The rise of sustainable investing and its effect on financial performance of high-rated ESG companies

(1)

“The Rise of Sustainable Investing and its Effect on

Financial Performance of High-Rated ESG Companies”

By Fransje Puts, 10658076

MSc Finance – Banking & Regulation

Supervisor: prof. dr. T. Yorulmazer

1 July 2018

Abstract

As testified by Eurosif (2016), Social Responsible Investing (SRI) is gaining ground rapidly, indicating the consciousness of investors to social and environmental issues, in addition to obtaining financial returns. However, no consensus exists whether sustainable investing leads to better financial performance. This study analyses the relationship of sustainability performance, measured by a company’s ESG-rating, and firm value, proxied by Tobin’s Q. Using data on 1,048 European public listed companies in the period 2004-2016, a fixed effects OLS regression as well as an instrumental variable approach supports the view that enhancing ESG performance results in higher firm value. Moreover, when controlling for the industries that are material to the different subcategories Environment, Social and Governance, the results seem to be more pronounced, implicating the importance of materiality. Lastly, the relationship of Environmental performance and firm value appear to be negative in countries providing a higher than average amount of renewable energy subsidies, suggesting that investors in these countries do not incorporate the cost-cutting advantages of subsidies in their stock price expectations.

(2)

Statement of originality

This document is written by Fransje Puts who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1. Introduction

Traditional finance focuses on financial performance and considers the financial sector as isolated from the society and environment it is embedded in. An increasing awareness of climate change, especially since the Paris Climate Agreements in 2015, where the associated countries agreed to diminish CO2-emissions, asks for a

transition to a more sustainable world. To reach this goal, funding is needed, as sustainable technologies are still very expensive. Because finance connects people who are in need of money, with those who have abundant, the financial sector might be the ultimate bridge builder to a greener and more social society (Beslik, 2017). This is contrary to the traditional finance perspective, and is called “Sustainable Finance”, which considers financial, environmental, social, and governance returns jointly (Schoenmaker, 2017).

In line of this goal, the past decade, as testified by Eurosif (2016), Social Responsible Investing (SRI) is gaining ground rapidly, indicating the consciousness of investors to social and environmental issues, on top of obtaining financial returns. Edith Siermann, former manager Sustainable and Responsible Investing at Robeco, explains how society but also regulation asks from the financial industry to behave in a sustainable way, and even expects Sustainable Investing to become mainstream by 2020 (Robeco, 2015).

Although institutional investors are still the main participants in this market, the number of retail investors is increasing significantly. But also mainstream

financial firms, and pension funds are currently incorporating sustainability measures. For example, the Dutch pension fund PME announced recently to stop investing in black coal, to help reducing the CO2-emmissions (2018).

The above named example is one of the many sustainable investment strategies, and is called “negative screening”. This strategy simply excludes assets with very negative impact and is still the most dominant variant, covering 48% of all European professionally managed assets (Eurosif, 2016). Among other sustainable investment strategies are positive selection, ESG (Environment, Social, Governance)-integration, voting, engagement, and impact investing (VBDO, 2016). Whereas the last variant is gaining more and more interest, and aims to quantify and measure the precise impact of projects and firms. Primarily energy efficiency- and renewable energy- firms are connected with this strategy.

(5)

The rise of sustainable investing and its various strategies are partly an outcome of the increasing required disclosure of ESG practices of public listed companies. Twenty years ago, the most basic data about sustainability was

unobtainable, whereas now an increasing number of individual rating agencies rank companies on Environmental, Social and Governance performance. Because of these publicly available rankings, all sorts of investors are able to compile an investment portfolio based on firm’s ESG performance.

The disclosure of sustainability data enables the market to act and profit upon the problem of climate change. Following the theory of the classical economist Adam Smith, pursuing the individual first-best interest would eventually benefit society as a whole (1776). Investors generally see profit opportunities in innovations and new technologies and currently, one of the leading innovations are in renewable energy technologies. In this manner, serving the individual interest of investing in innovative projects would simultaneously profit the whole society.

Accordingly, the New-Keynesian economist Shiller (2013) emphasizes that the financial system of capitalism need to be used for the forces of a sustainable society. He states that the financial sector has to be continually renewed through innovation in order to achieve society’s goals. One of these recent innovations in the financial sector is the emergence of crowdfunding. The market for crowdfunding is growing rapidly, of which many projects have a sustainability motive (VBDO, 2016).

However, no consensus exists whether incorporating ESG practices actually lead to better financial performance, and thus it is not clear whether sustainable investing optimally serves the individual best interest. Various studies about investor characteristics show that a lot of investors believe sustainable investments result in lower returns, and are thus willing to forego higher returns to invest socially and responsibly (Riedl & Smeets, 2017, Morgan Stanley, 2015). This trade-off believe might be a result of the deeply rooted, short-term focused Shareholder Theory, which is still one of the main challenges moving to a sustainable society. Next to the

inconsistent investors beliefs, empirical studies also show ambiguous findings on the effect of sustainability on financial returns (Bassen, Busch & Friede, 2015). With the aim to provide more clarity and additional findings on this topic, the current study will test the relationship of a firm’s sustainability performance, measured by ESG rankings, and market firm value, proxied by Tobin’s Q. Tobin’s Q is a widely used

(6)

measure of market expectations about growth opportunities of a company, and thus captures the investor’s perception how much a stock is worth (Orlitzky, Rynes, & Schmidt, 2003). The line of reasoning used in this research is that if

investors/shareholders regard a higher sustainability level of a firm as a form of intangible value, the enhancement of a firm’s sustainability performance would improve firm value.

Furthermore, to shed some light on how innovations of the financial sector could help moving to a more sustainable society, a case study about a Dutch solar panel crowdfunding organization “Zonnepanelendelen” is included. It also gives some practical insights about the sustainable investor characteristics, the availability of funding, how this funding has evolved over time, how profitable the business is, and to what extent governmental renewable energy subsidies are a necessity. As is stated by the politician and environmentalist Al Gore, the pace of moving to a sustainable society is depended on policy measures like subsidies and tax reductions for sustainable technologies (2018). To test whether this statement is actually true, the effect of subsidies for renewable energy on firm performance is analyzed.

As opposed to most of the previous research, which is merely based on American firms, this empirical study will look at European public listed companies and covers the time period 2004-2016. Because ESG data became available quite recently, especially in Europe, every year of extra data adds value to the existing literature. The ESG ratings are obtained from the ASSET4 Europe Thomson Reuters database, whereas the financial data is retrieved via Datastream. The applied dataset is left with yearly panel data about ESG- and financial performance of around 1,000 companies.

The effect of the overall ESG performance on firm value is tested by conducting both, a Fixed Effects OLS regression method, and an Instrumental

Variables Approach to minimize endogeneity concerns. This research finds consistent support for a positive and significant overall effect of ESG performance on firm value. Moreover, the effects of the subcategories Environment-, Social- and Governance-ratings on firm value are regressed separately on the whole dataset, as well as controlling for different industries which are most related to the respective subcategories. The latter method shows positive results, and appears to be significant for the Environmental- and Social-ratings, indicating the importance of materiality. Lastly, the relationship of the Environmental-ranking and firm value suggest to be

(7)

negative in countries providing a higher than average amount of renewable energy subsidies, which contradicts the expectations that subsidies should lower the costs and hence enlarge financial performance.

The remainder of this study is structured as follows. Section 2, the literature review, explains the main relevant theories, but also discusses the preeminent

empirical findings, and subsequently derives the tested hypotheses. Section 3 outlines a case study about the Dutch crowdfunding start-up “Zonnepanelendelen”. Thereafter, in section 4, the methodology and a description of the data are presented. Section 4 provides the results and its accompanying analysis. Conclusively, a review of the main results, limitations of this research, policy- as well as business implications, and suggestions for future research are provided.

(8)

2. Literature Review

2.1 Theoretical Background

Already explained in the introduction, the last decade, the interest in Sustainable Investing has increased significantly. However, various studies find that a large number of investors believe sustainable investments result in lower returns, and are thus willing to forego higher returns to invest socially and responsibly (Riedl & Smeets, 2017, Morgan Stanley, 2015). As a large number of empirical studies have shown that sustainable investment portfolios perform better or at least not worse than the traditional ones, this investor’s belief might be a result of the deeply rooted Shareholder Theory.

2.1.1. Shareholder Theory

According to the shareholder theory, a company should be served following the aims of its owners or shareholders and is thus mainly focused on maximizing profits. This view is in line with the ideas of the neo-classicist Milton Friedman who states that the only social responsibility of a firm is to optimally use its resources to maximize profits as long as it stays within “the rules of the game” (1962). He argues that the government is responsible for meeting social and environmental goals and to implement these rules of the game for sustainability. When a company would undertake Corporate Social Responsibility (CSR) activities, it could distract firms from spending their resources on the most profitable investments. Also, implementing environmental standards could lead to higher costs and thus to a competitive

disadvantage (Derwall, 2007). Furthermore, engaging in CSR may lead to an agency conflict between shareholders and managers (Jensen & Meckling, 1976). It could give an incentive for managers to over-invest in sustainability practices as it might

improve their reputation, which consequently results in lower returns (Barnea & Rubin, 2010).

As the shareholder approach contains build-in short-term incentives like quarterly reporting and incentive payment schemes (Schoenmaker, 2017),

inconclusiveness exists about the effect of CSR on firm performance. Even though incorporating sustainability practices could lead to higher costs in the short-term, it is able to generate a competitive advantage in the longer run (Hart & Ahuja, 1996). And

(9)

thus, a stakeholder-based, longer-term perspective, could give a non-negative outcome of incorporating sustainability activities.

2.1.2. From shareholder to stakeholder

The stakeholder theory, introduced by Freeman (1984), states that a company should maximize the value of all involved stakeholders; like shareholders, employees, customers, suppliers and the local community. Thus, instead of only maximizing firm value, the goal is to maximize total value. This aim can be reached by implementing a longer-term horizon. Various big corporates are already taking decisive steps to shift away management’s thinking striving for short-term results. As an example, Unilever stopped reporting quarterly profits (Clark, Feiner, & Viehs, 2015).

An extension of the stakeholder theory is the instrumental stakeholder theory, which argues that CSR activities will result in a competitive advantage of the firm and thus leads to an improvement of firm value (Jones, 1995). Both, direct and indirect effects can explain this improvement. In a direct way, engaging in CSR practices could result in enhanced management of a company and leads to risk mitigation of a company in two separate manners. Firstly, company specific risks are mitigated by the avoidance of fines and settlements concerning ESG issues (Clark, Feiner, & Viehs, 2015). Furthermore, external costs will be reduced by anticipating on the possible supply chain disruptions as a result of climate change such as droughts or floods.

Sustainability activities might also in an indirect manner result in financial improvement. Corporate reputation appears to be an important factor of continuous value maximization (Dowling & Roberts, 2002). One aspect of good reputation relates to human capital and job satisfaction. A pleasant work environment ensures that talented workers stay with the company and might improve firm value through employee motivation (Edmans, 2011). Besides, inferior ESG performance could threaten a firm’s reputation from a customer’s perspective. When ESG-related scandals are revealed in the news, it could result in reputational damage. Examples are disclosures about the deplorable working conditions in the factories of Foxconn (Apple’s manufacturer in China) which halved its market capitalization, the still present managerial and engineering damage of the Deepwater Horizon oil spill in the Gulf of Mexico operated by BP in 2010, and the huge outflow of private client assets

(10)

as well as fines of $780 million at the Swiss bank UBS as a result of unauthorized trading (Eccles & Serafeim, 2013).

Other indirect positive sustainability effects on financial performance are process- and product innovation. By implementing long-term sustainability programs, companies are able to make use of their resources completely, efficiently and

effectively. As an example, the last decade Coca-Cola has diminished its water intensity by 20% (Coca-Cola Company, 2013). Also, as sustainability is becoming a market driver (PricewaterhouseCoopers, 2010), innovating sustainable products, which are for instance recyclable, lifetime reliable, or energy efficient, can increase a company’s revenues significantly. Thus, incorporating ESG factors into a company’s sustainability framework could lead to cost cuttings through process efficiency, and an increase in profits via product innovation. Consequently, this should result in margin enhancements (Eccles & Serafeim, 2013).

It seems to be that a growing number of companies are taking these advantages into account, as the average ESG ranking of European firms increased substantially over the past fifteen years.

Graph 1. Average ESG rating of European Graph 2. Average Tobin’s Q of European public listed companies over the years 2004-2016 public listed companies over the years 2004-2016

As is shown in the first graph, the average ESG score increased substantially from 2006 until 2010, which might be a reaction of the Kyoto Protocol that entered into force on 16 February 2005 (UNFCC, 2018). This agreement engaged parties of the

(11)

United Nations Climate Change by setting binding emission reduction targets. Furthermore, the average ESG score decreased a bit in the years 2010-2015. When looking at the second graph, a steep decrease in firm’s Tobin’s Q is visible from 2007. One explanation for this decrease might be the Global Financial Crisis of 2007/2008. Either a delay in the reaction of ESG-rating on the global financial crisis is existent, as a decline is visible from 2009 to 2010, or the decrease in rating was a consequence of the European Debt Crisis starting at the end of 2009. In times of crisis, companies may see sustainable technologies as “luxury products” and hence provide fewer resources to invest in the possibly more expensive sustainable products.

Interestingly, a steep increase in the average ESG score is visible from 2015 to 2016. Along with the current European economic up-phase, the Paris Climate

Agreements, made in 2015, could be a reason for this significant grow. To meet the goals of these Agreements of reducing the CO2-emmissions, companies need to incorporate sustainable business processes. The business sector seems to be fulfilling an increasing job in the realization towards a more sustainable society (Motivaction, 2013). Specifically the big corporations and market leaders play an important role. Simultaneously, the number of sustainable investors is increasing. In the Netherlands for example, the sustainable investment market has increased by 30.6% from 11.5 billion to 15.1 billion in 2015 relative to 2014. This amounts to a market share of 13%. The rise in sustainable investing might be an implication of investors attaching higher (intangible) value to more sustainable companies.

2.2 Empirical Findings

Already since the 1970s, empirical research is done about the topic whether social and responsible firms are able to compete or even outperform the market. While the findings are inconclusive, 90% of the studies show a positive relationship between CSR and financial performance (Bassen, Busch & Friede, 2015). An explanation for these ambigous results might be the use of different proxies for financial performance as well for CSR. Some studies use accounting-based numbers to measure financial performance, like Return on Assets, others use market-based measures, either based on stock performance or on firm value. As the aim of this research is to look at the effect of sustainability ratings of a company on firm value from an investor’s perspective, it is most relevant to look at market-based measures of financial performance. Along with the method used in this study, this section is focused on

(12)

previous empirical findings that look at the effect of sustainability measures on firm value measured by Tobin’s Q.

One of the earlier studies researching the relationship of sustainability on firm performance is the study by Dowell, Hart & Yeung (2000). They identify a statistical relationship between environmental firm standards and firm value of American Multinational Enterprises covering the time period 1997-1999, using t-tests and multiple regression analysis. To address the problem of probable reversed causality, they use a Granger causality method. Their main finding is that higher quality companies, measured by Tobin’s Q, have better environmental standards and thus seem to pollute less. However, no clear causal relationship is found.

Accordingly, the study of Konar & Cohen (2001) researches the relationship of financial and environmental performance of manufacturing companies listed in the S&P 500. They find that inferior environmental performance has a significant

negative effect (both statistically and economically) on the value of intangible assets of S&P500 listed companies. This result is symmetric, which means that firms with a good environmental reputation also have higher valued intangible assets. Again, it is not evident whether this relationship is truly causal.

Jiao (2010) contributes to the existent research by applying a two-stage-least square regression to overcome the problem of endogeneity. The study looks at the relationship of stakeholder welfare and firm value and finds that an increase in 1 of stakeholder value results in an increase of 0.587 in Tobin’s Q, and thus has a positive effect. These positive results are mainly driven by its effects on environmental issues and employee relations.

Furthermore, to evaluate different pathways leading environmental performance to financial returns, both accounting-based as well as market-based measures are used in the study of Derwall (2007). As a measure for environmental performance he uses eco-efficiency data about US public listed companies. He follows a multivariate model to analyze the relationship between environmental performance and Return on Assets. For the relationship of environmental performance and Tobin’s Q, an OLS regression is applied, where Tobin’s Q is measured in levels, logs, and trimmed. The study, covering the sample period 1997-2004, points to a positive relation of financial performance and a strong corporate eco-efficiency policy.

(13)

The research of Marsat & Williams (2011) is most similar with the current study but is based on American data. It analyzes the relation of market value, proxied by Tobin’s Q and CSR, measured by MSCI ESG data. In line with the previous discussed research, they apply an OLS regression where Tobin’s Q is measured in levels, logs, and trimmed, covering the time period 2005-2009. They find a robust negative effect, indicating that either investors are not pricing in positive externalities of CSR in their equity asset valuation approach, or underestimate the influence of CSR performance.

Moreover, a similar time period is studied in the paper of Nishitani and

Kokubu (2011). It examines the influence of a company’s reduction in greenhouse gas emissions on firm value using data of 641 Japanese manufacturing firms. They not only investigate the relation of the reduction in greenhouse gas emission and firm value, but also the effect of the market discipline imposed by shareholders in terms of the reduction in greenhouse gas emissions. Applying a random effect instrumental variable approach, they find support for their hypothesis that companies with a strong market discipline are more prone to diminish greenhouse gas emissions and

consequently more likely to enhance firm value.

2.3 Hypotheses

From the above-described literature, a couple of hypotheses are derived with the aim to answer the research question whether the rise in sustainable investing has a positive effect on firm value.

As the number of sustainable investors is increasing, demand for sustainable stocks is growing, which accordingly leads to an overvaluation of these companies (Marsat & Williams, 2011). Measuring the sustainability level of a company by its ESG (Environment, Social, Governance)-rating and firm value by Tobin’s Q, results in the following hypothesis:

H1: A high ESG-rating has a positive effect on Tobin’s Q

To see whether this effect varies across the different subcategories of ESG, the above-described hypothesis is also tested for Environmental-, Social-, and Governance-ratings separately.

Furthermore, measuring the sustainability level of a company based on three subcategories happens for a reason. Being “sustainable” as a company varies in

(14)

practice hugely across different sectors and industries. Each subcategory might be more or less relevant or “material” among different industries (Eccles & Serafeim, 2013). Firstly, companies operating in polluting industries have a higher incentive to improve their environmental performance to avoid reputational damage but also to mitigate risks related to environmental damage (Reverte, 2012). Subsequently, the second hypothesis is as follows:

H2: The Environmental-rating of a company has a more pronounced positive effect on firm value in emission-sensitive industries

Secondly, social performance is specifically relevant in industries with higher chances of having for example critical labor laws; human rights are more material for

companies making use of low-cost labor in emerging or developing markets than for firms using qualified workers in developed markets. The first explained companies are more inclined to enhance social issues, preventing reputational damage, resulting in the next hypothesis:

H3: The Social-rating has a more pronounced positive effect on firm value in related industries

Thirdly, one of the main issues of the last financial crisis was the lack of an appropriate corporate governance structure among financial companies. Risk

exposures did not reach senior levels of management, while these supposed to be their responsibility (OECD, 2009), but also remuneration packages with high variable pay-schemes led to excessive risk-taking, especially in the market for mortgage-backed-securities. As the financial industry is exposed to a lot of risk, good corporate governance is essential. Hence, it would be interesting to see whether corporate governance has a more pronounced effect in this industry, leading to the subsequent hypothesis:

H4: The Corporate Governance-rating has a more pronounced positive effect on firm value in the financial sector

Lastly, as this research is based on European firms, and each country provides different levels of subsidies for renewable energy, this might influence the effect of the sustainability level of a company on its firm value. When a company for example obtains subsidies to install solar panels on its roof, it will get a higher Environmental

(15)

rating, but at the same time investors expect higher profits as the costs to install these solar panels are lower. As these subsidies are specifically used for renewable energy technologies, only the Environmental-rating is relevant for this hypothesis. The last hypothesis is phrased as follows:

H5: The Environmental rating of a company has a more pronounced positive effect on firm value in countries providing high renewable energy subsidies

2.4 Contribution

As is shown in the literature review, the relationship of CSR and financial

performance has been studied extensively, with a large variation in applied empirical methods. Because the results are inconsistent, on going research is needed to find which sustainability factors are most relevant for investment returns (Clark, Feiner, & Viehs, 2015).

Up until now, studies are mainly focused on American data, whereas the current research studies the effect of sustainability measures on firm performance of European companies. As environmental regulations and subsidies, as well as social issues and governance rules vary across Europe and the United States, this might lead to different results and implications.

Additionally, because of the quite recent availability of ESG data and the fast growing interest in sustainable investing, it is of great relevance to make use of the most recent provided data. This study covers the time period 2004-2016, in which the first effects of the Paris Climate Agreements of 2015 might be incorporated.

Also, because the three different subcategories E, S and G are relevant in different industries, these separate measures are controlled for different related industries, to investigate the materiality of each category across different industries. Previous studies have also incorporated this issue but they only focused on the ESG measure or Environmental subcategory regressing on emission-intensive industries. This research will additionally look at the effect of the other subcategories Social and Governance and their respective relevant industries.

Furthermore, the case study of the crowdfunding start-up

“Zonnepanelendelen” gives some insights whether they believe sustainable investing leads to higher financial returns, and how they think we could move to a sustainable society. According to them, providing subsidies, and taxation on emissions would be important steps in that direction. Therefore, the relationship of firm value and

(16)

provided subsidies for renewable energy in Europe is tested. If this effect turns out to be positive, this might be an implication for policy makers to extend the amount of renewable energy subsidies.

(17)

4. Methodology

4.1 Data description

To measure the sustainability performance of companies in Europe, ESG ratings are obtained from the ASSET4 database. This database is provided and sold by Thomson Reuters to investors and corporations and administers objective data about

Environmental, Social and Governance (ESG) issues which is found on more than 250 key performance indicators. The data provides a ranking on each topic between 0-100% that is based on the performance of a company benchmarked to all other companies in the dataset.

It contains data of listed companies globally and is collected from publicly available information sources. Even though currently most companies present their ESG performance publicly, the ASSET4 database is standardized and thus easy to compare different companies globally.

As this research will be focused on Europe, the ASSET4 Europe dataset is used which gives information about companies in 20 European countries, and currently provides data of around 1,090 European companies. The financial data, needed for the firm value and control variables of this study is obtained via DataStream.

The research covers the time period 2004-2016. Even though the ASSET4 dataset started to provide the ESG rankings from 2002 on, a lot of companies did not report their ESG ranking yet. Also, so far, not enough data is available for the year 2017, and therefore the years 2002, 2003 and 2017 are excluded from this research. After cleaning the data, such as taking care of outliers by winsorizing non-normal distributed variables, and removing firms with too many missing data points

(companies who have less than two years of data available are removed), the dataset is left with yearly panel data about ESG- and financial performance of 1,048

companies. The advantages of using panel data is to diminish omitted variables bias, as well as increasing the precision of the estimates, simply by using more data (Stock & Watson, 2007).

(18)

4.2 Model

The above-described data allows me to test the relationship of ESG-rankings on firm value, and looks in its simplest form accordingly:

𝑓𝑖𝑟𝑚 𝑣𝑎𝑙𝑢𝑒!" = 𝛼! + 𝛽!𝐸𝑆𝐺!"+ 𝛽!𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠!"+ 𝜀!

The main dependent variable, firm value, will be measured by a Tobin’s Q ratio, as this appears to be an adequate market-based measure for financial

performance (Tobin, 1969; Dowell, Hart, & Yeung, 2000; Marsat & Williams, 2011). Tobin’s Q is a commonly used proxy for firm value in the finance field and is a ratio of the market value of a firm relative to its replacement cost of capital. Intuitively, if a firm has a Q bigger than one, it generates more value using a specific amount of capital than another company would generate with this same amount of capital. Tobin’s Q is a forward-looking measure and relies on the investor’s perception how much a stock is worth (Orlitzky, Rynes, & Schmidt, 2003). Investors consider Q as the intangible value of a company and thus capture the value investors appoint to ESG policies (Derwall, 2007). As this study aims to research whether the increase in

sustainable investing results in a higher firm value, Tobin’s Q is the appropriate measure. In the existing literature there is not one specific functional form to calculate Q, but in this study the following formula is used:

𝑇𝑜𝑏𝑖𝑛′𝑠 𝑄 =𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 𝐵𝑜𝑜𝑘 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠

As the market value of assets is not an observable variable, the specification of Marsat & Williams (2011), is applied. The market value of assets is calculated by adding book value of total assets to market value of common stock and subtracting the book value of common stock.

The main independent variable in this study is the ESG-rating of European companies. To test the different hypotheses, both the equal-weighted aggregate ESG-score as provided by ASSET4, as well as the three sub rankings, Environmental, Social, and Governance, are tested.

To avoid omitted variable bias, based on previous studies and correlation of variables, controls are included that affect Tobin’s Q but are also related to the sustainability standards of a company. These are commonly used firm characteristics related to a company’s capital structure and (intangible) assets. Also, to reduce endogeneity problems, cross-sectional company fixed effects (𝛼_!) controlling for variables that stay the same over time but vary among companies, year fixed effects

(19)

(𝜇_!) to control for variables that are similar across firms but change over time, but also year-industry paired fixed effects (𝛾_!") will be included. The tested regression equation looks as follows:

𝑇𝑜𝑏𝑖𝑛!_{𝑠 𝑄}

!" = 𝛼!+ 𝜇!+ 𝛾!"+ 𝛽!𝐸𝑆𝐺!"+ 𝛽!𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒!"+ 𝛽!𝑆𝑖𝑧𝑒!"

+ 𝛽!𝐶𝑎𝑝𝑒𝑥𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦!"+ 𝜀!

The first control variable Leverage, is measured by the long-term debt of a firm, relative to its total assets. This measure, also used by many others in this field, (Waddock & Graves, 1997; Derwall, 2007; Jiao, 2010; Marsat & Williams, 2013) appeared to have a negative effect on firm value. As the probability of financial distress increases with the leverage ratio, and thus the amount of risk a company bears increases, this measure is negatively related with financial performance. Moreover, ESG is also expected to be negatively correlated with leverage; the higher the ESG performance, the lower the level of risk as a result of better management-skills and lower costs related to fines or lawsuits (Nordea, 2017).

The second control variable, Size, is expected to affect both, firm value and ESG performance and is calculated as the natural logarithm of total assets. This measure appears to be one of the most frequently used proxy for firm size (Orlitzky, Rynes, & Schmidt, 2003). Although the findings in the existing literature on the relationship of firm size and financial performance are inconclusive, one way of reasoning might be that bigger firms have stronger competitive competences and thus higher financial returns. On the other hand, investors might see bigger growth

opportunities in smaller firms, which is also an important component of Tobin’s Q. The relationship of ESG rating and size is expected to be positive. An explanation might be that larger firms have more financial resources available to spend on sustainability issues (Waddock and Graves, 1997).

Finally, the amount of R&D expenses and Capital Expenditures suggest to have a positive relationship with financial performance as it is an indication of higher innovation expenditures, leading to competitive advantages which could consequently result in higher profits (Dowell, Hart & Yeung, 2000). Moreover, higher R&D- and Capital- Expenditures are also expected to be positively related with ESG

(20)

performance, as R&D- and capital-intensive companies are more dependent on their reputation and human capital, and are therefore more inclined to invest in stakeholder welfare (Jiao, 2010). As firms are not obliged to report their R&D expenses, the data obtains a lot of missing values and hence this variable will only be used as a control for the robustness checks.

Other commonly used control variables in this field of study, like Return on Assets and the sales growth of a company, were also used in the current research, but appeared to have insignificant correlation with both Tobin’s Q and the ESG-ranking, and are hence left out.

4.3 Method

Hypothesis 1

To test the relationship of ESG performance on firm value, a Fixed Effects OLS regression method is applied. The concern of omitted variables, which is especially apparent when the main independent variable is a choice variable (in this case the ESG-ranking), is partly addressed by the inclusion of fixed effects. The following equation is specified to test whether ESG performance has a positive effect on firm value:

!" = 𝛼!+ 𝜇!+ 𝛾!"+ 𝛽!𝐸𝑆𝐺!"+ 𝛽!𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒!"+ 𝛽!𝑆𝑖𝑧𝑒!"

+ 𝛽_!𝐶𝑎𝑝𝑒𝑥𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦_!"+ 𝜀_! (1)

Where the firm value (Tobin’s Q) of firm 𝑖 in year 𝑡 is the main dependent variable and is determined by the main independent variable, the overall ESG-ranking, and

control variables, the long-term debt to assets ratio (Leverage), the log of its total

assets (Size), the amount of capital expenditures relative to its total sales (CapexIntensity), and firm (𝛼_!), year (𝜇_!), and industry-year (𝛾_!") fixed effects. Because the fixed effects estimator is quite restrictive and can deliver biased

estimators in case of measurement error (Hsiao, 2003), a Pooled OLS regression, as well as regressions including different fixed effects individually, but also company-, year-, and paired industry-year fixed effects all together are carried out.

Furthermore, it is not fully clear in what direction the causality goes and hence the concern of simultaneous causality exists. It could both be true; that a higher

(21)

ESG-ranking results in higher firm value, but also that higher firm value leads to a better ESG performance. If reversed causality holds, OLS estimators might be biased. To overcome these sorts of endogeneity problems, but also to address further concerns of omitted variable bias, an Instrumental Variables Regression is applied. This method is commonly used to obtain a consistent estimator of the unknown coefficients when the main independent variable is correlated with the error term (Stock & Watson, 2007). For this method, an instrument needs to be found, which is correlated with the main independent variable (instrument relevance), but is not correlated with the error term (instrument exogeneity). Following the approach of Cheng, Ioannou, & Serafeim (2013), the subsequent two instruments are used; the average ESG-ranking for each country-sector pair, and the average ESG-ranking for each country-year pair. The rationale behind these instruments is that a firm’s ESG performance is affected by the ESG performance of other companies within the same country-sector set, as well as by the ESG performance of other companies in the same country over time, but is not related with firm value. After testing for instrument relevance and instrument

exogeneity, a Two Stage Least Squares method is used by regressing the following equations: First Stage: 𝐸𝑆𝐺_!" = 𝛿_! + 𝜇_!+ 𝜑_! + 𝛽_!𝑍 + 𝛽_!𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_!" + 𝛽_!𝑆𝑖𝑧𝑒_!"+ 𝛽_!𝐶𝑎𝑝𝑒𝑥𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦_!" + 𝜀! (2) Second Stage: 𝑇𝑜𝑏𝑖𝑛!_{𝑠 𝑄} !" = 𝛿! + 𝜇!+ 𝜑! + 𝛽!𝐸𝑆𝐺!"+ 𝛽!𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒!" + 𝛽!𝑆𝑖𝑧𝑒!" + 𝛽_!𝐶𝑎𝑝𝑒𝑥𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦_!"+ 𝜀_! (3)

Where in the first stage the ESG-ranking of firm 𝑖 in year 𝑡 is a function of the instruments Z, the same as previously used control variables, and country- (𝛿_!), year- (𝜇_!), and industry- (𝜑_!") fixed effects. The second stage is again Tobin’s Q

determined by the ESG measure, previously used control variables, and country-, year- and industry fixed effects.

Hypothesis 2, 3, & 4

The following three hypotheses test whether the different subcategories have a more pronounced effect in their related industries. Whether one of the subcategories

(22)

significantly affects a firm’s ability to create long-term shareholder value depends on the industry the company operates in. To help firms determine their material ESG concern, The Sustainability Accounting Standards Board (SASB) devised a

framework, which displays for each industry and sector its ESG materiality. Based on this materiality framework, the related industries for the different subcategories are chosen.

For the second hypothesis, following the method of Reverte (2012), an industry-sensitivity dummy is included which is equal to 1 if the company operates in an emission intensive industry, and equal to 0 otherwise. This dummy variable

interacts with the Environmental-ranking to take into account the effect of the related industry.

Table 1. Overview Emission-Intensive versus Emission-neutral industries Emission-intensive

industries Industry code

Emission-neutral

industries Industry code

Oil & Gas 1 Consumer Services 5000

Basic Materials 1000 Telecommunications 6000

Industrials 2000 Financials 8000

Consumer Goods 3000 Technology 9000

Health Care 4000

Utilities 7000

And thus the tested regression equation looks as follows:

𝑇𝑜𝑏𝑖𝑛′𝑠 𝑄_!" = 𝛼_! + 𝜇_!"+ 𝛽_!𝐸_!"+ 𝛽_! 𝐸_!" ∗ 𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝐼𝑛𝑠𝑒𝑛𝑡𝑖𝑣𝑒 + 𝛽_!𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_!" + 𝛽!𝑆𝑖𝑧𝑒!"+ 𝛽!𝐶𝑎𝑝𝑒𝑥𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦!"+ 𝜀! (4)

Where the firm value (Tobin’s Q) of firm 𝑖 in year 𝑡 is the main dependent variable and is a function of the main independent variable, the Environmental-ranking, the same as previously explained control variables, and company- (𝛼_!) and year- (𝜇_!) fixed effects.

As stated by the third hypothesis, the subcategory Social also expects to have a positive effect in its related industries. Following the materiality framework of SASB, the materiality for the Social subcategory is not as clear as the Environmental

(23)

subcategory. However, materiality of the Social measure seems to be very similar as the materiality of the Environmental measure, and thus also in this regression

specification an interaction term of the Social-ranking and the emission-intensive industry dummy is included. Subsequently, the following regression specification is tested:

𝑇𝑜𝑏𝑖𝑛′𝑠 𝑄_!" = 𝛼_!+ 𝜇_!+ 𝛽_!𝑆_!"+ 𝛽_! 𝑆_!"∗ 𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝐼𝑛𝑠𝑒𝑛𝑠𝑡𝑖𝑣𝑒 + 𝛽_!𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_!" + 𝛽_!𝑆𝑖𝑧𝑒_!" + 𝛽_!𝐶𝑎𝑝𝑒𝑥𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦_!" + 𝜀_! (5) Where the firm value (Tobin’s Q) of firm 𝑖 in year 𝑡 is the main dependent variable and is determined by the main independent variable, the Social-ranking, the same as previously explained control variables, and company- (𝛼_!) and year- (𝜇_!) fixed effects.

The fourth hypothesis expects a positive effect of the Governance

ranking in the financial sector, and thus a financials dummy is included which is equal to 1 if the company operates in the financial sector, and equal to 0 otherwise. Though, the materiality map of SASB suggests that other industries might be relevant too for the Governance subcategory. But also the materiality of this subcategory is not as clear as the materiality of the Environmental subcategory, and thus I will stick to the financials industry as most relevant industry, based on theory. The financials industry dummy variable is interacted with the Governance ranking to see what the effect of the related industry is on the subcategory Governance. This results in the subsequent regression equation:

!" = 𝛼! + 𝜇!+ 𝛽!𝐺!" + 𝛽! 𝐺!"∗ 𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙𝑠 + 𝛽!𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒!"+ 𝛽!𝑆𝑖𝑧𝑒!"

+ 𝛽_!𝐶𝑎𝑝𝑒𝑥𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦_!"+ 𝜀_! (6) Where the firm value (Tobin’s Q) of firm 𝑖 in year 𝑡 is the main dependent variable and is determined by the main independent variable, the Governance-ranking, the same as previously explained control variables, and company- (𝛼!) and year- (𝜇!)

fixed effects.

Hypothesis 5

The last hypothesis tests whether the subcategory Environment has a more

pronounced effect on firm value in countries issuing higher amounts of renewable energy subsidies. Based on the data provided by the Council of European Energy

(24)

Regulators in the report “Status Review of Renewable Support Schemes in Europe” (2017), the level of renewable energy subsidies for each country is examined. The provided subsidies data only covers the years 2014 and 2015, so the regression for this hypothesis is regressed on the data for 2014-2015 only. Furthermore, a couple of countries, studied in the rest of this research didn’t provide subsidy data and therefore they are removed from this sample. These countries are the Netherlands, Turkey, and Switzerland. As there is no dataset available that provides information on how much subsidies each subsequent company in the dataset attracts to invest in sustainable technologies, there will be only controlled for the height of subsidies for each country in total. A dummy variable is included which is equal to 1 when the country provides above weighted average level of support for both years, and 0 otherwise. In the Appendix, a table is displayed that shows which countries issue above average subsidies, and which countries issue below average subsidies. To measure the effect of Environmental performance on firm value in countries issuing higher amounts of renewable energy subsidies, the subsidy dummy variable is interacted with the Environmental rating. And thus, the tested regression equation is as follows:

!" = 𝛽!𝐸!"+ 𝛽! 𝐸!"∗ 𝐻𝑖𝑔ℎ𝑆𝑢𝑏𝑠𝑖𝑑𝑦 + 𝛽!𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒!"+ 𝛽!𝑆𝑖𝑧𝑒!"

(25)

5. Descriptive Statistics

Table 2 displays the summary statistics for both the dependent and independent variables. Based on these summary statistics as well as on the histograms, variables are corrected for outliers. Through trial and error, Tobin’s Q is winsorized at the 99th percentile, whereas Capital Expenditures- and Research & Development Intensity are winsorized at the 95th percentile, to reach a smooth distribution. The other variables used in this study, already exhibited a close to normal distribution, and are therefore not winsorized.

Table 2. Descriptive Statistics

Variables _Mean _{St. Dev.} _Min _Max _Obs

Tobin's Q 1.96 1.06 0.95 7.00 12,280 ESG 64.21 29.51 2.51 98.33 10,476 E 63.83 29.36 8.53 97.5 10,476 S 65.30 28.37 3.47 98.88 10,476 G 55.94 27.42 1.27 97.68 10,476 Size 15.55 2.12 7.43 23.15 12,887 Leverage (%) 19.03 18.23 0 269.78 12,879 Capex/Sales (%) 7.91 11.02 0.15 44.48 12,662 R&D/Sales (%) 4.77 10.31 0 77.84 5,055

The mean ESG-ranking is 64.21, and for the subcategories E, S and G 63.83, 65.30, 55.94 respectively. The average Tobin’s Q varies between 0.95 and 7.00 for the various firms, and is on average 1.96. This means that a lot of firms in the dataset are overvalued, as the companies’ stock is more expensive than its replacement costs of assets. The averages for the firm characteristics are 19.03% for the leverage ratio, 15.55 for the size, 7.91% for capital expenditures intensity, and 4.77% for Research & Development intensity. The variable R&D intensity has compared to the other

variables a way lower number of observable data as companies are not required to report this data.

In Table 3 a pairwise correlation matrix is provided. Contradicting the main

hypothesis, there appears to be a significant negative relationship between Tobin’s Q and ESG performance. Furthermore, against the findings of several previous studies (Waddock and Graves, 1997; Jiao, 2010), but along with the findings of Dowell, Hart & Yeung (2000) the size of the company suggest to be negatively correlated with

(26)

Tobin’s Q. Additionally, the relationship between capital expenditures and Tobin’s Q appears to be negative, opposing the view that higher capital expenditures eventually leads to a higher firm value (Jiao, 2010). Lastly, the correlation matrix shows a positive relationship between leverage and ESG performance, which contradicts previous findings (Nelling & Webb, 2009). The remaining correlations between Tobin’s Q, ESG performance and the independent variables are significant and consistent with other studies.

Table 3. Pairwise Correlation Matrix

Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) 1. Tobin's Q 1.000 (0.0000) 2. ESG -0.0688 1.0000 (0.0000) (0.0000) 3. E -0.1222 0.8409 1.0000 (0.0000) (0.0000) (0.0000) 4. S -0.1414 0.8790 0.7633 1.0000 (0.0000) (0.0000) (0.0000) (0.0000) 5. G -0.0725 0.6594 0.3848 0.4441 1.0000 (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) 6. Size -0.3165 0.4202 0.4713 0.4531 0.1223 1.0000 (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) 7. Leverage -0.3008 0.0934 0.0870 0.0848 0.1356 0.1532 1.0000 (0.0000) (0.0681) (0.0030) (0.0010) (0.0000) (0.0452) (0.0000) 8. Capex/Sales -0.1172 0.0566 0.0460 0.0774 0.0277 0.1520 0.3331 1.0000 (0.0000) (0.0000) (0.0002) (0.0000) (0.5244) (0.0055) (0.0000) (0.0000) 9. R&D/Sales 0.3084 -0.1691 -0.1810 -0.1398 -0.0414 -0.1971 -0.1762 -0.0927 1.0000 (0.0000) (0.0000) (0.0000) (0.0000) (0.0016) (0.0000) (0.0000) (0.2757) (0.0000)

(27)

6. Results

Before running the different regressions to test the hypotheses, a couple of tests need to be carried out. Firstly, to test whether heteroskedasticity exists in the model, the modified Wald-test is executed. As can be seen in Appendix 4, the null-hypothesis that no heteroskedasticity exists is rejected, and thus robust standard errors are included. Furthermore, the Wooldridge test for autocorrelation in panel data is applied. Displayed in Appendix 5, the null-hypothesis of “no serial correlation” is rejected. Not correcting for autocorrelation could lead to too low standard errors (Stock & Watson, 2007). Hence, clustered standard errors at the firm level are used throughout the analysis to diminish serial correlation within a company over time.

ESG performance and its effect on Firm Value: Fixed Effects OLS

As the fixed effects estimator is quite restrictive and can deliver biased estimators in case of measurement error (Hsiao, 2003), Table 4 displays a Pooled OLS regression, as well as regressions including different fixed effects individually, but also company-, year-company-, and paired industry-year fixed effects all together.

Table 4. Regression Results: Effect of ESG-performance on Tobin’s Q

The dependent variable is firm value measured by Tobin’s Q. The main independent variable is the sustainability level of a company proxied by its ESG-ranking. The ESG-ranking is controlled by size, leverage, capital expenditures intensity, and firm-, year-, and year-industry fixed effects. Regressions 1, 2, 3, 4, and 5 display a pooled OLS without fixed effects, with firm fixed effects, with time fixed effects, with firm and time fixed effects, and with firm, time, and year-industry fixed effects respectively. Standard errors are clustered at the firm level. The observed time period is 2004-2016

Variables ₍₁₎ ₍₂₎ ₍₃₎ ₍₄₎ ₍₅₎ ESG 0.0028*** 0 .0006 0.0011** 0.0011* 0.0010* (0.0003) (0.0007) (0.0006) (0.0006) (0.0006) Size -0.2225*** -0.2882*** -0.2523*** -0.2771*** -0.2877*** (0 .0048) (0 .0485) (0.0274) (0.0550) (0.0578) Leverage -1.1164*** -0.9344*** -0.8735*** -0.8384*** -0.7059*** (0.1020) (0.2037) (0.1872) (0.1983) (0.1940) Capex/Sales -0.2984*** 0 .3402* -0.3141** 0 3013* 0.2888* (0 .0890) (0 .1833) (0.1387) (0.1839) (0.1775) Cons 5.5390*** 6.6323*** 5.9743*** (0 .0860) (0.7341) (0.4100)

Firm FE No Yes No Yes Yes

Time FE No No Yes Yes Yes

Time-Ind FE No No No No Yes

𝑅! _0.2026 _0.1883 _0.2121 _0.8178 _0.8312

N 10,245 10,245 10,245 10,214 10,214

Standard errors are in parentheses and in italic .*, **, *** indicating significance at the 10%, 5%, and 1% respectively.

(28)

As stated by the first hypothesis, ESG performance is expected to have a positive effect on firm value. The first regression, the pooled OLS regression, shows that if the ESG measure increases by 1, Tobin’s Q increases by 0.0028 and appears to be

significant at the 1% significance level. Including only firm fixed effects, ESG performance remains positive, though not significant. By the inclusion of only year fixed effects, ESG becomes significant again now at the 5% level, and stays positive. Specification 4 and 5, includes firm- and year- fixed effects as well as firm-, year, and industry-year fixed effects, where ESG performance appears to have a positive and significant effect on Tobin’s Q at the 10% significance level. Given these findings, the null-hypothesis that ESG performance has no effect on firm value, can be rejected. This overall positive effect suggests that the more sustainable a company is, the higher value investors attach to a company.

The control variables leverage and size remain strongly significant across the different regression specifications, and show a negative relationship with Tobin’s Q. An increasing amount of studies find a negative relationship of size and Tobin’s Q (Dowell, Hart, & Yeung 2000), which is in line with the above displayed results. The control variable Capital Expenditure Intensity, which represents potential investment opportunities, displays different signs and significance levels over the various

regression specifications. This variable appears to be negative in the specifications not controlling for firm fixed effects. Because capital expenditures vary broadly across firms and industries, the estimates of regressions not controlling for these effects might be biased. Hence, a positive relationship of Tobin’s Q and capital expenditures, shown in regressions 2, 4, and 5 in Table 3, are most likely.

ESG performance and its effect on Firm Value: Instrumental Variables Regression

The displayed table below shows the results from the first and second stage of the instrumental variables regression. A couple of postestimation tests are carried out to see whether the equation is identified, how relevant and strong the instruments are, and whether the instruments are exogenous. Firstly, underidentification is tested by the Lagrange multiplier (LM) Kleibergen-Paap rk statistic, and shows that the model is always identified. Secondly, weak identification is tested by the Kleibergen-Paap rk F-statistic. This F-statistic is bigger than 10 (following the theory of Stock & Watson (2007) using this number as a simple rule of thumb), indicating that the instruments

(29)

are relevant and strong. Lastly, overidentification is tested by the Hansen’s J statistic. For this test, the null-hypothesis states that the instruments are exogenous, and thus if it is significant, it is an indication that the instruments are not exogenous. As the test statistic in the table below is insignificant and has a high p-value, the null-hypothesis of exogenous instruments is not rejected. Thus, according to the applied tests, the instruments satisfy the conditions for a valid instrument.

The coefficient on the ESG measure is positive and significant, implying that the exogenous component of the ESG-ranking positively affects firm value. These results are similar to the Fixed Effects OLS estimation. Though, the magnitude and

significance of the ESG coefficient (0.0043 at the 5% significance level) are larger in Table 5. Regression Results Instrumental Variables Approach

In the first stage, the dependent variable is the ESG-ranking. The Country-Sector Mean ESG and Country-Year Mean ESG are the instruments for the endogenous variable ESG-ranking. Both stages are controlled by the variables size, leverage, capital expenditures intensity, and country-, industry-, and year fixed effects. Standard errors are clustered at the firm level. The observed time period is 2004-2016.

Variables _{(1) First Stage} _{(2) Second Stage}

ESG 0.0043**

(0.0021)

Country-Sector Mean ESG 0.7442***

(0.0296)

Country-Year Mean ESG 0.7160***

(0.0658) Size 6.2070*** -0.2382*** (0.3546) (0.0254) Leverage -1.9269 -1.1367*** (3.1963) (0.2586) Capex/Sales 1.5266 0.0568 (5.4895) (0.2143)

Country FE Yes Yes

Industry FE Yes Yes

Year FE Yes Yes

(Centered) 𝑅! _0.5336 _0.3464 N 10,245 10,245 Kleibergen-Raap rk Wald F 395.029 Statistic Kleibergen-Raap rk LM 233.373 Statistic (0.0000) Hansen J Statistic 0.659 (0.4169)

(30)

the Instrumental Variable Regression compared to the Fixed Effects OLS regression. Also from these results it appears that investors have higher financial expectations about high-rated ESG companies. The control variables size and leverage remain negative and significant. Furthermore, the coefficient of capital expenditures intensity is positive, however not significant.

E-, S-, and G- performance and its effect on Firm Value in related industries

In the regression specifications 1, 3 and 5 of Table 6, the subcategories E, S and G are regressed separately on the whole sample. Strangely, they all show a negative

relationship with Tobin’s Q, while the ESG measure consistently presents a positive relationship with Tobin’ Q, as is shown previously. Though, the coefficients of the subcategories are not significant, and thus no conclusions can be drawn. These negative and insignificant coefficients might be caused by the fact that each

subcategory is to a greater or lesser extent relevant in different industries as stated by the hypotheses 2, 3 and 4.

More specifically, according to the second hypothesis, the Environmental ranking should have a positive effect on firm value in emission-intensive industries. Displayed in regression 2 of Table 6 the Environmental coefficient remains negative and is significant at the 10% level. But the interaction variable of the emission-intensive industry dummy with the Environmental subcategory appears to be positive and significant at the 5% significance level. This indicates that an increase in 1 of the Environmental-rating, decreases Tobin’s Q by 0.0018 in less-emission-intensive industries, but increases Tobin’s Q by 0.0013 (-0.0018 + 0.0031) in emission-intensive industries. And thus the environmental subcategory appears to be positive and significant in its related industry, supporting the second hypothesis. This means that investors attach higher value to sustainability issues that are material for a specific industry.

Regression 4 of Table 6 specifies the results for hypothesis 3 and expects a positive effect of the Social subcategory on firm value in emission-intensive

industries, including an interaction variable of the polluting-industry dummy with the Social-ranking. Again, the Social coefficient remains negative and is significant at the 10% level. Still, the interaction variable of the emission intensive industry dummy

(31)

with the Social subcategory is positive and significant at the 5% significance level. As before, this suggests that also this subcategory has a negative effect on the whole sample, but a positive effect on the emission-intensive sample. An increase in 1 of the Social-ranking, decreases Tobin’s Q by 0.002 in less-emission intensive industries, but increases Tobin’s Q by 0.0014 (-0.0020+0.0034) in polluting industries. These results support hypothesis 3, and rejects the null-hypothesis that Social performance does not have an effect on Tobin’s Q in its related industry. Same as in the previous hypothesis, these results mean that investors attach higher value to sustainability issues that are material for a specific industry.

Table 6. Regression Results: Effect of the subcategories E, S, and G separately on firm value

The dependent variable is firm value measured by Tobin’s Q. The main independent variable of regression 1 & 2 is the subcategory Environmental-ranking, of regression 3 & 4 Social-ranking, and of 5 & 6 Corporate Governance-ranking. Each regression is controlled by the variables size, leverage, capital expenditures intensity, and fixed effects. Regression 1, 3 and 5 show the results for the subcategories without controlling for the relevant industries. In regression 2 an interaction variable is used of an intensive industry dummy variable and the subcategory Environment. In regression 4 an interaction variable is used of an emission-intensive industry dummy variable and the subcategory Social. In regression 6 an interaction variable is used of the financials industry and the subcategory Governance. Standard errors are clustered at the firm level. The observed time period is 2004-2016.

Variables ₍₁₎ ₍₂₎ ₍₃₎ ₍₄₎ ₍₅₎ ₍₆₎ E _-0.0005 _-0.0018* _(0.0007) _(0.0010) S -0.0006 -0.0020* (0.0007) (0.0011) G -0.0008 -0.0009 (0.0006) (0.0009) Emission-intensive 0.0031** 0.0034** Industry*E-S (0.0013) (0.0014) Financials Industry*G 0.0011 (0.0011) Size -0.2787*** -0.2604*** -0.2781 -0.2609*** -0.2781*** -0.262*** (0.0586) (0.0322) (0.0581) (0.0594) (0.0576) (0.0592) Leverage -0.7128*** -0.8385*** -0.7113*** -0.8428*** -0.7112*** -0.8288*** (0.1935) (0.2058) (0.1938) (0.2078) (0 .1935) (0.2087) Capex/Sales 0.2801 0.1161 0.2801 0.1137 0.2755 0.1009 (0.1770) (0.1856) (0.1768) (0.1867) (0.1778) (0.1874)

Firm FE Yes Yes Yes Yes Yes Yes

Time FE Yes Yes Yes Yes Yes Yes

Time-Ind FE Yes No Yes No Yes No

𝑅! _0.8311 _0.8262 _0.8311 _0.8262 _0.8311 _0.8258

N 10,214 9,609 10,214 9,609 10,214 9,609

(32)

The fourth hypothesis is displayed by specification 5 of Table 6 and shows the results for the Governance subcategory on firm value in the financials industry, including the interaction variable of the financials industry with the Governance performance. Again, the subcategory Governance appears to be negative, and the interaction of the financials industry dummy with the Governance ranking is positive, but both

coefficients are not significant. This lack of significance might implicate that also other industries besides the financials industry are relevant for Governance

performance. As the coefficients lack significance, no conclusions can be drawn about the magnitude of the overall direction.

Environmental performance and its effect on firm value in high-subsidized countries

In the first regression of Table 7, the results are shown of the effects of the

Environmental-ranking and the previously used control variables on firm value. As for this hypothesis only the years 2014-2015 are observed, the number of observations reduced significantly from 10,214 to 1,644, and therefore it is important to first

analyze the effect of Environmental performance on firm value without including the subsidy dummy. The Environmental-rating shows a positive sign, is not significant. However, the control variables are significant, and show the same sign and magnitude as in the previous regression specifications, showing the robustness of these variables.

Regression 2 of Table 7 displays the results with the inclusion of the

interaction variable of the Environmental-ranking with the high-subsidy dummy. The Environmental-ranking appears to be positive and significant at the 10% significance level. This result means that if the Environmental-ranking goes up by 1, Tobin’s Q increases by 0.0015 in countries that provide lower than average renewable energy subsidies, but decreases by 0.0019 (0.0015-0.0034) in countries that provide higher than average renewable energy subsidies. This result is contrary to what is expected by hypothesis 5, which states that Environmental performance has a more pronounced positive effect on firm value in countries providing higher renewable energy

(33)

Table 7. Regression Results: Effect of Environmental-rating on firm value of high-subsidized countries

The dependent variable is firm value measured by Tobin’s Q. The main independent variables are the environmental ranking and the interaction of the environmental ranking with the subsidy dummy. Robust standard errors are used, to account for heteroskedasticity. Due to lack of data, the

Netherlands, Turkey, and Switzerland are excluded from the dataset. The subsample covers the time period 2014-2015.

Variables (1) (2) E 0.0009 _0.0016* (0.0009) (0.0009) Subsidy*E -0.0016*** (0.0010) Size -0.2273*** -0.2265*** (0.0197) (0.0203) Leverage -0.9304*** -0.9315 (0.1818) (0.1823) Capex/Sales -0.1402 -0.1590 (0.1552) (0.1573) Constant 5.7450*** 5.7316*** (0.3319) (0.3397) 𝑅! _0.1693 _0.1822 N 1,652 1,644

Countries that issue an above average amount of subsidies are Germany, Belgium, Greece, Czech Republic, and Italy (Appendix 6 displays a table for both, above-average and below average provided subsidies in each country). This set of countries seems like a good representation of the great variety in Europe in terms of GDP, political structure and climate, and thus it is less likely that these factors influence the results. An explanation for the negative results is that the above named countries attain a relatively low share of their energy usage from renewable energy sources (Eurostat, 2018). Various researches find a negative effect of renewable energy policies on the creation of renewable energy (Kleßmann, 2012, Scholtens & Veldhuis, 2015). The interest in sustainability might be lower in these countries and accordingly investors may have lower growth expectations about sustainable

technologies, which eventually result into a lower Tobin’s Q. Though, the above displayed results should be taken with caution, which is further explained in the discussion/conclusion.

The rise of sustainable investing and its effect on financial performance of high-rated ESG companies