Beyond the Green Premium: Assessing the Cost-Benefit Threshold of ESG Scores in Europe Daniel Björknert and Linus Kidmark Master’s thesis in Accounting and Financial Management Supervisor: Aineas Mallios Spring Semester 2024 Graduate School, School of Business, Economics and Law, University of Gothenburg, Sweden Abstract This study examines the relationship between ESG scores and the cost of equity for European firms, focusing on identifying a potential ESG score threshold. The results of our study confirm our first hypothesis that ESG score has a significant impact on the cost of equity. The results also validate the existence of an ESG score threshold, where benefits in terms of reduced cost of equity diminish beyond a certain point. The threshold, found within the mean range of ESG scores in our sample, suggests that investors may not fully integrate ESG considerations into their investment decisions. This may indicate a misalignment that challenges the effectiveness of the current integration of non-financial metrics into financial decision making by investors. Our study contributes to the ongoing discourse on sustainable finance by challenging the prevailing assumption that the relationship between ESG scores and cost of equity is linear. Keywords: ESG, ESG scores, Cost of Equity, CoE, Cost of Capital, Threshold effect, European companies, sustainable finance, CSRD, NFRD. Acknowledgements We extend our profound thanks to our supervisor, Aineas Mallios, for his invaluable guidance, support, and constructive feedback throughout the process of our master thesis. His readiness to help, deep expertise, and constant encouragement have been crucial in shaping our thesis. Additionally, we are grateful to our fellow students from the seminar sessions for their insightful contributions to our thesis. Their thoughtful critiques and varied viewpoints have deepened our understanding of the subject and pushed us to engage both critically and creatively with our work. Thank you! Daniel Björknert and Linus Kidmark Gothenburg, May 2024 Table of Contents 1. Introduction ........................................................................................................................ 1 1.1 Background ...................................................................................................................... 1 1.2 Problem discussion ........................................................................................................... 2 1.3 Purpose of the study and research question. .................................................................... 3 2. Literature review ................................................................................................................. 4 2.1 Theoretical framework ..................................................................................................... 4 2.1.1 Legitimacy theory ...................................................................................................... 4 2.1.2 Stakeholder theory and shareholder theory ............................................................... 5 2.2 Sustainable investments ................................................................................................... 5 2.3 ESG and cost of capital .................................................................................................... 6 2.4 Hypothesis development ................................................................................................ 11 3. Method .............................................................................................................................. 13 3.1 Methodology .................................................................................................................. 13 3.2 Data collection................................................................................................................ 13 3.3 Descriptive statistics ....................................................................................................... 15 3.4 Regression variables ....................................................................................................... 16 3.4.1 Dependent variables ................................................................................................ 16 3.4.2 Independent variables .............................................................................................. 16 3.4.3 Control variables...................................................................................................... 17 3.4.4 Empirical identification ........................................................................................... 18 4. Results and Robustness Checks ........................................................................................ 19 4.1 Results ............................................................................................................................ 19 4.2 Robustness Checks ......................................................................................................... 21 5. Analysis and Discussion ................................................................................................... 22 6. Conclusion & Final discussion ......................................................................................... 24 6.1 Conclusion ...................................................................................................................... 24 6.2 Limitations of the study and suggestion for further research ......................................... 25 References ................................................................................................................................ 26 Appendix .................................................................................................................................. 30 Appendix 1. Test for multicollinearity ................................................................................. 30 Appendix 2. Robust and clustered standard errors ............................................................... 31 1. Introduction In this chapter we introduce the topic and context for our study. We explore the challenges and gaps in existing literature that have shaped the direction of our study. The problem discussion forms the basis upon which we formulate our research question. 1.1 Background The Paris Agreement, adopted in 2015 and entered into force in 2016, is a global commitment to combat climate change and to accelerate and intensify the actions and investments needed for a sustainable low-carbon future. Its main goal is to limit global warming to well below 2 degrees Celsius above pre-industrial levels and to pursue efforts to limit the temperature increase even further to 1.5 degrees Celsius (UNFCCC 2024). The focus on limiting global warming has highlighted the financial risks associated with climate change, driving investors to incorporate ESG criteria into their decision-making processes. Companies that demonstrate high ESG standards are increasingly seen as better equipped to mitigate climate-related risks and to capitalize on opportunities presented by the transition to a low-carbon economy. This recognition has led to a significant rise in the demand for sustainable investment options, prompting a shift in how financial institutions evaluate the environmental and social impacts of their investments. Historically, the field of financial accounting faced criticism for its narrow focus on financial performance, often at the expense of environmental and social considerations. This gap in reporting and accountability led to the development of frameworks aimed at integrating non- financial factors into corporate disclosure. The European Union has been at the forefront for this movement, implementing the Non-Financial Reporting Directive (NFRD) in 2014, followed by the more comprehensive Corporate Sustainability Reporting Directive (CSRD) (European Commission, nd). The NFRD required large corporations to report on ESG matters, thereby laying the groundwork for more transparent and responsible business practices. Building on this foundation, the CSRD expanded the scope to encompass a wider range of companies and introduced more stringent reporting standards, aiming to uniform sustainability reporting across the EU. (European Commission, nd) 1 While the regulatory frameworks signify an effort to align corporate activities with sustainable development goals, the heterogeneity in reporting standards poses challenges for investors and stakeholders in assessing and comparing the ESG performance of companies (Madison and Schiehll 2021). This has paved the way for ESG ratings as tools for evaluating a company’s sustainability efforts and its potential risks and opportunities from an ESG perspective. Rating agencies have a central role in this area as they collect a vast array of data from corporate disclosures, regulatory filings, and alternative sources, including insight from governmental and non-governmental organizations (ibid). These ratings are often calculated on a scale from 0 to 100 across environmental, social and governance dimensions and then aggregated into an overall score with the purpose of enabling more informed assessments by investors and other stakeholders (MSCI, 2023). Existing research (Velte 2017; Ng and Rezaee 2015; El Ghoul et al. 2011; MSCI, 2020) suggests a negative correlation between high ESG scores and lower cost of capital across both developed and emerging markets. MSCI’s findings highlight that companies within the highest ESG quantile experienced, on average, a cost of capital significantly lower than those in the lowest quantile, illustrating the financial benefits of superior ESG practices. 1.2 Problem discussion In the wake of the Paris Agreement and subsequent global commitments to a sustainable low- carbon future, investors and regulators have increasingly emphasized the integration of ESG criteria into financial decision-making. High ESG scores are often viewed as indicators of a company’s resilience against climate-related risks and its readiness to capitalize on opportunities within the transitioning economy. This has not only shifted investment paradigms but also transformed how financial performance is assessed, moving beyond traditional financial metrics to include sustainability metrics. The European Union’s efforts through directives such as the NFRD and CSRD have been important tools in promoting transparency and accountability in corporate ESG practices. These regulatory frameworks aim to standardize reporting and enhance comparability across companies, fostering a more uniform approach to evaluating corporate sustainability. Despite these advancements, the pursuit of high ESG scores raises complex questions about the economic viability of such investments. While there is a general consensus that superior 2 ESG scores may correlate with reduced capital costs due to perceived lower risks (El Ghoul et al., 2011; Ng and Rezaee, 2015), this relationship is not always straightforward (Magnanelli and Izzo, 2017). The costs associated with achieving and maintaining high ESG scores can be substantial, and the financial benefits are not always proportional, suggesting a non-linear relationship between ESG performance and cost implications. While ESG scores offer a standardized method to evaluate corporate sustainability, the actual effectiveness of these scores in accurately reflecting a company’s financial health and risk management capacity is not well-documented. Particularly, research by Bae, Chang and Yi (2018) and Ye and Zhang (2011) suggests that the benefits of high ESG practices are not linear, and there may be a threshold beyond which further investments do not yield sufficient returns and may even increase financial costs. Specifically, there is a lack of understanding regarding the point at which the costs of improving ESG scores outweigh the financial benefits, particularly in terms of capital costs. This gap highlights an underexplored area: the potential for overinvestment in ESG practices where the marginal cost of achieving higher ESG scores may not be justified by the corresponding reduction in capital costs. 1.3 Purpose of the study and research question. The principal aim of this study is to empirically assess the impact of varying ESG scores on cost of capital among European companies, particularly focusing on cost of capital as influenced by different levels of ESG performance. We aim to identify and analyze the potential financial inefficiencies that arise from high ESG scores and to discern whether there exists a non-linear relationship between ESG scores and financial outcomes. Central to this is the hypothesis that there exists a threshold in ESG scores beyond which the incremental costs of improvement may not be justified by the benefits of reduced capital costs. Based on the problem discussion and stated purpose of the study, the following research question has been formulated: “Is there a threshold beyond which a higher ESG score outweigh the financial benefits in terms of capital costs?” 3 2. Literature review In this chapter, we discuss the theories underpinning our study specifically focusing on Legitimacy Theory alongside Stakeholder and Shareholder Theories. Additionally, we review the existing body of research relevant to the impact of ESG scores on the cost of capital. Lastly, we provide a hypothesis development. 2.1 Theoretical framework 2.1.1 Legitimacy theory Legitimacy theory, which has significantly influenced various theoretical frameworks, focuses on understanding organizational behavior and structure. Meyer and Rowan (1977) highlight that organizations operate within a framework of institutionalized rules shaped by societal norms, values, and expectations. These rules can be explicit and enforced or implicit, emerging from social interactions. Organizations conform to these institutional structures to demonstrate their legitimacy and to shield themselves from scrutiny. This concept of institutional isomorphism - mimicking the structures of the institutional environment - seen as crucial for an organization’s success and survival. It potentially increases commitment from both internal and external stakeholders who perceive these structures as legitimate and provides protection against questioning. However, this approach leads organizations to seek legitimacy through conformity to societal expectations, rather than solely based on actual performance or outcomes. An unintended consequence of strictly following formal structures is the loss of effective internal coordination and control. Hence, the solution to this was decoupling or separating formal rules from the organization’s actual routines, to protect its core functions. (Meyer and Rowan, 1977) Furthermore, Deephouse and Carter (2005) differentiate between legitimacy and reputation. Legitimacy pertains to an organization’s social acceptance that justifies its existence, while reputation involves relative comparisons among organizations. Their research suggests that isomorphism is critical for legitimacy and that financial performance benefits lower- performing firms, with diminishing returns for high performers. 4 2.1.2 Stakeholder theory and shareholder theory Stakeholder theory is trying to explain that a firm’s management has a responsibility to create value for all stakeholders. This theory extends beyond shareholders to encompass a broader range of entities such as customers, suppliers, employees, and creditors. The theory emphasizes the interdependence of these stakeholders, arguing that value cannot be created in isolation. Decisions made by the firm should consider the impact on these various relationships. Neglecting this interrelation can lead to deep mistrust among stakeholders, potentially harming the firm’s reputation and profitability. Therefore, it is crucial for management to act in the interests of all stakeholders, maintaining and nurturing these essential relationships. (Freeman et al. 2010) Conversely, shareholder theory, often associated with Milton Freidman, describes that the foremost duty of a corporation is to its shareholders. This theory emphasizes that the primary aim of a business should be to enhance shareholder wealth and profitability, as shareholders are the owners of the company (Schaefer, 2008; Tse 2011; Friedman, 2007). Furthermore, according to Freidman (2007), the only “social responsibility’’ of the corporate executives is to maximize the corporation’s profits. Friedman (2007) further highlights the significance of generating value for shareholders, who bears all the risk associated with the firm and thus should be rewarded. As a result, allocating resources to other areas (like activities related to ESG) may not align entirely with the shareholders’ best interests. 2.2 Sustainable investments The domain of sustainable investments and the integration of sustainable practices in corporate strategies has obtained substantial scholarly attention recently. Despite a general consensus within existing literature, notable divergences exist. These differences primarily arise from the researchers’ focus on varied facets of capital costs and the adoption of diverse methodological approaches in data analysis. Previous studies have demonstrated a link between companies with strong CSR performance or high ESG scores and a reduced cost of capital. Yet, the exact reasons behind this relationship, such as whether it is due to a decreased cost of equity or reduced cost of debt is ambiguous. 5 Velte (2017) conducted a study focusing on the individual ESG components. The author compares their relationship to return on assets (ROA) and Tobin’s Q on a larger sample of companies in Germany. The study revealed a more pronounced negative correlation between the governance component and return on assets when compared to the environmental and social components. Additionally, their study found no significant relationship between any of the ESG components and Tobin’s Q. The author suggests that one explanation for their finding might be the more established tradition of corporate governance reporting in Germany, which dates to the introduction of the GCGC in 2002, or the heightened importance of these reports to stakeholders. 2.3 ESG and cost of capital El Ghoul et al. (2011) explores the impact of corporate social responsibility (CSR) on cost of equity for a large sample of U.S. firms between the years 1992-2007. The study utilizes various methods to estimate firms’ ex ante cost of equity and discovers that firms with higher CSR scores tend to have lower costs of equity financing. Furthermore, the study emphasizes that investments in responsible employee relations, environmental policies, and product strategies lower the cost of equity for firms. In line with El Ghoul et al. (2011), Bhuiyan and Nguyen (2020) conducted a study focusing on the impact of CSR activities on a company’s financial performance, with a specific focus on the cost of debt and the overall cost of capital. The authors employ a methodology like El Ghoul et al. (2011), and their study concludes that higher CSR performance results in a lower cost of capital, thereby indicating a negative relationship between these two factors. Furthermore, Ng and Rezaee (2015) conducted a similar study, but their focus was on the relationship between ESG factors and the cost of equity. Their study contained a sample of 3,000 firms over the period from 1990 to 2013. From the data set, the authors concluded that only the environmental and governance components of ESG exhibited a significant negative relationship to the cost of equity. Apart from prior CSR/ESG studies and its relationship to cost of capital, Eliwa et al. (2021) instead investigated non-financial firms within the European Union to explore a potential relationship between ESG performance and the cost of debt. The study utilized a sample of 6018 firms, with data collected over the period from 2005 to 2016. The results of their research 6 indicate that firms stand to gain from enhancing their ESG performance and disclosure levels. ESG improvements are reflected in a reduced cost of capital imposed by lending institutions. Furthermore, their findings suggest that market forces, manifested by lending institutions, play a crucial role in elevating the reliability of ESG disclosure and in influencing sustainable development. Moreover, their study proposes that the influence of ESG practices on cost of debt is more pronounced in countries that are oriented towards stakeholder interests, and they further argue that ESG practices might be effectively evaluated by civil society as a potential agent for driving change in business behavior. A substantial amount of previous research indicates a negative relationship between ESG or CSR and cost of capital, but there are studies presenting contrasting views. Magnanelli and Izzo (2017) offer one such perspective. Their research diverges from the common trend, revealing a positive relationship between CSR performance and the cost of debt, thus suggesting a different dynamic in how CSR impacts financial outcomes. In addition, Chava (2014) identified a positively significant relationship between expected returns and environmental concerns. However, noted was that there was no statistically significant correlation between expected returns and the strength of a company’s disclosed environmental profile. A study providing a behavioral aspect of ESG and CSR was conducted by Harjoto and Jo (2015) focusing on the interpretation of CSR activities by sell-side analysts. Their research explores the varying effects of diverse types of CSR (overall, legal, and normative) on various financial aspects. These aspects include the dispersion of analysts’ earnings forecasts, the volatility of stock returns, the cost of equity capital, and the overall value of the firm. Their study showed evidence that higher levels of overall CSR activities lead to a decrease in the dispersion of analysts’ earnings forecasts, a reduction in stock return volatility, and a lower cost of capital. Additionally, these activities contributed to an increase in the overall value of the firm. Furthermore, when the authors categorized CSR activities into legal or normative types, the outcomes differed significantly. Legal CSR tend to decrease the dispersion of analysts’ earnings forecasts, reduce stock return volatility, and lower the cost of capital. Conversely, normative CSR exhibited the opposite effect in these areas. The concluding remarks from this study suggest that sell-side analysts face less information asymmetry when dealing with legal CSR, primarily because it is mandated by law and thus provides a more standardized basis for comparison with other companies. In contrast, when companies engage 7 primarily in normative CSR, it tends to create greater information asymmetry, making comparisons and analysis more challenging. In the article from Baldini et al. (2018), the authors examine how numerous factors at the country and firm levels influence the practices of ESG disclosure by companies. The study used a sample of 14,174 firm-year observations from 2005 to 2012. The findings of the research indicated that characteristics at both the country and firm levels play a significant role in shaping firms’ practices of ESG disclosure. At country level, the authors argue that the characteristics of “Corruption” and “UnemploymentRate” have a homogeneous effect on ESG disclosure. Baldini et al. (2018) concludes that countries with higher levels of corruption tend to have lower disclosure of ESG-related practices, as they are more prone to engaging in unethical practices. In contrast, the “UnemploymentRate” characteristic suggests that companies in countries with higher unemployment rates tend to disclose more about their ESG practices. This is attributed to the desire of managers to attract higher-skilled employees by demonstrating a greater commitment to ESG standards. Meanwhile, characteristics specific to a firm, especially those linked to its visibility, consistently showed a positive influence on the disclosure of all ESG aspects and the authors discussed that managers tend to increase ESG disclosure when exposed to public visibility. Furthermore, the authors argue that it is consistent with legitimacy theory as the results confirm that ESG disclosure can be used as a tool to appear socially aware. Nollet et al. (2016) examine the relationship between ESG scores and return on capital and return on assets. They utilize both linear and non-linear models to investigate this relationship. The study focus on firms listed on the S&P 500, using ESG scores between the years 2007 to 2011. Their analysis reveal a significant negative correlation between ESG scores and return on capital in the linear model. However, the non-linear model exhibit a U-shaped relationship. Additionally, a deeper examination of ESG’s individual components by Nollet et al. reveal that only the governance aspect demonstrate the U-shaped correlation with the return on capital as identified in the non-linear model. Research by Bae et al. (2018) challenges the conventional wisdom that more CSR investment invariably leads to better financial performance. The core question to their research is to explore whether the relationship between CSR activities and the cost of debt is monotonic, meaning does increasing CSR investments continuously decrease the cost of debt, or if there is 8 a point beyond which further CSR investments do not yield financial benefits. The authors focus on loans issued by U.S firms and how these firms’ CSR activities affect the spread on these loans. The spreads represent the interest rate premiums over the baseline interest rate that firms must pay on their borrowed funds. The results suggest that the benefit of CSR strengths is not monotonic. Initially, as CSR activities increase, there is a corresponding decrease in the cost of debt, reflected through lower loan spreads. However, this benefit declines at a decreasing rate (ibid). Furthermore, Bae et al. (2018) provide evidence that there exists an optimal level of CSR engagement. When firms exceed this level, their CSR investments are viewed by lenders as ineffective or even wasteful, suggesting that overinvestments in CSR can be financially disadvantageous (ibid). By exceeding the optimal level of CSR, firms might not only gain additional benefits but might incur higher costs of debt (ibid). Similar results are found by Ye and Zhang (2011) who examines the relationship between CSR investments and cost of debt in China. The authors discover a U-shaped relationship illustrating that while CSR initially reduce the cost of debt, excessively high levels of CSR lead to increased financing costs. The results suggest a threshold beyond which the cost associated with additional CSR activities begin to outweigh the benefits in terms of reduced debt costs. Additionally, the authors differentiates the impact of CSR on firms of different seizes, finding that smaller firms have a higher optimal CSR level compared to larger counterparts. By illustrating that not all CSR-, or by extension, ESG efforts are financially beneficial when scaled, the results from both Ye and Zhang (2011) and Bae et al. (2018) directly support our hypothesis that there might exist a threshold beyond which a higher ESG score outweigh the financial benefits in terms of capital costs. 9 Table 1. Summary of literature. Author (s) Region Time period ESG or CSR Relationship ESG or CSR Bae et al. (2018) US 1991 - 2008 Non-linear relationship to cost of debt. CSR Baldini et al. (2018) Global 2005 - 2012 Relationship between sustainability disclosure and ESG practices. ESG Bhuiyan and Nguyen (2020) Australia 2004 - 2016 Negative relationship to cost of debt and cost of equity. CSR Chava (2014) Global 1992 - 2007 Positive relationship to cost of debt and cost of equity. Not specified El Ghoul et al. (2011) US 1992 - 2007 Negative relationship to cost of equity. CSR Eliwa et al. (2021) Europe 2005 - 2016 Negative relationship to cost of debt. ESG Harjoto and Jo (2015) Global 2003 - 2012 Negative relationship to cost of debt and cost of equity. Both Magnanelli and Izzo (2017) India 2015 - 2021 Positive relationship to cost of debt. CSR Ng and Rezaee (2015) Global 1990 - 2013 Negative relationship to cost of equity. ESG Negative relationship to cost of equity & Nollet et al. (2016) US 2007 - 2011 a non-linear relationship to the governance component in ESG. ESG P. Velte (2017) Germany 2010 - 2014 Negative relationship to return on assets. ESG Ye & Zhang (2011) China 2007-2008 Non-linear relationship to cost of debt. CSR Source: Created by the authors. 10 2.4 Hypothesis development Based on our literature review, the general consensus suggests that high ESG scores correlate with reduced capital costs, attributing to a perceived lower risk by investors and lenders (El Ghoul et al., 2011; Ng and Rezaee, 2015; Bhuiyan and Nguyen, 2020). This is based on the premise that higher ESG scores, reflecting superior sustainability practices, benefit companies through enhanced investor appeal and potentially lower cost of capital. This leads us to our first hypothesis: H1: ESG scores significantly impact the cost of equity. However, there seems to be a lack of research investigating any drawbacks of striving for high ESG scores. The increased costs associated with achieving and maintaining high ESG scores raise critical questions about the economic efficiency of such investments. Recent studies, including those by Velte (2017) and Eliwa et al. (2021) suggest that the benefits of high ESG scores are not linear. In fact, studies from Bae et al. (2018) demonstrate that CSR practices, closely related to ESG practices, exhibit a non-linear effect on the cost of debt. They find that while CSR investments initially decrease the cost of debt, there exists an optimal level of CSR beyond which additional investments increase debt costs. Similarly, Ye and Zhang (2011) find a non-linear relationship between CSR and the cost of debt in China, suggesting that both high and low CSR levels can lead to increased financing costs compared to moderate levels. While some of the existing literature extensively examines the impact of CSR on financial metrics, we choose to focus on ESG scores as a parallel measure of corporate sustainability performance. Both CSR and ESG frameworks are rooted in the principle of creating long-term value through responsible business practices, but ESG scores offer a more standardized method of assessment. The inclusion of environmental and governance dimensions provides a comprehensive view that aligns with current sustainability reporting standards and investor criteria. This alignment justifies the use of ESG scores as a modern extension of CSR activities, and as such, our study not only builds upon previous CSR research but also more contemporary studies using ESG metrics. Based on our review of the existing literature and prevailing theories on the financial implications of ESG scores, our study identifies a gap in understanding the impacts of high 11 versus moderate ESG scores on the cost of equity. While numerous studies have explored the general negative relationship between ESG scores and cost of capital, detailed examinations of the non-linear effects of high ESG scores remain sparse. To the best of our knowledge, no study has specifically investigated the threshold at which the benefits of increasing ESG scores cease to outweigh the associated costs from the lens of shareholders. H2: There exists an ESG score threshold beyond which an increase in these scores has a diminishing impact on the cost of equity. 12 3. Method In this chapter, we present our research framework and methodologies used to answer whether there exists a threshold beyond which increasing a company’s ESG score has a diminishing impact on reduced cost of equity. Furthermore, we explain our chosen statistical models used for our statistical tests and regression analysis. 3.1 Methodology A quantitative research design has been adopted for this study to systematically investigate the relationship between ESG scores and cost of equity. This approach is chosen for its effectiveness in analyzing numerical data and for providing objective results that can be generalized to a broader population. 3.2 Data collection First, we determined which database would be most suitable for our statistical tests and regression analysis in the study. Several databases were available, including Capital IQ, Bloomberg, and Refinitiv Eikon. For our purposes, an important factor in data collection was the accessibility of ESG scores. Li and Polychronopoulos (2020) suggest that sourcing ESG scores from corporate websites and non-governmental organizations may not be optimal due to the absence of a standardized ratings methodology and the lack of comprehensive company ESG scores. Thus, we choose to rely on established providers like Bloomberg and Refinitiv Eikon for our data collection needs. However, due to limited access to Bloomberg’s database, we picked Refinitiv Eikon for the gathering of all required data. Data was collected from all listed companies in Europe to ensure a comprehensive understanding of the ESG-cost of capital relationship within this geographical context. The screening process yielded a sample of 423 firms. We employed a 9-year time frame and ESG scores were provided annually in the Refinitiv Eikon database, which gave us a total of 9 observations for each company and a total of 3 879 observations across 11 different sectors. (Refinitiv, 2023) In figure 1 below the distribution between the sectors is presented. 13 Figure 1. Distribution between sectors in our data set. Source: Refinitiv Eikon. Made by the authors. Refinitiv Eikon’s ESG scoring framework covers a range from 0 to 100, marking the lowest to the highest possible scores. This aggregated score is an integration of three distinct pillars: environmental, social, and governance. Each pillar allows for a firm to score anywhere between 0 to 100 and hence contributing to the overall ESG rating (Refinitiv, 2023). Additionally, the values for cost of equity and our other control variables, except the leverage ratio, were obtained from Refinitiv Eikon. We were unable to source the leverage ratio from Refinitiv Eikon and thus had to calculate it manually. We accomplished this by dividing the total debt by total equity for all companies in the data set. Moreover, to address our first hypothesis, we initiated our study by examining a potential significant relationship between ESG scores and cost of equity. For this phase, we incorporated all available ESG scores within our data set. In exploring our second hypothesis, our attention was narrowed to the higher ESG scores in the data set. Our preliminary action for this involved determining the median ESG score, which served to exclude the lower half of our ESG data and thereby centering our analysis for scores slightly below and values surpassing our calculated median value. Furthermore, we segmented these ESG scores into three separate percentile categories. Our first percentile included ESG scores between 55 – 60, the second percentile included all scores between 60 – 65 and the third percentile included all scores between 65 – 70. Surprisingly, all ESG values above 70 showed an insignificant relationship with our dependent variable and could therefore not be included to answer our second hypothesis. 14 3.3 Descriptive statistics Table 2 displays all the variables used in the study. In STATA, we applied the winsorized command to trim all variables, excluding values below the 1st percentile and above the 99th percentile in our data set. This method is used to exclude extreme values in the dataset. For the study’s independent variable, “ESG Scores”, we obtained a minimum value of 3.44, a maximum value of 92.23 and an average value of 61.39. It is noteworthy that our mean value is relatively high compared to ESG values used in previous research. For instance, studies by Velte (2017) and Bassen et al. (2022) presented mean values of 56.66 and 55.66, respectively. Nonetheless, there are studies with mean values higher than to ours as well, for example Eliwa et al. (2021) reported a mean value of 66.1 in their study. Furthermore, our minimum ESG value of 3.44 are in line with previous studies such as Bassen et al. (2022). As our reported mean ESG value is notably high, this suggests a prevalence of higher values over lower ones within our data set. Regarding our dependent variable, “Cost of Equity”, the values from our data set are somewhat higher than those reported in prior studies, although not significantly so. As illustrated in Table 2, the mean value is 0.08, which closely aligns with the mean value of 0.056 reported by Ng and Rezaee (2015) in their study. 15 Table 2. Descriptive statistics. VARIABLES N Min p25 Mean p75 Max sd ESG Scores 3,879 3.44 49.86 61.39 75.52 92.23 15.57 Cost of Equity 3,879 0.042 0.0589 0.0811 0.0992 0.204 0.0356 Weighted Average Cost of Capital 3,879 0.033 0.0472 0.0658 0.0814 0.161 0.0288 Volatility Earnings per share 3,879 0.008 0.019 0.162 0.152 2.670 0.352 Number of employees 3,879 17 4.1* 42.2* 43.8* 672.8* 80.4* Market Cap 3,879 17.4** 2.5*** 18.6*** 18.4*** 442.4*** 34.9*** Revenue 3,879 25.6* 1.5** 12.9** 12.5** 356.3** 26.7** Total Assets 3,879 59.5* 2.8** 24.0** 20.1** 978.3** 59.4** Total Equity 3,879 3.2** 1.2*** 8.0*** 7.5*** 203.2*** 17.0*** Leverage Ratio 3,879 0.009 0.320 0.906 1.091 6.041 4.881 Number of companies 423 423 423 423 423 423 423 * in thousands, ** in millions, *** in billions Source: Refinitiv Eikon. 3.4 Regression variables 3.4.1 Dependent variables In our stated regression analysis, we have chosen to include one dependent variable: “Cost of Equity”. Prior studies in the fields of accounting and finance have employed various methods for manually calculating these variables. For instance, determining the implied cost of equity is considered an effective approach as it distinctly separates the impacts of cost of capital from those related to growth and cash flow effects (Hail and Leuz 2006; Hail and Leuz 2009; Chen et al. 2009; El Ghoul et al. 2011). However, given that Refinitiv Eikon provides published values for the variables we are examining, we choose to maintain consistency in our data gathering process by only using data from Refinitiv Eikon for all variables. 3.4.2 Independent variables In our regression analysis, we included the following independent variables: "ESG Scores," "ESG First Percentile", "ESG Second Percentile" and "ESG Third Percentile”. For our first hypothesis, we employed the first independent variable and for our second hypothesis we also 16 employed the percentile-based independent variables. When we segmented our ESG scores by percentiles, it naturally resulted in fewer observations in each category. Specifically, the first percentile contained 369 observations, the second had 410, and the third 428 observations. 3.4.3 Control variables Control variables are used to isolate the effect of the independent variables on the dependent variable, thereby enabling a more accurate interpretation of their relationship. In our study, we incorporate a set of control variables that are grounded in both theoretical reasoning and empirical evidence from prior studies (Hail and Leuz 2006; Gebhardt et al 2001; Dhaliwal et al. 2006). These variables include the natural logarithm of market capitalization (LN Market Cap), the natural logarithm of the number of employees (LN Number of Employees), leverage ratio (Leverage Ratio), and industry classification (Industry). LN Market Cap and LN Number of Employees serve as proxies for company size. In line with Hail and Leuz (2006) and Gebhardt et al. (2001), size is a critical determinant of a firm’s cost of capital. Larger firms typically benefit from economies of scale, more diversified operations, and greater market recognition, which can translate into a lower cost of capital due to perceived lower risk by investors and creditors. By logging these variables, we mitigate the influence of extreme values. The leverage ratio, computed as the ratio of total debt to the market value of equity, is included as a control variable to account for the financial structure’s impact on the cost of capital. Consistent with the findings of Dhaliwal et al. (2006), a higher leverage ratio is expected to increase the cost of capital due to the increased financial risk associated with higher debt levels. Leverage affects not only the risk perception among investors but also the potential for financial distress. Lastly, we control for industry effects to account for the systematic differences in the cost of capital across sectors, in line with the methodology of El Ghoul et al (2011). Industries vary in their capital intensity, regulatory environment, and cyclicality, which can significantly influence their financing costs. 17 3.4.4 Empirical identification For the study’s regression models, we created two different equations to answer our hypothesizes. The first equation investigates the direct relationship between ESG scores and the cost of equity, while controlling for various factors that may impact the cost of equity. The control factors include the natural logarithm of market capitalization, the natural logarithm of the number of employees, the leverage ratio, and the industry. Additionally, an error term is incorporated to account for the variation in the cost of equity not explained by the model. Hypothesis 1 Regression Model Cost of Equity as a dependent variable 𝐶𝐶𝐶𝐶𝐶𝐶it = 𝛽𝛽0 + 𝛽𝛽1 × 𝐶𝐶𝐸𝐸𝐸𝐸_𝐸𝐸𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠𝑖𝑖,𝑡𝑡 + 𝛽𝛽3 × 𝐶𝐶𝑆𝑆𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐶𝐶𝑠𝑠𝑖𝑖,𝑡𝑡 + 𝜖𝜖𝑖𝑖,𝑡𝑡 The study’s second equation instead examines how different percentiles of ESG scores influence the cost of equity. In this model, ESG scores are divided into three distinct groups. As in Model 1, we control for various factors that might affect the cost of equity: the natural logarithm of market capitalization, the natural logarithm of the number of employees, the leverage ratio, and the industry. An error term is also included to capture the variation in the cost of equity that the model does not explain. Hypothesis 2 Regression Model Cost of Equity as a dependent variable with percentiles 𝐶𝐶𝐶𝐶𝐶𝐶it = 𝛽𝛽0 + 𝛽𝛽1𝐶𝐶𝐸𝐸𝐸𝐸_𝐸𝐸𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠𝑖𝑖,𝐶𝐶 + 𝛽𝛽2𝐶𝐶𝐸𝐸𝐸𝐸 𝐸𝐸𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠𝑖𝑖,𝑡𝑡 × 𝐶𝐶𝐸𝐸𝐸𝐸 𝐹𝐹𝑖𝑖𝑆𝑆𝑠𝑠𝐶𝐶 𝑃𝑃𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶𝑖𝑖𝐶𝐶𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽3𝐶𝐶𝐸𝐸𝐸𝐸 𝐸𝐸𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠𝑖𝑖,𝑡𝑡 × 𝐶𝐶𝐸𝐸𝐸𝐸 𝐸𝐸𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝑆𝑆 𝑃𝑃𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶𝑖𝑖𝐶𝐶𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽4𝐶𝐶𝐸𝐸𝐸𝐸 𝐸𝐸𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠𝑖𝑖,𝑡𝑡 × 𝐶𝐶𝐸𝐸𝐸𝐸 𝑇𝑇ℎ𝑖𝑖𝑆𝑆𝑆𝑆 𝑃𝑃𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝐶𝐶𝐶𝐶𝑖𝑖𝐶𝐶𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛽𝛽5𝐶𝐶𝑆𝑆𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐶𝐶𝑠𝑠𝑖𝑖,𝑡𝑡 + 𝜖𝜖𝑖𝑖,𝑡𝑡 COE = Cost of Equity (dependent variable) ESG Scores = Environmental, Social and Governance Scores Controls = LN Market Cap, LN Number of employees, Leverage Ratio and Industry. First percentile = ESG Scores between 55-60. Second percentile = ESG Scores between 60-65. Third percentile = ESG Scores between 65-70. i = firm t = year 18 4. Results and Robustness Checks In this chapter, we present the results obtained from our statistical models that were designed to first investigate the impact of ESG scores on the cost of equity, and later examine the potential ESG threshold. The chapter presents findings that either confirm or disprove our hypotheses. Lastly, we present robustness test to test the validity of our results. 4.1 Results As presented in table 3 model (a), we can confirm our first hypothesis. Interestingly, compared to prior research, our findings indicate a much lower reaction where a unit increase in ESG scores corresponded to only an 8.17 bps increase in cost of equity, this result is statistically significant at a 1% level. LN Market Cap and LN Number of employees are also statistically significant on a 1% level, the former has a negative correlation of -0.007, indicating that companies with higher market capitalization have a lower cost of equity in our study. However, a unit increase in LN Number of employees will increase the cost of equity with 20 bps. Volatility Earnings per share the other control variable Industry are statistically significant on a 5% significance level. For our second hypothesis, illustrated in model (b), we explored whether there exists an ESG threshold beyond which an increase in these scores has a diminishing impact on the cost of capital. According to our second regression analysis below, ESG scores have a positive and significant impact on the cost of equity for companies in the first percentile, at a 5% significance level. A unit increase in ESG scores will increase the cost of equity with 8.12 bps, indicating that higher ESG scores in this group correlate with a higher cost of equity. However, for companies in the second percentile, the table shows a negative correlation for cost of equity, indicating that a unit increase in ESG score within this group would lower the cost of equity with -0.0019. This result is significant at a 5% significance level. Consistent with the outcome of our first hypothesis, our second regression analysis demonstrated low coefficient values. This indicates that, despite the distinct delineation between the first and second percentiles, the divergence are relatively minor. 19 Table 3. Regression Analysis for hypothesis 1 and 2. Regression Analysis VARIABLES Cost of Equity Cost of Equity Model (a) Model (b) ESG Scores 0.000817*** 0.000812*** (0.0000539) (0.0000541) ESG First Percentile 0.00185** (0.000941) ESG Second Percentile -0.00190** (0.000902) ESG Third Percentile 0.0000144 (0.0000231) LN Market Cap -0.00701*** -0.00677*** (0.00101) (0.00101) LN Number of employees 0.00225*** 0.00249*** (0.000395) (0.000397) Volatility Earnings per share 0.00997*** 0.00975*** (0.00246) (0.00246) Industry 0.000377** 0.000400** (0.000188) (0.000188) Constant 0.152*** 0.141*** (0.0227) (0.0228) Observations 3,879 3,879 R-squared 0.083 0.086 Number of companies 423 423 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 20 4.2 Robustness Checks To ensure the validity of our results and construct solid arguments based on our data, it is crucial to confirm that the statistical assumptions underlying the OLS model are intact. The data in this study are cross-sectional, with observations recorded annually. Thus, as outlined by Wooldridge (2010) and Hausman (1978), we have conducted three specific statistical tests: a Hausman test, a test for multicollinearity, and robustness check using robust and clustered standard errors in our models. Hausman test from, Hausman (1978), follows a chi-squared distribution with degrees of freedom corresponding to the number of variables is utilized to determine the appropriate model for panel data analysis, choosing between fixed effect and random effect regression. The null hypothesis supports a random effect model, and the alternative hypothesis favors the fixed effect model in the test. After conduction the Hausman test we could determine that our regression model is better suited to a fixed effects approach, which we adopted. To enhance the robustness of our results, we included a test for multicollinearity, which tests for high correlations among variables. High correlations suggest that variables may convey similar information, potentially rendering some unnecessary for the regression model (Poole and Farrell, 1971). In our analysis, we utilized a Variance Inflation Factor (VIF) test. The VIF scores for our independent variables ranged from 1.00 to 1.73, indicating that multicollinearity is unlikely to be inflating the variance of the estimated coefficients in our model. See Appendix 1. Lastly, we conducted a test for heteroscedasticity for our regression models by using robust and clustered standard errors instead of regular standard errors. As shown in Appendix 2, the model values remained relatively unchanged and continued to be significant. This indicates that the results are reliable and not excessively affected by heteroscedasticity. 21 5. Analysis and Discussion This chapter discuss the findings from our regression analyses with the existing theoretical and empirical literature to draw conclusions about the impact of ESG scores on the cost of equity and to explore the complexities of the potential ESG score threshold. Our findings are aligned with Legitimacy Theory which posits that firms gain legitimacy through conformity to social norms, including environmental and social standards (Meyer and Rowan, 1977). This adherence can be reflected in their ESG scores. However, contrary to some expectations, the results from our first hypothesis suggest that higher ESG scores correlate with a higher cost of equity. This positive relationship indicates that while higher ESG scores may enhance a firm’s legitimacy, they also reflect increased costs associated with maintaining these standards, which can be perceived as higher cost by investors. Similarly, Stakeholder Theory, which advocates for the importance of valuing all stakeholder interests, supports our findings that higher ESG scores have a significant impact on the cost of equity. Freeman et al. (2010) emphasizes the benefits of addressing stakeholder interests, yet our results align with Friedman (2007), who suggests that not all stakeholder-oriented actions directly translate to increased value. There relatively low marginal effect of ESG scores on cost of equity suggests a cautious approach to ESG investments. While initial investments in ESG may significantly reduce the cost of equity, the benefits diminish as scores increase. This aligns with the results of Magnanelli and Izzo (2017), who observed a positive relationship between CSR performance and the cost of debt, suggesting that overinvestments in CSR activities could potentially lead to increased cost of capital. The observed threshold effect where the benefits of higher ESG scores diminish aligns with the theoretical discussions by Bae et al. (2018) and Ye and Zhang (2011), who noted similar non-linear relationships in their studies on CSR investments and cost of debt. Our results, which demonstrates a diminishing impact of ESG scores on the cost of equity beyond a 60-65 ESG score level, provides empirical support for these theoretical predictions. Our results suggest the existence of an optimal ESG score threshold beyond which additional investments in ESG practices may not yield proportional financial returns in terms of cost of capital. 22 Our results suggest that the ESG threshold falls within the 60-65 score range, unexpectedly lower than what we initially presumed. The results are statistically significant and further suggests that investors consider ESG scores to be economically significant only up to a certain threshold. Beyond this point, the influence of ESG scores on investment decisions becomes marginal. This suggests a misalignment between investor behavior and the integration of ESG scores into investment decisions. The observed threshold for ESG scores impacting the cost of equity, falling within the mean range of our sample, indicates that investors may not be fully accounting for ESG scores when evaluating investment opportunities. This lag could stem from two main factors that would have dual implications for policymakers. Firstly, the perceived complexity of ESG assessments may hinder investor’s understanding of how ESG scores can positively affect the cost of equity. This could be due to a lack of knowledge and understanding of ESG metrics among investors. For policymakers, this highlights the need for standardized reporting and clearer guidelines to help investors make more informed decision based on ESG metrics. Secondly, the lag may reflect a broader hesitation for investors to shift from traditional focus on financial performance to more integrated approaches that also consider non-financial metrics. To address this, policymakers could consider not only enhancing transparency through standardized reporting but also incentivizing ESG integration. Our results contribute to the ongoing discourse as outlined by several studies, (El Ghoul et al. 2011; Ng and Rezaee, 2015; Bhuiyan and Nguyen, 2020; Eliwa et al. 2021) who identified a negative correlation between ESG factors and cost of capital. By confirming these findings within the European context, our study strengthens the case that ESG scores have a significant impact on the cost of capital. However, the identified threshold introduces a new perspective that questions the linear assumption common in prior studies, like El Ghoul et al. (2011), which did not consider potential non-linear effects. 23 6. Conclusion & Final discussion In this chapter, we address our research question and discuss the broader implications of these findings for investors and policymakers. We reflect on the validation of our results and propose directions for future research to build upon this study. 6.1 Conclusion Our study examines the relationship between ESG scores and the cost of equity, driven by our research question: “Is there a threshold beyond which a higher ESG score outweigh the financial benefits in terms of capital costs?”. The result of our study confirms our first hypothesis that ESG score has a significant impact on the cost of equity, as a unit increase in ESG score amount to an 8.17 bps increase in cost of equity. Although the effect is minor, it is statistically significant on a 1% significance level. Furthermore, our findings also confirm our second hypothesis that there exists a threshold beyond which an increase in ESG scores has a diminishing impact on the cost of equity. This threshold illustrates the same non-linear ESG- cost relationship that Bae et al. (2018) and Ye and Zhang (2011) found in their respective studies. We found that ESG scores influence the cost of equity differently across different percentiles, with significant variances between the 55-60 range and the 60-65 range. Although we expected the threshold to lie in a higher ESG score range, our results might indicate that investors are lagging behind and don’t seem to incorporate company ESG scores in their investment decisions as the threshold lies within the mean ESG score from our sample. The implications of this lag for policymakers suggest that they may need to consider more standardized, transparent, and incentivized ESG reporting to facilitate a more widespread adoption of ESG scores in investment decision making. To the best of our knowledge, this is the first study examining the potential ESG score threshold within European companies. Hence, the study contributes to a better understanding of the complexities of integrating non-financial metrics into financial decision making. Lastly, the result of our study challenges the predominant assumption of a linear relationship between ESG score and cost of equity. 24 6.2 Limitations of the study and suggestion for further research While our study provides new insights within the field of sustainable finance, we recognize that our study has several limitations. Firstly, only a limited number of firms possess an ESG- score, potentially causing our study to overlook some firms due to the absence of these scores. Secondly, the ESG-scores are sourced from just one of many agencies that provides them, which could introduce bias into our findings. Thirdly, the scope of the data is limited to European companies, this may affect the results in primarily two ways; it may affect the generalizability to other regions outside of Europe with different regulatory standards and on the contrary, there may be significant variations even among European countries. Lastly, our study does not take industry factors into account to the extent that we would have wished for, as the industry in which companies operate has a significant impact on their ESG score. To build on current research and address these limitations, we propose further research to examine the relationship between ESG scores and cost of equity among companies within the same industry. This would allow to determine whether certain industries exhibit more pronounced benefits among high ESG scores compared to others, or if various types of industries feature distinct ESG thresholds. The results would provide interesting knowledge on investors’ preferences regarding ESG scores, segmented on industry. Additionally, given the current lack of data on ESG scores, it would be interesting to conduct this study in a few years when more countries and companies are required to disclose more of their sustainability practices with proper benchmarks. 25 References Bassen, A., Kordsachia, O., Lopatta, K., & Tan, W. (2022). Revenue Alignment with the EU Taxonomy Regulation. Bhuiyan, M. B. U., & Nguyen, T. H. N. (2020). Impact of CSR on cost of debt and cost of capital: Australian evidence. Social Responsibility Journal, 16(3), 419-430. Bae, Sung C., Kiyoung Chang, and Ha-Chin Yi. (2018). “Are More Corporate Social Investments Better? Evidence of Non-Linearity Effect on Costs of U.S. Bank Loans.” Global finance journal 38: 82–96. Baldini, M., Maso, L. D., Liberatore, G., Mazzi, F., & Terzani, S. (2018). Role of country-and firm-level determinants in environmental, social, and governance disclosure. Journal of Business Ethics, 150, 79-98. Chava, S. (2014). Environmental externalities and cost of capital. Management science, 60(9), 2223-2247. Chen, K. C., Chen, Z., & Wei, K. J. (2009). Legal protection of investors, corporate governance, and the cost of equity capital. Journal of corporate finance, 15(3), 273-289. Deephouse, D. L., & Carter, S. M. (2005). An examination of differences between organizational legitimacy and organizational reputation. Journal of management Studies, 42(2), 329-360. Dhaliwal, D., Heitzman, S., & Zhen Li, O. L. I. V. E. R. (2006). Taxes, leverage, and the cost of equity capital. Journal of Accounting Research, 44(4), 691-723. El Ghoul, S., Guedhami, O., Kwok, C. C., & Mishra, D. R. (2011). Does corporate social responsibility affect the cost of capital? Journal of banking & finance, 35(9), 2388-2406. 26 Eliwa, Y., Aboud, A., & Saleh, A. (2021). ESG practices and the cost of debt: Evidence from EU countries. Critical Perspectives on Accounting, 79, 102097. European Commission. (nd). Corporate Sustainability Reporting. https://finance.ec.europa.eu/capital-markets-union-and-financial-markets/company-reporting- and-auditing/company-reporting/corporate-sustainability-reporting_en [2024-01-14] Freeman, R. E., Harrison, J. S., Wicks, A. C., Parmar, B. L., & De Colle, S. (2010). Stakeholder theory: The state of the art. Friedman, M. (2007). The social responsibility of business is to increase its profits. In Corporate ethics and corporate governance (pp. 173-178). Berlin, Heidelberg: springer berlin heidelberg. Gebhardt, W. R., Lee, C. M., & Swaminathan, B. (2001). Toward an implied cost of capital. Journal of accounting research, 39(1), 135-176. Hail, L., & Leuz, C. (2006). International differences in the cost of equity capital: Do legal institutions and securities regulation matter?. Journal of accounting research, 44(3), 485-531. Hail, L., & Leuz, C. (2009). Cost of capital effects and changes in growth expectations around US cross-listings. Journal of financial economics, 93(3), 428-454. Harjoto, M. A., & Jo, H. (2015). Legal vs. normative CSR: Differential impact on analyst dispersion, stock return volatility, cost of capital, and firm value. Journal of Business Ethics, 128, 1-20. Hausman, J. A. (1978). Specification tests in econometrics. Econometrica: Journal of the econometric society, 1251-1271. J. W. Meyer & B. Rowan. (1977). "Institutional organizations: formal structure as myth and ceremony," American Journal of Sociology, 83, 340-63. 27 Li, F., & Polychronopoulos, A. (2020). What a difference an ESG ratings provider makes. Research affiliates, 24. Madison, N., & Schiehll, E. (2021). The effect of financial materiality on esg performance assessment. Sustainability (Basel, Switzerland), 13(7), 3652. Magnanelli, B. S., & Izzo, M. F. (2017). Corporate social performance and cost of debt: The relationship. Social Responsibility Journal, 13(2), 250-265. Morgan Stanley Corporate International (MSCI). (2020). ESG and the cost of capital. https://www.msci.com/www/blog-posts/esg-and-the-cost-of-capital/01726513589 [2024-01-14] Morgan Stanley Corporate International (MSCI). (2023). ESG Ratings Process. https://www.msci.com/documents/1296102/34424357/MSCI+ESG+Ratings+Methodology+- +Process.pdf/820e4152-4804-fe33-0a67-8ee4c6a8fd7d?t=1666300410683 [2024-01-14] Ng, A. C., & Rezaee, Z. (2015). Business sustainability performance and cost of equity capital. Journal of Corporate Finance, 34, 128-149. Nollet, J., Filis, G., & Mitrokostas, E. (2016). Corporate social responsibility and financial performance: A non-linear and disaggregated approach. Economic Modelling, 52, 400-407. Poole, M., & Farrell, P. (1971). The Assumptions of the Linear Regression Model. Transactions - Institute of British Geographers, 52(1971), 145-158. UNFCCC. (2024). The Paris Agreement. https://unfccc.int/process-and-meetings/the-paris-agreement [2024-02-27] Refinitiv. (2023). Environmental, Social and Governance Scores from Refinitiv https://www.lseg.com/content/dam/data-analytics/en_us/documents/methodology/lseg-esg- scores-methodology.pdf [2024-02-23] 28 Schaefer, B. P. (2008). Shareholders and social responsibility. Journal of Business Ethics, 81, 297-312. Tse, T. (2011). Shareholder and stakeholder theory: After the financial crisis. Qualitative Research in Financial Markets, 3(1), 51-63. Velte, P. (2017). Does ESG performance have an impact on financial performance? Evidence from Germany. Journal of Global Responsibility, 8(2), 169-178. Wooldridge, J. M. (2010). Econometric analysis of cross section and panel data. MIT press. Ye, Kangtao, and Ran Zhang. (2011). “Do Lenders Value Corporate Social Responsibility? Evidence from China.” Journal of business ethics 104.2: 197–206. 29 Appendix Appendix 1. Test for multicollinearity In the table below our VIF values are presented. All variables have a VIF value below 2 which suggest a low level of multicollinearity among the predictor variables in our regression models. Hence, the estimates of the regression coefficients are likely to be stable and reliable. Variance Inflation Factor (VIF) VARIABLES ESG Scores 1.35 ESG First Percentile 1.06 ESG Second Percentile 1.04 ESG Third Percentile 1.04 LN Market Cap 1.73 LN Number of Employees 1.49 Volatility 1.03 Industry 1.02 Leverage Ratio 1.00 Mean VIF 1.20 30 Appendix 2. Robust and clustered standard errors In the model below, we observe some minor differences when using robust and cluster standard errors compared to the model presented in Table 3. However, these differences do not significantly impact the results, as the result remains economically insignificant. Regression Analysis (including robust and clustered standard errors for both models). (a) (b) VARIABLES Cost of Equity Cost of Equity ESG Scores 0.000816*** 0.000812*** (0.0000896) (0.0000881) ESG First Percentile 0.000532** (0.0000385) ESG Second Percentile -0.000815** (0.0000352) ESG Third Percentile -0. 0000144 (0.0000247) LN Market Cap -0.00702*** -0.00677*** (0.00206) (0.00207) LN Number of employees 0.00226*** 0.00237*** (0.000568) (0.000550) Volatility Earnings per share 0.00998*** 0.00974*** (0.00238) (0.00237) industry 0.000377*** 0.000380*** (0.0000998) (0.0000974) Constant 0.152*** 0.145*** (0.0454) (0.0458) Observations 3,879 3,879 Number of companies 423 423 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 31