NLKDNJKLnkjls
The relationship between
CSR and the interest bearing
cost of debt
An analysis of Nordic countries
Authors: Oscar Carlsson & Jonas Wahlström
Authors: Oscar Carlsson, Jonas Wahlström
Department of Business Administration Civilekonomprogrammet Degree Project, 30 Credits, Spring 2020
ii
Abstract
Corporate social responsibility (CSR), or perhaps the lack of it, is widely debated today. Regulations and international discussions surrounding the topic is taking up a larger room in the public domain with the Paris Climate Accords being a pinnacle where a number of countries decided to agree to quite ambitious climate goals. For companies, there could be incentives to early adapt to these accords to be better suited for the future. But even during present time, it could be beneficial to as early as possible readjust into a more sustainable direction. The Nordic countries are ranked among the top in the world when it comes to sustainability. Can this however, lead to a reduction in costs for Nordic companies that invest into CSR?
The purpose of this study is to examine if there is a link between socially responsible companies and the cost of interest bearing debt. This was done by conducting a number of regression models, both linear and non-linear. The data used have a time frame of 10 years and span between 2009 to 2018 with a total number of 106 firm’s and 796 firm-year observations. CSR scores were collected using Thomson Reuters and the results show that there is a statistically significant negative relationship between CSR and the cost of interest bearing debt. The effects CSR have on the cost of interest bearing debt are somewhat inconsequential and the conclusion is made that CSR can be a major determinant for interest bearing debt, but it is related to the size of the debt portfolio. Further, the results of this research could be viewed as a component in a more comprehensive model.
Keywords: ESG, CSR, sustainability, cost of interest bearing debt, risk management,
iv
Acknowledgments
We would like to thank our supervisor Jörgen Hellström for his invaluable input, feedback and patience. This thesis would not possible without his aid.
We would also like thank each other, this time has been, for better and worse, to say the least eventful. But we are proud to say that we made it through together and are more accomplished because of it.
Oscar Carlsson & Jonas Wahlström
vi
Table of Contents
1. INTRODUCTION ... 1
1.1PROBLEM BACKGROUND ... 1
1.2PROBLEM DISCUSSION ... 2
1.3PURPOSE AND RESEARCH QUESTION ... 3
1.4PRACTICAL AND THEORETICAL CONTRIBUTION ... 4
1.5DELIMITATIONS ... 4 2. SCIENTIFIC METHOD ... 6 2.1CHOICE OF PARADIGM ... 6 2.1.1 Ontological assumptions ... 6 2.1.2 Epistemological assumptions ... 7 2.2RESEARCH DESIGN ... 7 2.3RESEARCH APPROACH ... 7 2.4LITERATURE SEARCH ... 8 2.5SOURCE CRITICISM ... 9 3. PREVIOUS LITERATURE ... 10 3.1ARTICLES ... 10
3.1.1 ESG relationship to financial performance. ... 10
3.1.2 Corporate social performance relationship to corporate financial performance. ... 10
3.1.3 Corporate social performance relationship to idiosyncratic risk. ... 11
3.1.4 CSR relationship to firm risk. ... 11
3.1.5 ESG relationship to the cost of capital. ... 12
3.1.6 CSR relationship to cost of debt. ... 12
3.1.7 CSR relationship to the costs of bank loans. ... 13
3.2IMPLICATIONS FROM PREVIOUS LITERATURE. ... 13
4. THEORETICAL FRAMEWORK ... 15
4.1THE THREE MAIN CATEGORIES WITHIN ESG ... 15
4.1.1 Environmental ... 15
4.1.2 Social ... 15
4.1.3 Governance ... 16
4.2ESG,RISK AND AGENCY THEORY ... 16
4.2.1 Risk mitigation view ... 16
4.2.2 Overinvestment view ... 17
4.2.3 Efficient Market Hypothesis ... 18
4.3THE COST OF CAPITAL ... 18
4.3.1 Weighted Average Cost of Capital ... 19
4.3.2 The Cost of Equity ... 19
4.3.3 The Cost of Debt ... 20
vii
5.1DATA COLLECTION AND PROCESSING ... 21
5.2MULTICOLLINEARITY ... 22
5.3HETEROSCEDASTICITY ... 23
5.4OUTLIERS ... 24
5.5THE ESGSCORE ... 25
6. RESEARCH METHOD ... 27
6.1POPULATION AND SAMPLE ... 27
6.2STATISTICAL HYPOTHESIS ... 28
6.3REGRESSION ANALYSIS ... 28
6.4THE LEAST SQUARES ASSUMPTIONS IN MULTIPLE REGRESSION ... 28
6.5DESCRIPTIVE STATISTICS. ... 29
6.6MODELS ... 30
6.6.1 Regression model ... 30
6.6.2 Non-linear regression model ... 30
6.7VARIABLES ... 31
6.7.1 Dependent variable ... 31
6.7.2 Independent variables ... 31
7. EMPIRICAL RESULTS ... 33
7.1RESULTS ... 33
7.2LINEAR REGRESSION RESULT ... 33
7.2.1 The linear relationship ... 33
7.2.2 The non-linear relationship ... 34
7.3COUNTRY SPECIFIC TESTS ... 35
8.1ANALYSIS OF LINEAR RESULTS ... 37
8.2NON-LINEAR REGRESSION AND COUNTRY SPECIFIC ANALYSIS ... 39
8.3SUMMARY ... 40 8.4MODEL ANALYSIS ... 40 8.5CONCLUSION ... 41 8.6TRUTH CRITERIA ... 42 8.6.1 Reliability ... 42 8.6.2 Validity ... 43 8.6.3 Generalizability ... 43
8.7ETHICAL AND SOCIETAL CONSIDERATIONS AND IMPLICATIONS ... 44
8.8FUTURE RESEARCH ... 45
viii
Definitions
CSR = Corporate Social Responsibility, the stakeholder-, social-, economic-, voluntariness- and the environmental aspects of business from a firm’s perspective. Eikon = Database provided by Thomson Reuters.
ESG = Environmental, Social and Governance, an investors perspective of CSR. Key rate = The interest rate that determine banks lending rates and the cost of credit for borrowers (Investopedia, 2020).
Nordic Countries = Denmark, Finland, Norway and Sweden.
1
1. Introduction
1.1 Problem Background
In this day and age, companies not only have to focus on maximizing profits and keep within the legal limits. They are also expected to take on bigger responsibilities outside of the classical sphere of business. Larry Fink, the CEO of BlackRock sent a warning to executives in his last yearly letter to CEO’s where he pushed leaders to establish a corporate purpose. The notion with the purpose is to find a way where the corporation can co-exist with society and survive long-term (Fink, 2019). The purpose Larry Fink refers to are the corporations raison d’etre and to widen the view on stakeholders. The classical view is that stakeholders are directly involved with the corporation, for example as a lender, supplier or an employee. A different and broader view regarding stakeholder theory points to stakeholders as someone who is affected by externalities, and therefore, puts them at the receiving end of the negatives surrounding some industries. One way stakeholders and corporations can reach a destination of co-existence is to strengthen their Corporate Social Responsibilities (CSR).
2
1.2 Problem Discussion
CSR as a concept have been discussed since the 1950’s. Bowen (1953, referenced by Moura-Leite & Padgett, 2011, p. 529) describes firm’s as powerful entities and that managers should reflect on their social responsibility. Drucker (1954, referenced by Moura-Leite & Padgett, 2011, p. 530) identifies public responsibilities as one of the eight key areas for strategic planning. There was however not much discussion about the potential business benefits on CSR in the 1950’s. In the 1960’s Friedman (1962, referenced by Moura-Leite & Padgett, 2011, p. 530) presented a view that the main social responsibility for firm’s is to provide as much value to shareholders as possible. This view was previously described by Levitt (1958, referenced by Moura-Leite & Padgett, 2011, p. 530), they argued that CSR could be beneficial.
But if CSR was to interfere with profit maximization, shareholders value was preferred. However, around the same time that legislation added protection for consumers and employee’s, regulations where implemented on how corporations handled CSR (Lee, 2008, p. 57). The change in regulation still goes on today and firms are required to be proactive towards their social responsibilities and to take further action outside the voluntary realm. The definition of CSR is widely debated and to the authors knowledge there is not a set consensus at this point in time. CSR is closely related to ESG, which is an abbreviation for Environmental, Social and Governance. ESG and CSR are merely different perspectives on the same topic. ESG is an outside look on organizations work with social responsibilities and are strategically implemented into corporations to reflect their CSR. Therefore, these terms will be used interchangeably within the study.
An analysis of 37 different definitions of CSR concludes that they all refer to five dimensions. They include the stakeholder-, social-, economic-, voluntariness- and the environmental dimension (Dahlsrud, 2006, p. 5). These are in line with Carrolls (1991, p. 42) previous framework where he describes CSR through four responsibilities that laid out a framework for corporations to use when trying to implement the strategic CSR decisions into everyday operations. The European Union updated their definition of CSR in 2011 to; “the responsibility of enterprises for their impacts on society” with the follow up; “To fully meet their corporate social responsibility, enterprises should have in place a process to integrate social, environmental, ethical, human rights and consumer concerns into their business operations and core strategy in close collaboration with their stakeholders, with the aim of;
3 As previously noted, CSR as a research subject have been relevant since the 1950s. During this time, several thousand studies have been completed, most of them have been investigating if there are potential financial performance gains to be obtained through CSR. The general notion is that a firm performance does benefit from CSR, according to a meta study conducted by (Friede et al., 2015, p. 210) who covered over 2000 articles. Other potential benefits include a lessened risk of default (Sun & Cui, 2014, p. 282), lower credit risk (Stellner et al., 2015, p. 538) and a reduction in firm’s risk as shown by Jo and Na (2012).
Another area of research related to CSR covers the relationship between cost of capital and CSR. This research area is not as extensively covered as financial performance and the results are somewhat ambiguous, but it does nonetheless show promise. The areas which have been covered are genereal cost of capital (Sharfman & Fernando, 2008, Suto & Takehara, 2017), cost of equity (Dhaliwal et al., 2015; Ghoul et al., 2011) and bank loans (Goss & Roberts., 2010; Izzo & Magnanelli, 2017) in relationship to CSR. The last two are the only two papers that was identified which covered CSRs relationship to debt the debt side of financing.
As CSR now is one of the most researched subjects in business it is surprising that the relationship between CSR and cost of debt have not been more thoroughly investigated. Furthermore, it is interesting how previous research have found either inconclusive or a negligible impact on the cost of debt given that studies have demonstrated how CSR can be used to decrease a firm's risk profile which should lower a firm's debt costs since the debt cost, at least in theory, always is influenced by a firm's perceived risk.
1.3 Purpose and research question
How companies finance themselves are generally categorized through debt or equity financing, where most companies have a mix of the two. What effects CSR have on the equity side of financing have been more thoroughly researched compared to the debt side. Further, there are valuation models that is starting to emerge where CSR is a determinant for the valuation of corporations. In light of this, it is believed that the understanding of what effects CSR have on the debt side of financing needs more research. As a step to understand the effects further, only Nordic countries will be researched in this study. The reason for this is that Nordic countries have a reputation of being progressive in many different areas and this holds true for CSR as well, Nordic countries are ranked top in the world when it comes to sustainability. (Robeco, 2019). This implies a high interest for CSR related topics and the study could benefit from investigating potential effects in countries where the interest in CSR already is established to be high.
This study mainly aims to clarify if there is a relationship between CSR and the cost of interest bearing debt. This is as previously mentioned, not a well-researched subject even though CSR as a concept is studied extensively. The study aims to build on previous research in risk management, CSR and the connection between CSR and different financial key figures and investigate the Nordic countries and its publicly quoted companies.
4 ● What effects, if any, does CSR have on the cost of interest bearing debt for
publicly quoted companies in Nordic countries?
1.4 Practical and theoretical contribution
Given the importance of CSR, all research that can widen the perspective on it can be both of practical and theoretical use. The most obvious practical contribution this study have is to clarify if the cost of interest bearing debt can be affected by CSR. This can be useful as a variable in a model for valuing firm’s using CSR scores in the valuation. But it can also be useful for CFOs or executive groups when making strategic decisions. Given how important the interest bearing cost of debt are to most corporations the ability to better understand the underlying factors is crucial. An economic incentive for firms to be more sustainable is a win-win situation for both firms and stakeholders and something that holds a potential practical impact.
On a theoretical basis, this thesis tries to expand on previous research within CSR to create a more thorough understanding, since there is a lack of studies on the relationship between debt cost and CSR and previous results are inconclusive. Furthermore, this study will target CSR within a Nordic context, an environment where 4 of the 5 Nordic countries are ranked among the top 5 in the world in terms of sustainability (Robeco, 2019). Most of the articles that expands on CSR do not cover the Nordic countries, because of this it would be even more interesting to obtain a knowledge of the relevant aspects for firms finances in a context where CSR is particularly prioritized, and to what extent CSR makes a difference.
1.5 Delimitations
The delimitations chosen to investigate the link between ESG and costs of interest bearing debt are the following:
• The scope of the study will focus on Nordic firms located in Sweden, Denmark, Finland and Norway listed on Nasdaq with a standardized ESG rating.
• The dependent and independent variables including the ESG rating has been retrieved through Thomson Reuters Eikon. The ESG rating (refered to as the ESG Score in this study) consists of the ESG Score and ESG Controversy Score provided by Thomson Reuters Eikon (Thomson Reuters, 2017, p. 7). The data from Thomson Reuters Eikon has been the source of data for many peer-reviewed articles and is considered a trusted source (Magnanelli & Izzo, 2017, p. 14; Garcia et al., 2017, p. 135).
6
2. Scientific method
2.1 Choice of paradigm
The first aspect a researcher needs to consider is what paradigm best suits the intended research. A paradigm includes a philosophical framework which is used as a guide when executing scientific research. Philosophy is in the case of scientific research one's perception of the elemental nature of knowledge, existence and reality (Collis & Hussey, 2014, p. 43). The Researchers view of reality is referred to as their ontological assumption, their view of knowledge is referred to as their epistemological assumption. Positivism has been identified as the appropriate paradigm for this study. Theories provide the foundation which can explain phenomena when conducting research with a positivism view, further the theories facilitate the anticipation of phenomena, forecast their occurrence and through this process allow them to be controlled. Understanding of phenomena contain identifying one or several causes for a relationship among variables (Collis & Hussey, 2014, p. 4). One can accomplish this through identifying laws of causation, these laws claim that the changes in nature is always the result of a cause (Merriam-Webster, 2019). The identified laws are then linked to either a deductive or integrated theory.
According to Collis and Hussey (2014, p. 45) the positivism view has it flaws, critic questioning this philosophy argue that: People and social context are two variables that coexist and cannot be divided. It is essential to understands people’s perceptions to understand people. This type of research design is rigid and may ignore valuable findings. Researchers within the study can´t act objectively when studying a phenomenon. Phenomena could potentially be oversimplified when using a single measure.
When identifying, what paradigm would be appropriate the following aspect was taken into consideration: What kind of data was needed (the study will require a large sample of quantitative data to compare the ESG score and the cost of interest bearing debt for the publicly quoted Nordic firm’s). Hypothesis testing through statistical analysis. The findings should reflect the general effects of CSR on financial costs for firm’s that operate in Nordic countries.
Collis and Hussey (2014, p. 50) claim that these traits are typical for the positivism paradigm. The above-mentioned reasons make the positivism paradigm appropriate to use as a guide through this study´s process of research.
2.1.1 Ontological assumptions
As previously noted, the term ontology refers to one's perspective on reality (Collis & Hussey, 2014, p. 47). The positivism view argue that social reality is objective, which means that only one reality exists, and all people experience reality in the same way. A researcher who accepts this philosophy presume that reality can be investigated objectively without being disturbed (Collis & Hussey, 2014, p. 44).
7 researchers accept the positivism view the results the test provide should enable understanding of how the ESG scores affects firms cost of interest bearing debt.
2.1.2 Epistemological assumptions
The term for a researcher’s perception of what should be viewed as acceptable knowledge is epistemology. Within positivism information must be scientifically verifiable, this means that only phenomena that are observable can be considered knowledge (Bryman, 1997, p. 24). Appropriate information should be mathematical or in other ways provide logical proof for the phenomenon which is investigated (Collis & Hussey, 2014, p. 44). Positivism presumes that it is possible to measure social phenomena and is linked to quantitative methods of analysis which is conducted through statistical analysis with quantitative data (Collis & Hussey, 2014, p. 44). Data in numerical form is quantitative and this data, if considered precise and objective, is common within the positivism paradigm (Collis & Hussey, 2014, p. 50). The hypothesis will be investigated through statistical testing and since the researchers presume that the data used is objective it is viewed upon as appropriate for the study.
2.2 Research Design
To investigate the hypothesis, a longitudinal study design has been chosen. This design enables researchers to continuously follow the selected variables over a time-frame suitable for the study (Collis & Hussey, 2014, p. 64). The repetitive observations have the benefit of enabling detection of the studied relationships stability over the course of the studies time-frame (Collis & Hussey, 2014, p. 64).
In this study’s case, it will enable the researcher to follow the selected firms ESG Scores and costs of interest bearing debt over the chosen time-frame. It will reveal if there is a relationship between the two, if the relationship changes as time passes and if the firms interest bearing cost of debt increase or decrease during this time.
2.3 Research approach
8 Figure 1. The deductive research approach (Bryman, 1995, p.31)
The hypothesis is based on the theories within the chosen theoretical framework and the hypothesis will be tested through statistical analysis. The intention is to assess if the theories within the theoretical framework can explain the results, this as previously mentioned is typical for a deductive approach. Also, typical for a positivism view is the quantitative data and statistical analysis used to carry out the study. For the above mentioned reasons, the deductive approach is assumed to be appropriate for this study.
2.4 Literature search
That this study would include sustainability and finance was clear from the start. Sustainability is a pressing topic and the affect most companies have on the environment and other sustainability aspects are substantial. To possibly find incentives for companies to further their efforts towards sustainability would mean a great deal to the researchers. Both researchers, majors in finance and knew that a combination of the two subjects would be interesting to them and relevant to their education.
During the search of finding appropriate literature the first matter was to identify databases which could give access to academic journals and other scholarly literature. The databases used are Umeå University’s library database and Google Scholar which both provide scholarly literature, including peer-reviewed scientific articles. Through these databases the search for study´s on sustainability and its relation to different financial aspects began. The intention was to provide a base of knowledge surrounding the kind of study´s related to sustainability which previously have been conducted and where further research might be appropriate.
Key Words
9
2.5 Source Criticism
To ensure that the literary sources are credible, most of the articles used in the study are peer-reviewed. Primary referencing has been prioritized to further ensure that references consistently maintain a high quality.
A significant number of previous articles which have been referred to were published quite a few years ago. The researchers are aware that old articles possibly will not provide information which is up to date, especially since sustainability is an area which has become more and more relevant and what is viewed as sustainable firm procedures continuously is developing. However, some of the most important CSR studies which are referred to are the Meta studies by Friede et al., (2015) and Orlitzky et al., (2003, p. 427). Despite that some of the articles included in these studies are quite old, the sheer volume of articles and different time spans covered make them necessary to include since they provide the most in depth information about CSRs relationship to finance available.
10
3. Previous literature
The following section will screen some of the previous literature within CSR, it will include a couple of studies the authors perceive to be relevant for their study. The screening will begin to cover some of the most thoroughly covered relationships of CSR. Then two studies focusing on CSR and firm risk will be introduced, these are relevant since firm risk is assumed to be positively correlated with the cost of debt. Then a study of CSR and cost of capital is covered since the even more specific areas of CSR relationship within the cost of capital, introduced later, probably would not exist if this broader area was not previously covered. Lastly CSR and the cost of debt and CSR and the cost of bank loans are covered, since these are the most specific research areas and the area this study intend to contribute to.
3.1 Articles
3.1.1 ESG relationship to financial performance.
ESG and financial performance: Aggregated evidence from more than 2000 empirical studies. Friede, Bush & Bassen, 2015.
Friede et al. (2015), released what might be the most comprehensive study on the relationship between CSR and corporate financial performance (CFP). The study is a second-hand review of both vote count, meta and econometric review studies. They reviewed 60 other reviews with a total of 3718 underlying studies. Around 2200 unique studies remained after adjusting for overlap in the reviews (Friede et al. (2015, p. 210). The result shows that a high ESG score, in general, is positive for corporate financial performance (Friede et al., 2015, p. 427). However, there are some mixed results, mainly when comparing portfolio to non-portfolio based studies. Portfolio studies have a much higher spread between being positively and negatively affected by CSR and Friede et al. (2015, p. 220) argues that a major problem with portfolio studies is that the aggregated data from a portfolio often do not add up to the individual company data.
The notion of how CSR affects CFP is relevant, mainly when it comes to looking at CSR as a whole, including both performance and costs.
3.1.2 Corporate social performance relationship to
corporate financial performance.
Corporate Social and Financial Performance: A Meta-analysis. Marc Orlitzky, Frank L. Schmidt, Sara L. Rynes, 2003.
11 corporate environmental performance. In other words, it is similar to the ESG score but without the environmental aspects. Others studies only use corporate environmental performance. What was clear from dividing and comparing the two was that the social performance relationship with CFP was nearly twice as strong as environmental performance, but booth do positively impact CFP (Orlitzky, et al., 2003, p. 415).
The relationship between CSP and CFP represents the largest body of knowledge surrounding what potential value firms can achieve through CSR. It seems to be the vantage point from where all other CSR related studies have emerged from and for this reason is necessary to cover.
3.1.3 Corporate social performance relationship to
idiosyncratic risk.
Corporate Sustainability Performance and Idiosyncratic Risk: A Global Perspective. Lee and Faff, 2009.
Lee and Faffs study investigated the link between firm´s Corporate sustainability performance (similar to other CSR scores) and their idiosyncratic risk. The study’s sample is comprised of firms from 34 different countries across 51 different industries (Lee & Faff, 2009, p. 217). The results are adjusted for size and book-to-market, since larger and highly valued companies in general experience lower risk (Lee & Faff, 2009, p. 216). They compare leading and lagging Corporate Sustainability Performance firms, their results indicate that leading firms have significantly lower idiosyncratic risk than lagging firm’s (Lee & Faff, 2009, p. 216). Previous studies have used so called negative screens where some low Corporate Sustainability Performance industries among others tobacco, firearms and alcohol have been excluded. This study does however use an index that do not exclude industries and the results should because of this be more reliable (Lee & Faff, 2009, p. 216).
This study hypothesis assumes that lower firm risk does reduce costs of interest bearing debt. Since leading CSP firm’s experience lower idiosyncratic risk, as shown by Lee and Faff (2009), it is reasonable to assume that firm’s cost of interest bearing debt should therefore, as well be lowered through Corporate Sustainability Performance and other CSR Scores.
3.1.4 CSR relationship to firm risk.
Does CSR Reduce Firm Risk? Evidence from Controversial Industry Sectors. Jo and Na 2012.
non-12 controversial industries (Jo & Na, 2012, p. 452-453). The authors suggest that this could be due to that firm-risk is a more severe problem for non-controversial industries and might use CSR as a risk management tool (Jo & Na, 2012, p. 453).
Jo and Na´s study, as well as Lee and Faff´s study (2009, p. 228), demonstrate how firm’s risk profile benefit from CSR. The hypothesis in this study is in part based on the assumption that a firm's perceived risk is a relevant aspect for a firm's cost of interest bearing debt, the two above mentioned studies provide support for the reasoning behind the hypothesis.
3.1.5 ESG relationship to the cost of capital.
The link between ESG, alpha, and the cost of capital: Implications for investors and CFOs. Kölbel and Busch 2017.
In an article published in the Corporate Finance Biz, Kölbel and Busch (2017), reviews what effects ESG has on firm’s alpha and cost of capital. The method is to review existing research with the aim to use that research to draw conclusions on residual earnings and risk (Köbel & Busch, 2017, p. 82). Note that this article is not a full meta study, but a review based on what the authors claim to be high quality papers.
Since the article is a review there is no set theoretical framework. Instead it consists of the 14 reviewed articles combined frameworks (Köbel & Busch, 2017, p. 83). The study separates existing literature into alpha and risk studies (Köbel & Busch, 2017). Alpha studies try to establish if a portfolio that is generated using ESG criteria, will yield abnormal returns compared to its given benchmark. The assumption is that financial markets are inefficient when it comes to pricing ESG criteria, and therefore mispricing can occur (Köbel & Busch, 2017, p. 83). The risk studies however assume that financial markets are efficient and that ESG measures are priced correctly. This leads to a reduction in capital cost with an increase in ESG performance.
The authors conclude that the financial risk is lower for firms with high ESG performance. This enables firms to save a substantial amount of money as they raise debt due to the lower rates. Further, this study concludes that there is little to no evidence that an ESG focused firm will financially outperform its non ESG focused counterparts. This is contradicting other studies, including meta studies (Friede, et al., 2015) and (Orlitzky, et al., 2003), that states that different types of CSRs do have a positive effect on financial performance.
3.1.6 CSR relationship to cost of debt.
Corporate social performance and cost of debt: The relationship. Magnanelli and Izzo, 2017.
13 to this one since potential benefits include the possibility of a decrease in debt financing costs.
The results of the study are in contrast with the hypothesis. They show that CSR scores in fact is positively correlated with the cost of debt and therefore a higher CSR will most likely not be the driver behind a decrease in risk (Magnanelli & Izzo (2017, p. 257). On the other hand, it could lead to an increase in the risk profile. This is explained by the overinvestment view which is a form of agency theory where managers invest too much into CSR, potentially to gain benefits, on the expense of the shareholders (Magnanelli & Izzo (2017, p. 257). The results are interesting and could be interpreted in different ways. One being that managers might need more time to learn how to efficiently implement CSR into their businesses without losing confidence from stakeholders.
3.1.7 CSR relationship to the costs of bank loans.
The impact of corporate social responsibility on the cost of bank loans. Goss and Roberts, 2010.
Goss and Roberts (2010), explores the relationship between bank loans and CSR from a lender perspective. Banks are interesting as a subject because of the special inside information banks have compared to other lenders. In the study banks are assumed to only hold one agenda and that is the borrower’s ability to repay its debt (Goss & Roberts, 2010, p. 1795). The sample is made up of almost 4000 loans given to US firms (Goss & Roberts, 2010, p.1797). The authors differentiate between firms in two separate ways. First through CSR, mainly how strongly the firm invests in it and secondly with the use of collateral as a signal to differ between high- and low-quality borrowers to examine if there is a difference in bank loan costs given the quality of the firm (Goss & Roberts, 2010, p. 1794). The results show that low quality borrowers are given a penalty in the form of higher interest rates when investing in to CSR (Goss & Roberts, 2010, p. 1807). However, the increase in rate is modest and the reasoning for that is believed to be that banks consider CSR to be a second-tier determinant (Goss & Roberts, 2010, p. 1807).
Goss and Roberts study is similar to the study conducted by Magnanelli and Izzo (2017) with the exception that they only focus on bank loans, while this study tries to capture a larger portion of the total debts. One interpretation of the result is that a higher CSR investment acts as a buffer for reputation risk. And that buffer manages to make the change in collateral less significant than it in fact is. Goss and Robert (2010, p. 1794-1795), hypothesis builds on the risk mitigation view and agency theory. Both are used in this thesis and is more thoroughly described in the theoretical framework.
3.2 Implications from previous literature.
15
4. Theoretical Framework
4.1 The three main categories within ESG
4.1.1 Environmental
The extent of financial consequences companies face because of environmental issues they or others have created are not yet known. There is however, a growing demand among investors for firms to face topics which affect the environment. Climate change being the main topic, the sub-topics are constituted of operations within firm’s that cause and reduce the impact they have on the environment. Among others they include environmental hazards in the air, water and soil, recycling, energy and energy extraction-related risks and energy efficiency (Robeco, 2019). Potential consequences when not dealing with these risks are of course increased costs related to among others reputational damage, restoration of polluted landscapes and litigation costs (Robeco, 2019). Of course, there are potential benefits to integrating environmental considerations within corporate strategy. Being efficient with resources does decrease costs and the increasing demand for environmental responsibility among potential customers can be viewed as an opportunity rather than a threat. Some emphasize that it rather should be viewed as a necessity given how much neglecting climate change might cost companies moving forward.
The cost associated with a rise in global temperature is hard to accurately estimate, but if the temperature rises to a point where damages to the natural ecosystem is irreversible then it is natural that the uncertainty surrounding having businesses in general will be high. All companies should, as a long-term plan, strive for as small changes in the ecosystem as possible even as a strategic business decision given the uncertainty that potentially could occur. In the US some scientists have estimated a 2 % GDP loss every year if the temperature rises with 4 degrees (Brookings, 2019). This is a significant amount and that figure is mainly made up off the increase in natural disasters and the cost associated with them.
4.1.2 Social
16
4.1.3 Governance
Governance within corporations is their governing practices and involve principles, responsibilities, rights and the expectations one has on an organization's board of directors.
Deliberate corporate governance practices make sure that all stakeholders needs are considered, this would include consideration of, shareholders, government, communities, clients, suppliers, financiers and management (Robeco, 2019). Basically, a firm's corporate governance need to aid a sustainable strategy.
4.2 ESG, Risk and agency theory
4.2.1 Risk mitigation view
To understand the Risk mitigation view, an understanding of what kinds of risks a firm can experience is essential. Different industries experience these risks in various degrees, Figure 2 depicts an overview of the relevant aspects that a lender need to take in to account when lending money to a firm. The exception being that industries outside of banking and finance, do not directly experience risks surrounding regulatory capital, since no other firms than banks and financial institutions have access to regulatory capital, which only is provided by a nation’s Central Bank or the equivalent.
Figure 2: The risks that banks and financial institutions face (Hull, 2015, p. 547).
In terms of CSRs relationship to risk, CSR could potentially influence all of these risks. But to the authors knowledge there have only been studies covering CSRs relationship to idiosyncratic risk and idiosyncratic risks do not include market risks. Lee and Faff (2009) and Jo and Na’s (2012) studies are used as a hypothesis builder for this study. Lee and Faff (2009, p. 228) finds that higher CSR scoring firms tends to have lower idiosyncratic risk than their lagging counterparts. Jo and Na (2012, p. 447) studied controversial industries and have similar findings. The literature on the link between CSR and CFP also indicates that investments in CSR will be beneficial for financial performance.
17 A vital aspect of any company is their reputation. Risks surrounding a company’s reputation are several, many of which are interconnected with CSR. Reputational risks are probable adverse outcomes to publicity, public impression or incidents beyond a company's control that affect its reputation, which through this decrease revenue (Reputation Management, 2019). These risks seem plausible to counteract through CSR. The CSR controversies score that is used in combination with the standard CSR score in this study is in some way a measurement on the effects of reputational risk. A controversy that potentially could be hedged through a better management of reputation risk can have a significant impact on the CSR score. When looking closer at the data it is evident that the more considerable moves in CSR score between years mainly are a product of a controversy and therefore, a mismanagement of the reputational risks.
4.2.2 Overinvestment view
One approach to social responsibilities for companies was launched by Milton Friedman during 1970. His attitude toward the matter, was that the only socially responsible factor that companies need to address is to maximize profits for their shareholders (Friedman, 1970, referenced by Goss & Roberts, 2010, p. 1795-1796). He claims out that a company, as an artificial person, can not have responsibilities other than maximizing wealth for the shareholders, but that the businessmen and women behind the company might feel that they have responsibilities, and if they do, they can voluntarily act on them if they want to (Friedman, 1970, referenced by Goss & Roberts, 2010, p. 1795-1796). In other words, CSR is argued by Friedman to be an individual concern and a waste of shareholder’s wealth when conducted by an agent for a company.
There are however, some benefits to CSR investments on a large scale. Fewer people must spend time to learn of where their investment can be utilized the most. Therefore, a type of discretionary management where experts makes rational decisions based on where the funds would be utilized in the most efficient manner could be argued being more efficient compared to every individual making that decision. Another potential benefit of a grouped philanthropy is the reduction in transaction costs associated with many smaller donations to places where it potentially could be very costly and complex to send funds. So, CSR as a consolidated effort could be viewed as something negative or positive depending on what view one would have on the subject. There are some concerns regarding CSR investments derived from agency theory, potentially it could be used by managers to obtain personal benefits on the expense of shareholders. It can be argued that CSR is the product of agency theory within a firm (Kruger, 2015, p. 305). In a case where managers act selfish in CSR investments for their own benefits, debtors would be correct to identify the overinvestment and in that case the interest bearing cost of debt is expected to rise given that value is not created by the investment. This will be considered when analyzing the results in this study.
18 and because of this, a non-linear regression will in addition to the linear regression be executed.
4.2.3 Efficient Market Hypothesis
The Efficient Market Hypothesis (EMH) was launched by Fama (1970) as a review of previous research. The reasoning behind the EMH is to optimize financial allocations and that a correct pricing is needed (Fama, 1970, p. 1575). What the EMH states is that every asset is priced according to the available information. The market will quickly adapt to all available information and therefore, assets will be priced correctly. The technological progress has been immense since the EMH was launched. Information is now instant and according to the EMH this leads to fewer occasions of mispricing (Fama, 1970, p 1601). The Efficient Market Hypothesis is divided into three forms of efficiency. The weak, the semi-strong and strong form. The weak form of market efficiency is when price is governed only by historical pricing. The semi-strong form is when the pricing reflects all available information on the market. This means that the pricing of an asset will instantly be adjusted as soon as new information is made public. The pricing is always a reflection of publicly available information. The strong form, however, occurs when pricing is adjusted given all available information. This includes private information that the market usually is not aware off. When it comes to the first two forms, abnormal return could be possible, however in a strong form of market efficiency no abnormal returns are possible since all investors are sufficiently informed. Financial markets are believed to be semi-strong, and the efficiency of the financial markets are basically linear with the technological development as algorithms and faster information flows eliminate the majority of mispricing. However, it could not be strong as there still for example exist insider information that is not readily available to the public domain.
For the EMH to hold true a couple of assumptions must be fulfilled. Firstly, there should be no transaction costs between agents on the market. Secondly, all information should be free and available to all investors and finally there should be some sort of consensus surrounding how this information should be valued when pricing an asset. These conditions are rarely in place on a market, but Fama (1970, p. 1575) states that if enough participants on the market fulfill the conditions, the market will be efficient, at least in some form. The EMH is not without its critics or doubters. The very notion that some investors continuously over performs the market should not be possible over time if the market was efficient. Flash crashes is also an indication where companies are not valued correctly, albeit usually under a short time frame.
Even though this study’s focus is not on the financial performance, the EMH is still relevant. The reasoning behind this is that financial markets are highly efficient (Fama, 1970, p. 1601) which indicate that the possibilities for abnormal returns in the financial marketplace should be scarce. The EMH will be used to examine if it might be possible to receive a beneficial market position given each companies ESG rating and in return receive a lesser cost of interest bearing debt.
4.3 The Cost of Capital
19 investors want a premium to invest in a company to compensate for the risk. If there was no premium, an investor would be better off by not investing at all or invest in what is commonly referred to as a risk-free asset like a treasury bill as it adds limited or no risk to the investment. With that in mind, the price of capital can be seen as a pricing mechanic for a companies potential risk.
4.3.1 Weighted Average Cost of Capital
In general debt financing tends to be less expensive compared to equity financing. A shareholder do not have a direct claim on a company’s free cash-flow and the asset type is deemed riskier. Even though it is generally cheaper with debt financing compared to equity this cost cannot be isolated. As the debt portion of the total capital budgeting increases so does the leverage. With a higher leverage the risk will increase, and investors will because of this demand a higher yield to invest in the company and therefore offset the potential effect of the relatively cheaper debt cost of capital.
To capture the complete cost of a firm’s capital the Weighted Average Cost of Capital (WACC) is used. WACC is derived from Modigliani and Miller’s second proposition
“that the cost of capital of levered equity increases with the firm’s market value debt-equity ratio” (Berk and DeMarzo, 2014, p. 489).
𝑟!"## = 𝐸 𝐸 + 𝐷𝑟$+ 𝐷 𝐸 + 𝐷𝑟%(1 − 𝜏&) 𝐸 = 𝐹𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑓𝑖𝑟𝑚 𝑣𝑎𝑙𝑢𝑒 𝑓𝑖𝑛𝑎𝑛𝑐𝑒𝑑 𝑏𝑦 𝑒𝑞𝑢𝑖𝑡𝑦 𝐷 = 𝐹𝑟𝑎𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑓𝑖𝑟𝑚 𝑣𝑎𝑙𝑢𝑒 𝑓𝑖𝑛𝑎𝑛𝑐𝑒𝑑 𝑏𝑦 𝑑𝑒𝑏𝑡 𝑟$ = 𝐶𝑜𝑠𝑡 𝑜𝑓 𝑒𝑞𝑢𝑖𝑡𝑦 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 𝑟% = 𝐶𝑜𝑠𝑡 𝑜𝑓 𝑑𝑒𝑏𝑡 𝜏& = 𝐶𝑜𝑟𝑝𝑜𝑟𝑎𝑡𝑒 𝑡𝑎𝑥 𝑟𝑎𝑡𝑒 (Berk and DeMarzo 2014, p. 490)
The equity financed part of the equation is represented by the market capitalization of the firm’s shares and debt is equal to all interest bearing debt that the firm hold.
4.3.2 The Cost of Equity
20 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑠𝑡𝑜𝑐𝑘 𝑟𝑒𝑡𝑢𝑟𝑛 = 𝑟'+ 𝛽C𝑟(− 𝑟'D
𝑟' = 𝑅𝑖𝑠𝑘𝑓𝑟𝑒𝑒 𝑟𝑎𝑡𝑒
𝛽 = 𝐵𝑒𝑡𝑎 (𝑠𝑦𝑠𝑡𝑒𝑚𝑎𝑡𝑖𝑐 𝑟𝑖𝑠𝑘) 𝑟( = 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑚𝑎𝑟𝑘𝑒𝑡 𝑟𝑒𝑡𝑢𝑟𝑛
CAPM have over time been heavily criticized given a variety of reasons and the model have been reworked over time and is the basis for Fama & French Three-Factor Model which included book-to-market ratio and firm size and Carhart Four-Factor Model which builds even further on the previous two models.
4.3.3 The Cost of Debt
The cost of interest bearing debt is measured by the interest rate. The interest rate is the premium a stakeholder receives in exchange for lending an asset to the borrower (Brealy, 2017, p. 224). Interest bearing debts can include for instance bonds, bank loans and different kind of leases. Usually the risk profile is what dictates the interest cost. The risk profile is an aggregate combination of the company’s internal and external prerequisites. The internal prerequisites consist of a number of different variables such as financial leverage, size, credit history, tangible asset that can be collateralized and intangible assets like patents, licenses, copyrights and brands. Furthermore, a firm’s cash-flow, result and general financial health also determines the interest rate.
21
5. Data
5.1 Data collection and processing
The data, as previously stated is gathered from Thomas Reuters Eikon database using DataStream. This is in line with similar studies like (Magnanelli & Izzo, 2017, p. 14; Garcia et al., 2017, p. 135). The rationale for choosing Eikon over other similar ratings like the Standard & Poor’s sustainability index was based on two main points. Firstly, Eikon’s rating universe consist of a wider range of ESG ratings for Nordic companies. Secondly, studies similar to this have used Eikon which improves the standardization. Eikon´s database covers ESG ratings for over 6000 companies and uses more than 400 metrics to do so. For this dataset Eikon provided ESG scores for 199 companies spread over Nasdaq OMX Stockholm, Nasdaq OMX Copenhagen, Nasdaq OMX Helsinki and Oslo Børs. The data was collected through a screening process in Eikon where the Nordic stock exchanges was the first filter. This yielded a result of 785 companies. The second filter was added to confirm that a reported ESG rating was in place, this was done by screening for companies with a ESG score of greater than or equal to 1. This excluded all companies that do not have a rating and 199 companies remained.
Eikon’s ESG rating consists of three parts, the classical ESG rating and an overlay for controversies. The third part is the combined score between the previous two. The most reasonable independent variable to use in this study is the ESG combined rating since it includes controversies. Since the focus of this study is aimed at the interest bearing cost of debt, the risk profile of firm’s is of high importance. Controversies can be detrimental for a company’s public image for example, and this is assumed to change the risk profile of a given firm in a negative manner. A higher risk should in theory lead to higher interest bearing cost of debt if ESG is a driver for risk and therefore makes the combined variable the most useable one. Eikon reported on ESG controversies in the same extent as the regular ESG rating. Therefore, no companies were excluded with ESG combined being the independent variable.
An adjustment was made for financial institutions by controlling every firm’s description manually and the final collected data from DataStream consisted of 137 firms. Several financial variables where simultaneously collected, and these include:
● Net debt (ND) ● Total debt (TD)
● Interest expense on debt (IOED) ● Total assets (TA)
● Total capital (TC) ● Total liabilities (TL) ● Beta (Beta)
● Market capitalization (MarkCap) ● Return on invested capital (ROIC)
22 was missing. Some variables were created using existing key ratios from DataStream but were constructed manually. The dependent variable in this study is cost of interest bearing debt, it was created using Interest expense on debt divided by Total debt. Market capitalization (Mark. Cap.) measure the value of the firm’s in the sample. One problem with Market cap is the different currencies used between the included countries. This was solved by utilizing Euro as the main currency and then adjust the Swedish, Danish and Norwegian companies to the Euro. The currency is converted for the given year on the 31 of December because that is the most common closing day of the financial year for most Nordic companies. After the currency adjustment, the natural logarithm was calculated. The final Market cap variable is therefore adjusted for currency and logarithmic. The Financial leverage (Fin. lev.) variable is calculated by dividing Total debt with Total assets.
A total of three different sets of dummy variables is used in the study. The first set is a country specific dummy variable. The second one is a sector variable. The sectors were decided by using the description of sectors on each firm from the specific exchange. The sample consists of in total nine different sectors. The last dummy variable is based on years and one is made for every year between 2009-2018.
Lastly a couple of macro variables is used to adjust for differences in risk profile between firm’s from different countries. The first macro variable is the key rate for each country and year. The key rate is calculated through a weighted average of the key rate throughout the year for each individual country. The second macro variable is the GDP growth for each respective country.
In total 31 firm’s lacked data for the minimum requirement of 4 years and was therefore excluded. Of the 106 firm’s remaining there was a couple where there were holes in the data, the holes were imputed using a mean of the existing data. In total 24 impites were used and for two firm 2 years of imputes were used. On these occasions a primary impute was created by using the mean of the existing 8 years and then another impute was created using the existing data with addition to the previous impute.
The data is sorted as panel data in Excel, this is a prerequisite to use STATA, which is the chosen analyzing tool.
5.2 Multicollinearity
If an independent variable is a perfect linear function of one of the other independent variables there emerges a problem called perfect multicollinearity. If perfect multicollinearity occurs within the formula it is not able to calculate the OLS estimator, from a mathematical point of view the reason for this is that perfect multicollinearity results in division by zero (Stock & Watson, p 239).
Another aspect to look over in a regression model is imperfect multicollinearity which implies that one or several independent variables are highly correlated with one another. In this case the regression model will still be able to provide a result, but the independent variable estimates will for at least one variable be inadequate (Stock & Watson, 2012, p. 244).
23 that any of the correlation coefficients do not hold a value equal to or beyond 0,8, because this would indicate a high risk of multicollinearity (Studenmund, 2014, p. 271-272). Table 1 display the correlation matrix for the included variables.
Table 1. Correlation matrix
The correlations do not indicate any problems with heteroscedasticity except for relationship between Market Cap and Total Assets. That these variables are strongly correlated is expected given the natural relationship between the number of assets and the market valuation of the firm. Even though the market cap not necessarily is a indicator of size, firm’s usually try and use their assets in an efficient way to maximize profits. With that in mind it is reasonable that a mature company with a high number of assets also generate an expected profit that is reflected in the market capitalization. The same argument can be made for the correlation between Return on Invested Capital and the market cap. A high profit will help to push the market cap upwards, therefore it is also expected that market cap and ROIC will correlate to a certain degree.
5.3 Heteroscedasticity
24
Table 2. Heteroscedasticity test
5.4 Outliers
To be able to best describe what effect CSR have on interest bearing debt the issue with outliers must be addressed. Outliers are observations in the dataset that deviate
significantly from the rest of the sample. In Figure 3 a scatterplot of interest cost and the number of observations show the outliers in the dataset.
Figure 3. Scatterplot for interest costs.
It is important to be careful when adjusting outliers as they could just be natural
variances in the sample and not a calculation error or an indication of data that have not been correctly reported. However, if outliers are believed to be an error they should be removed or possibly corrected. In this case, it is believed that there are some outliers that should be removed. The reasoning behind this is that the interest cost is extremely high in some observations and this skews the results of the regression into a direction that is believed to not be representable of the population. What exact percentage of interest cost that makes an observation a probable outlier is a judgement call, and it is based on the average interest cost of the dataset and the average interest cost of new or renewed loans in Sweden during 2009-2019. Sweden has the highest amount of
companies in the sample and the average interest rate for all Swedish companies could be an indicator of what interest rate to expect and accept as normal.
25 observations that are above 18% interest cost and these are determined to be outliers and are removed from the sample. This is a decision believed to be a good balance between providing as accurate results as possible and at the same time be very cautious about removing data from the sample. The adjustment is done by creating a dummy variable where all observations above 18% interest rate is given a value of 1 and the rest 0. This is an efficient method to capture all the perceived outliers in one variable.
5.5 The ESG Score
Among the previous literature, the measurement intended to reflect firm’s CSR differ in several studies. They differ among other less common reasons because of what kinds of measurements were available at that point in time, which areas the different measurement providers covered and if several providers were available. In this study the researchers have settled on the ESG score provided by Thomson Reuters to reflect CSR. Among the available measurements the two most prominent are S&P and Thomson Reuters which booth are considered reputable sources with an up to date scoring process. The later was chosen because of their more extensive coverage of the Nordic countries.
The ESG score is used to quantify a corporation's CSR into financial figures, this will enable comparing and rating corporations from a sustainable point of view. Figure 4 display the ESG score.
Figure 4. The ESG score (Refinitiv, 2019, p. 3).
Thomson Reuters provide three separate ESG scores, the traditional score, controversies and combined (Thomson Reuters, 2017, p. 6). The traditional ESG scores consists of 400 ESG measures grouped into 10 categories that are weighted proportionally per the amount of measures used in each category (Thomson Reuters, 2017, p. 7). This modified ESG score (called ESGC by Thomson Reuters) adds controversies to the score (is referred to as the ESG score in this thesis). It measures 23 different categories of potential scandals and if a corporation is involved they will be penalized and the score is affected (Thomson Reuters, 2017, p. 7).
26 2017). ESG Combined score evaluates firm´s ESG performance through the knowledge in the ESG pillars.
The following sections explain how the ESG score is calculated.
There are three separate pillars, one for each aspect of ESG divided in to ten total categories. The categories do no not hold equal weights in the final score as some are viewed as more important than others. 400 separate firm level measures for ESG are part of the scoring process and the total maximum score is 178.
The environmental pillar consists of the following categories; Resource use, Emissions and Innovation. The Social pillar consist of the following categories; Workforce, Human Rights, Community and Product Responsibility. The Governance pillar consist of the following categories; Management, Shareholders and CSR Strategy.
Also, part of the score is ESG controversies. The controversies are coverage by media, larger negative stories increase the impact of an ESG incident. The score cover the categories Community, Human rights, Management, Product Responsibility, Resource Use, Shareholders and Workforce. Each Category include one or several controversy topics. Combined they cover 23 different topics. The scores intention is to assess firm's sustainability impact and conduct.
The combination of the two, calculated with equal weight for a fiscal period with new controversies within the latest complete period, this is what the final ESG Combined score for a firm include. The score is calculated in the following fashion.
First, a percentile rank scoring methodology is used in the scoring process. Three factors are considered in the methodology, how many firms are worse than the current one, how many firms have the same score and how many firms have a score at all. Since valuation with percentile rank is rank based, outliers are not a serious issue. This is the case because the score distribution nearly is flat, which do not make averages and standard deviation helpful.
The categories within the score are weighed equal to the sum of indicators it consists of. Calculations with the normalized weights are excluded from indicators without data in the public domain. The score weights within the three pillars is presented in Appendix 1.
Second, the TRBC Industry Group is used as the benchmark for the score calculation within the Environmental and Social categories. During the governance category calculations, country is the selected benchmark because the practices for governance usually is more consistent within countries.
Third, the overall score calculation has an applied and automated factual logic, this automated process decides the weights within the categories. The driver is the number of measures included in each category compared to the indicators in the score framework. The categories do thus include multiple issues and some categories will hold a more substantial weight than others.
27
6. Research method
6.1 Population and sample
The population of the study consists, as stated in the research question, off publicly quoted companies on Nordic exchanges. There are five major exchanges where four are owned by Nasdaq and one is independent. The four Nasdaq exchanges consists off Nasdaq OMX Stockholm, Nasdaq OMX Copenhagen, Nasdaq OMX Helsinki and Nasdaq OMX Iceland. The Norwegian stock exchange is independent and goes by the name Oslo Børs. However, to be able to study the relationship between ESG and the interest bearing cost of debt a viable ESG score is necessary. There are different ways to measure ESG and even though objective variables are used in the scoring process, there is always a subjective element to the valuation. In this study, Thomas Reuters Eikon database is used to provide ESG scores. The goal is to be able to standardize the scoring in the most accurate way possible. Others studies such as Magnanelli & Izzo (2017, p. 258) and Garcia et al. (2017, p. 135) used Eikon which should simplify comparison.
Since the ESG score is the independent variable of this study it is crucial that companies have been scored. It is also important that companies have been scored over the course of multiple years to be able to produce a time series of results. During the screening process, all companies from Iceland was excluded given that no companies from Iceland were given a score. Therefore, our true population will only consist of companies from Sweden, Denmark, Norway and Finland.
All companies are not given a score, the process of scoring as explained in the variable section of the research method is extensive and a large quantity of data is analyzed. Only companies of a certain size have a rating and in the screening process when adding a filter for ESG scores only large cap companies remained.
This leads us to an original population of 785 listed Nordic firms. From that population, the requirement of a greater than or equal ESG score of 1 excluded 586 companies. Furthermore, all financial institutions are excluded (Goss & Roberts, 2010, p. 1797). There are multiple reasons for this, the business model is significantly different from other sectors mainly because debt is used as an operational mean to create value. Financial institutions are in a way, also subsidized by the state, if the institutions can be deemed as too big to fail, as firm’s that are a backed by a sovereign state might have a different risk profile then counterparts in different sectors. In the sample, there were 62 financial institutions and after excluding these 137 companies remained. Some firm’s that could be defined as financial institutions do however remain in the sample, for example Intrum AB and Ratos. The sole reasoning for this is that those firm’s do not fit the definition for why a financial institution should be excluded. They are not subjects to a state backed subsidy and for investment firms the operational income is created on the operational side of the firms they invest in.
28 have a delay on the interest cost. Therefore, t-1 is used on the independent variable of ESG score which moves the time series of ESG to include the years 2009-2017. The dependent variables time series consists of data between 2010-2018. The minimum requirement of accessible data was set to 4 years of ESG scores, therefore the sample include companies that as a minimum have reported scores between 2014-2017 with ESG score t-1. This requirement excluded 31 companies and the final sample consists of 106 firms.
6.2 Statistical hypothesis
The use of hypothesis testing enables researchers to assess what might be true if the null hypothesis (H0) is not. If the null hypothesis is rejected from a statistical point of view, this do not necessarily mean that the opposite is true, merely that it with the evidence at hand, the alternative hypothesis can not be rejected (Stock & Watson, 2012, p. 112).
This study´s research question is defined as: What effect does CSR have on the cost of interest bearing debt for publicly quoted Nordic firm's?
The following hypothesis has been constructed to investigate the research question:
H01: There is no relationship between the cost of interest bearing debt and firm’s ESG
score
H11: There is a relationship between the cost of interest bearing debt and firm’s ESG
score
6.3 Regression analysis
A multiple regression model is the chosen measure which will estimate how the cost of interest bearing debt is affected by firms ESG Scores. This method of analysis is in line with other financing focused CSR studies (Goss & Roberts, 2010; Magnanelli & Izzo, 2017; Ghoul et al., 2011). The model enables estimation of how Yı is changed by Xı, when the other regressors, X2ı, X3… Xnı also referred to as independent variables in the model are held constant (Stock & Watson, p. 242).
6.4 The least squares assumptions in
multiple regression
For the estimates in the regression model to appropriately reflect the unknown regression coefficients there are four assumptions that need to be fulfilled.
1. Conditional distribution of Uı given that Xı, X2,…,Xnı has a mean of zero;
The assumption here is that Yı (the cost of interest bearing debt in this study) from time to time is above and below the population regression line, the average however follows regression line. If the assumption is fulfilled despite the value of the regressors one expects the value of Uı to always have a mean value of zero (Stock & Watson, 2012, p. 238).
2. X1ı,...,Xkı, Yi, i=1,….,n are identically distributed;
29 3. Large outliers are unlikely;
Observation with values significantly outside the normal span is uncommon. The intention with this assumption is to acknowledge that the OLS estimator of the coefficients within the model possibly are sensitive to larger values (Stock & Watson, 2012, p. 238).
4. No perfect multicollinearity;
If perfect multicollinearity occurs within the model it is impossible to compute the OLS estimator (Stock & Watson, 2012, p. 239)
6.5 Descriptive statistics.
Table 3 display the descriptive statistics of the dependent and the independent variables included in the regression.
Table 3. Descriptive statistics
30 (2017, p. 258), was significantly lower that this study at 0.6204. This is surprising considering the average Beta regardless of market always is 1. Possibly it is because mostly low risk companies have a score within the markets the sample was extracted from. The lowest observation for Market Cap. was 4.350782, the highest was 8.147381, the mean was 6.511606 and the standard deviation .6156505. The lowest observation for GDP Growth was -1.426, the highest was 4.457, the mean was 1.718857 and the standard deviation 1.311622. The lowest observation for Key rate was -.0069137, the highest was .0213699, the mean was .0035122 and the standard deviation .0078742.
6.6 Models
6.6.1 Regression model
To assess if CSR has an impact on the interest bearing cost of debt the following regression model have been constructed. The cost of interest bearing debt is the selected measure intended to reflect debt costs and the ESG score the measure intended to reflect the quality of firm’s CSR. The regression will be run with panel data and random effects.
Cost of interest bearing debt = 𝛼 + 𝛽1ESGit + 𝛽2BETAit + 𝛽3FinLevit + 𝛽4ROICit +
𝛽5MarkCapit + 𝛽6KeyRateit + 𝛽7GDPGrowthit + 𝛽8TAit + 𝛽9Industryit + 𝛽10Countryit + eit
Cost of interest bearing debt = Debt cost indicator ESG = ESG Score
Beta = Systematic risk measure FinLev = Financial leverage ROIC = Profitability measure MarkCap = Market Cap Key rate = Macro variable GDP Growth = GDP growth TA = Total Assets
Industry = Industry dummy Country = Country dummy 𝛽 = Beta, coefficient 𝛼 = Alpha, coefficient
6.6.2 Non-linear regression model
There is as previously mentioned a case to be made for examining both linear and non-linear links between ESG and the cost of interest bearing debt. As stated in the overinvestment view section (4.2.2), there are reasons to believe that an CSR investment can be viewed differently depending on individual aspect surrounding each company. Therefore, tests will be made for non-linearity. The test is made by multiplying the independent variable and then apply it in the regression model.
• The effects of ESG on the interest bearing cost of debt can possibly be present after a certain standard of ESG have been met.
