significant corporate events. According to Modigliani and Miller (1958), leverage can easily shows that the earning stream of the ordinary stockholders becomes more volatile with debt. This is consistent with the variable McGuire et al. (1988) selected. Therefore, debt to asset, operating leverage, and in addition, standard deviation of operating income are used.
3.5.1 Stock Market based Approach Control variables
a) Total return measures the actual return of the business and takes into account economic values such as interest, capital gains, dividends and distributions realized over a given period of time (McGuire et al., 1988). However, the information extracted from COMPUSTAT does not contain this data for every company over all studied years. Since this measurement will be closely related Page 29 of 64
to stock market, the total return rate will be calculated as the movements in stock market valuation, measured as follow (McGuire et al., 1988):
Total Return = [(Pricet+ Dividendt) − Pricet−1]/Pricet−1
b) Alpha measures the performance on a risk-adjusted basis. It takes the price risk and compares its risk adjusted performance to a benchmark index. The measurement is derived from the market model as following (McGuire et al., 1988; Beaver et al., 1970):
Ri = a�i+ b�iRm Where
Ri stands for return on company i in period t;
Rm stands for return on Standard and Poor’s 500 industries index in period t; Beta (β) stands for Cov (Ri,Rm)
Var(Rm) ; And alpha (a) is Ri - bRm.
c) In regards to the stock based risk measures, beta (β) (refer to (b) for the formula) is the covariance of return on a company and return on the stock market to variance of return on stock market (in this study, the data is obtained directly from the CRSP23 database by using corresponding CUSIP codes). d) Standard deviation of total return measures the variability of total return
measured as following (McGuire et al., 1988):
�(Total return − Total return����������������)2/n n
i=1
3.5.2 Accounting based Approach Control variables
23 CRSP is renowned for its expertise in building and maintaining historical, academic research-quality stock market databases. More information can be found on CRSP website.
Page 30 of 64
To align with McGuire et al (1988), the predictors to be used for analysis are listed as following respectively.
a) ROA is measured as (net income + interest expenses)/average total assets. However, since the interest expenses is only available for a few companies in a few years, to minimize the influence of data missing, ROA is calculated as net income / average total assets.
b) Average asset is measured by using ending assets minus beginning assets and then divided by 2.
c) Asset growth is measured by calculating the movement percentage from total asset of year t-1 to total asset of year t.
d) Operating income growth is measured by calculating the change from operating income before depreciation of year t-1 to operating income before depreciation of year t.
e) Sales growth is measured by calculating the movement percentage from sales of year t-1 to sales of year t.
f) In regards to the accounting based risk measures, standard deviation of operating income is defined by the equation as follow (McGuire et al., 1988):
∑ (Operating income − Operatıng ıncomen �����������������������)2/n
i=1 .
g) Debt to asset is total debt divided by total asset.
h) Operating leverage is defined as:
(sales − variable costs)/(sales − variable costs − fixed costs) Where
Fixed expenses = maintenance expense + advertising expense + staff expense + pension and retirement expense + rental expense + selling, general and administrative expenses
Variable expenses = total (operating) expense - fixed expense (Because variable expenses vary from changes of business level and business activities, it is very likely to omit some variable expenses if it is calculated by adding up expense items)
3.6 Correlation analysis and Regression analysis
To align with McGuire et al (1988), two statistic based analyses will be conducted to find the association between sustainability performance and financial performance. McGuire et al. (1989) do not provide detail description about these two models in their study. However, correlation analysis and regression analysis are two commonly used statistics models, some statistical implications will be explained in the following paragraph.
The first one is correlation analysis, which is used to find the strength of linear association between two numerical variables. The results are always between -1 and 1 where 1 indicates a strong positive relationship, -1 indicates a strong negative relationship, and 0 indicates no relationship at all. For example, a correlation of r = 0.6 suggests a strong positive correlation between two variables, whereas a correlation of r = -0.06 suggests a weak negative correlation. The general model for correlation analysis is as follow (Archdeacon, 1994):
𝑟𝑟 = 𝑛𝑛 ∑ 𝑥𝑥𝑥𝑥 − (∑ 𝑥𝑥)(∑ 𝑥𝑥)
�𝑛𝑛(∑ 𝑥𝑥2) − (∑ 𝑥𝑥)2�𝑛𝑛(∑ 𝑥𝑥2) − (∑ 𝑥𝑥)2 Where x = variable x, y = variable y, n = number of sample.
The graphs as below in Figure 2 depict distribution of data for four different scenarios.
Figure 2: Correlation distribution scenarios
Source: Applied Multiple Regression/Correlation Analysis for the Behavioural Sciences, 3rd edition
Therefore, to apply the correlation model in this case, the model will be:
𝑟𝑟 = 𝑛𝑛 ∑ 𝐸𝐸𝐸𝐸𝐸𝐸 × 𝐹𝐹𝐹𝐹𝐹𝐹 − (∑ 𝐸𝐸𝐸𝐸𝐸𝐸)(∑ 𝐹𝐹𝐹𝐹𝐹𝐹)
�𝑛𝑛(∑ 𝐸𝐸𝐸𝐸𝐸𝐸2) − (∑ 𝐸𝐸𝐸𝐸𝐸𝐸)2�𝑛𝑛(∑ 𝐹𝐹𝐹𝐹𝐹𝐹2) − (∑ 𝐹𝐹𝐹𝐹𝐹𝐹)2 Where
N = number of companies
ESG = ESG variable (e.g. environment number of strengths)
FIN = Financial performance variable (e.g. ROA)
The second analysis is regression analysis, which is used to test if the model used is valid. Regression analysis is one of the most widely used techniques for analysing multifactor data (Montgomery et al., 2012). Two sub-analyses are included in this section with one is called F-test (also known as ANOVA) and the other one is R analysis (R squared and adjusted R squared). The F test is used to test if a group of variables are jointly significant and R compares the descriptive power of regression models. R-squared values range from 0 to 100. A higher R-squared means that the Page 33 of 64
movements of a variable are completely explained by movements in the index. The models for F-test, R squared and adjusted R squared are presented as follow (Gravetter & Wallnau, 2009; Archdeacon, 1994).
F-test: F = 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 (𝑑𝑑𝑣𝑣𝑑𝑑𝑑𝑑𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑑𝑑) 𝑏𝑏𝑣𝑣𝑏𝑏𝑏𝑏𝑣𝑣𝑣𝑣𝑣𝑣 𝑑𝑑𝑣𝑣𝑠𝑠𝑠𝑠𝑠𝑠𝑣𝑣 𝑠𝑠𝑣𝑣𝑣𝑣𝑣𝑣𝑑𝑑 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 (𝑑𝑑𝑣𝑣𝑑𝑑𝑑𝑑𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑑𝑑) 𝑣𝑣𝑒𝑒𝑠𝑠𝑣𝑣𝑣𝑣𝑏𝑏𝑣𝑣𝑑𝑑 𝑏𝑏𝑏𝑏 𝑣𝑣ℎ𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 (𝑣𝑣𝑣𝑣𝑣𝑣𝑒𝑒𝑣𝑣) R squared: 𝑅𝑅2 = (𝑣𝑣 ∑ 𝑒𝑒𝑏𝑏−(∑ 𝑒𝑒)(∑ 𝑏𝑏))2 [𝑣𝑣(∑ 𝑒𝑒2)−(∑ 𝑒𝑒)2][𝑣𝑣(∑ 𝑏𝑏2)−(∑ 𝑏𝑏)2] Adjusted R squared: 𝑅𝑅� = 1 − (1 − 𝑅𝑅2) 𝑣𝑣−1 𝑣𝑣−𝑠𝑠−1
Where R2 = sample R squared, p = number of predictors, n = total sample size. Similarly, ESG and financial variable will be used in the models as above to run the regression test.
3.7 Validity and Reliability
Finally, since the method is largely based on research by McGuire et al. (1988), in which a different CSR data source24 is used back in 1980s, special attention will be paid to the validity of latest data. Hence, the following issues have been considered:
(1) Is the data up to dated?
(2) Is the data complete in all the fiscal years the research covers?
(3) Are the fiscal year periods for the companies in KLD social ratings and COMPUSTAT financial information the same?
(4) Is the data of social ratings from KLD suitable for correlation analysis and regression analysis?
Only the observations that meet the mandatory conditions (as above) like having a valid date/period for the disclosure, being trustworthy and being able to be verified will be applied in the findings in the next section in thesis. Therefore, when selecting the year to research, the latest available date was selected before deciding the starting year and the latest data which could be obtained meeting above criteria is 2013. When 24 Their CSR data source is Fortune magazine rankings.
Page 34 of 64
selecting the sample companies for this study only companies with complete data in all 11 years were retained. All selected companies have the same fiscal year period. In regards to concern (4), correlation analysis is a suitable model for analysing the correlation between two sets of variables from theoretical perspective (refer to section 3.5). And regression analysis is appropriate to examine the multivariate relationship between financial performance and corporate social responsibility (refer to section 3.5 for theoretical implications). However, before making more in-depth analysis in ESG and types of industries, a similar to McGuire et al. (1988) correlation analysis will be conducted based on all 653 firms in order to verify if the method is useful and feasible in this case. A detailed comparison between the result of this study comprise of concurrent data of 653 firms across a time span of 11 years from 2003 to 2013 (refer to Table 5) and the result of McGuire et al (1988) of data with 131 firms across a time span of 9 years from 1977 to 1985 (refer to Appendix 2) will be presented.
4. Findings
Two sets of data will be presented in this section. First the correlation analysis (refer to section 3.5 for the used model) between ESG and financial measures is presented, followed by regression analysis (refer to section 3.5 for the model explanation). McGuire et al. (1988) suggest that to consider financial performance as a variable that have an impact on sustainability than the reverse is more productive. Therefore, the following findings will start from the angle of financial performance.
4.1 Correlation Analysis 4.1.1 Full sample analysis
Table 5 (pg. 40) displays the result of the performed correlation analysis (refer to section 3.5 for more detail) across the entire dataset. The correlations are shown for all industries between CSR ratings with 7 different dimensions against accounting and market based financial performance measures. The total sample contains 653 companies, takes over a time range of 11 years (2003 – 2013).
Market based Financial Measures
Market-based financial performance measures including Alpha and total return both show insignificant correlations with ESG dimensions, suggesting that there is little association between them. The correlation between Alpha and social number of
