Are firms rewarded for their environmental performance? – an empirical examination of Swedish public firms

1

Are firms rewarded for their environmental performance? –

an empirical examination of Swedish public firms

Master Thesis Nikolay Marinov University of Groningen: S2350750 Uppsala University: 890507-P513 Supervisor: Dr. H. Gonenc Co-assessor: Dr. H.W.J. Vrolijk 10.01.2014

University of Groningen University of Uppsala

MSc International Financial Management MSc Business and Economics

2

Abstract

Many academicians claim that good environmental performance improves firms’ economic performance. Thus, the win-win reasoning, saying that corporate competitiveness is increased, if firms improve their environmental performance, justifies this study’s objectives. In this research multiple regression models and data from a sample of 44 Swedish public listed firms are used to study the relationship between greenhouse gas emissions and financial performance. The results show that finacial performance is not correlated to greenhouse gas emissions. This finding is contrary to most of the previous studies conducted on this topic. Furthermore, the results show that there is no correlation between finacial performance and companies that operate in emission-intensive industries, as well as between increase in emissions and financial performance. Thus, it is implied that there are no incentives for Swedish firms to invenst in emission reduction.

3 1. Introduction.

Porter and Van der Linde (1995) are among the first to challenge the traditional view of economists, who state that profit maximization determines production and lower profits will be made, if companies make additional efforts to improve their environmental performance. Their new point of view suggests that among many other things a better environmental profile of companies can improve market shares and financial performance.

In the past many researches have been made to test the relationship between corporate social performance (CSP) and corporate financial performance (CFP). However, no definitive answer has been given to the question “Does it pay to be green”. The level that a company is “green” is measured by its CSP, which in turn is measured differently in different studies, e.g. emission variation or eco-efficiency. However, in the last years greenhouse gas (GHG) emissions prevail as the most used variable to measure CSP (Delmas and Nairn-Birch, 2011; Busch and Hoffmann, 2011). GHG is any of the gases, e.g. carbon dioxide, methane, etc, that absorb solar radiation, trap heat in the atmoshpere, and thus contribute to the greenhouse effect. With the advance in technologies collecting GHG emission data is not something impossible. Hence, many academicians use these data to test the CSP-CFP relationship. Despite the fact that GHG emissions are an important part of CSP, they are different from other CSP measurements. GHG emissions affect both business environment and operations of companies (Busch and Hoffmann, 2011), and are responsible for the negative climate change, which attracts increasing attention from firms’ stakeholders and media. The business environment of a company is affected by GHG emissions in a sense that there might be changes in its bottom line, when a price is introduced, such as in the Emission Trading Schemes (ETS). This, in turn, influences management’s decisions and could bring up new opportunities or business risks. GHG emissions differentiate from other CSP measures as well by the fact that they are not specific to a particular region, but impact every company, sector and country. Inevitably with the increasing media coverage and social awareness of this environmental problem, researchers are conducting new studies, in which they try to explain this relationship. So far the results are mixed. Most of the previous studies show that there exist actuall benefits for companies to engage in CO2 reduction

efforts (Orlitzky et al., 2003; Telle, 2006), and that CO2 emissions have a significant negative

4

these studies are conducted with US, European, Canadian, Japanese or Australian data. The similar thing between these countries is that they are highly emission-intensive, i.e. countries with high GHG emission levels.

On the other hand, this study will give further understanding on the GHG emission-CFP relationship in a country that stands as an GHG reduction example, i.e. non-emission-intensive country. Sweden is seen as one of the most environmental thinking countries in the world, that fullfils all of the Kyoto protocol and EU ETS’s requirements. As such a bright example it is of particular interest to find out, how do the different business entities operate in such a business environment and how do the country’s environmetal policies affect them, i.e. what is the financial impact of GHGs reduction. Hence, the following research question is built: Are firms rewarded for their environmental performance?

There are two initial assumption in this study. First, GHG emissions affect financial performance in a negative manner. Second, increase in GHG emissions has a negative effect on financial performance. However, after taking in account the influence of different variables on firms’ financial performance, the results prove different. There is no relationship between GHG emissions and financial performance in Sweden, meaning that companies find no incentives to engage in CSP practices. Furthermore, financial performance of companies in emission-intensive industries is not impacted by GHG emissions. The results prove the same, when the relationship is tested for increase and decrease in emissions, implying that country-specific factors may be the reason for this. This research contributes to the existing literature by studying Swedish firms’ GHG emissions and gives new evidence on this relationship in an ETS member state. Hence, this study’s findings provide new evidence to the “pays to be green” literature and debate.

5

2. Theoretical background and Hypothesis development.

In the not so distant past a new demand is introduced by customers to companies. Firms are not only required to deliver quality products, but they have to operate in a social environmental manner. At first, this new perspective is not accepted by executives, because it is thought that the costs to operate in this new way will be too high and the benefits too low. In the next sections, a closer look is taken at CSP (and the way it is measured in different studies), EU ETS and GHG emissions, and what their effect is on firms’ financial performance. These elements are relevant to this research, because they give further understanding on the question “Does it pay to be green”. Different studies on this topic that are reviewed bellow.

2.1 Overall Corporate Social Performance-Corporate Financial Performance relationship Although there is a strong correlation between environmental performance and firm profitability (Cohen et al., 1995), an analysis of the existing literature regarding the relationship between CSP and CFP shows that at the beginning environment protection investments provide few financial benefits. Subsequently, Porter and Van der Linde (1995) suggest that there is a causal link between CSP and CFP and that pollution reduction provides future cost savings by increasing efficiency, reducing compliance costs, and minimizing future liabilities. The authors also theorize that the reason for opportunities for profitable pollution reduction to exist is that the majority of managers lack knowledge to understand the full cost of pollution. They posit that properly designed environmental regulations force firms to engage in more active innovation, which in turn brings benefits that offset the costs imposed by these regulations and make firms more competitive in the market. Hart (1997) builds further on that theory and posits that more companies are adopting the “going green” strategy, because they realize that at the same time they can reduce pollution and increase profits. He suggests that differences in the underlying environmental capabilities of firms result in excess returns, and managers might possess unique resources or capabilities that allow them to employ profitable environmental strategies that are difficult to imitate.

6

7

there is a positive relationship between eco-efficient companies and positive equity returns, whereas the same does not hold for less-eco efficient companies over the period 1995–2003.

Empirical studies, such as that made by Yang and Yao (2012), continue to build on the existing literature and to provide evidence for a positive relationship between environmental compliance and firm profitability. Their article uses a panel survey of 1200 firms, certified by ISO14000, from 12 Chinese cities. The authors posit that raising the level of environmental compliance within a firm is related to substantial preparation and re-engineering of the management structure and production process. The results of the study show that being certified by this standard increases Chinese firms’ profit rate by 3.5 percentage points. Using a sample of 156 Egyptian firms, Wahba (2008) studies the relationship between environmental responsibility and market value. The main result of this research is that corporate environmental responsibility is significant and positively related to firm market value, as measured by Tobin’s Q ratio. Hence, the market indeed rewards firms for their environmental consciousness.

Other articles, such as that of Margolis et al. (2008), use meta-analysis to research the CSP-CFP relationship. The results show that it is positive or at least non-negative. In their study Capon et al. (1990) show that there is a positive correlation between positive social performance and firm performance. Another meta-analytic review, made by Orlitzky et al. (2003), studies this relationship and shows that across studies, CSP is positively correlated with CFP, and that the relationship tends to be two-way and simultaneous. Furthermore, firm reputation is an integral part of this relationship.

8

2.2 EU Emission Trading Scheme.

This section introduces the major drivers for EU countries to perform environmental friendly. The EU ETS is the first and largest multinational programme that regulates the emission trading. As such it does influence not only countries, but companies in the individual countries as well. Some results of this influence on firm’s financial performance are presented bellow.

On January 1, 2005, the first and largest multi-country ETS in the world is created (Rosenzweig et al., 2005). As of January 2008, it applied to all 27 EU and to the three Member States of the European Economic Area (Norway, Iceland, and Liechtenstein). Over 10,000 installations (firms and other emission-intensive facilities) are covered by it in the energy and industrial sectors, which are responsible for approximately and 40% of its total GHG emissions (European Commission, 2008). This programme regulates emissions from key economic sectors (e.g. energy, transport, industry sectors, agriculture, forestry and waste management). It requires reductions in two phases, from 2005 to 2007 and from 2008 to 2012. At the end of 2007 concludes the first trading period, i.e. Phase I.

“Regarding allowances, each member state must develop a national allocation plan indicating the total amount of allowances that it intends to allocate for that period and how it proposes to allocate them” (Directive 2003/ 87/EC: European Commission, 2003). According to Engels (2009), new knowledge and competencies within the organization have to be developed in order to trade with emissions. In their study Gusbin and Kouvaritakis (2000) prove that the lack of emission market would results in USA and Japan, being the countries that would take on the highest costs.

9

of UK firms and show that the EU ETS is positively correlated to earnings before interest, tax, depreciation and amortization (EBITDA).

2.3 Greenhouse gas emissions and Financial Performance.

Some of the studies compare environmental to financial performance over time, while others analyze the effect of environmental performance on market value directly. In their studies, Delmas and Nairn-Birch (2011) and Busch and Hoffmann (2011), using Tobin’s Q, find that increase in carbon emission amounts has a negative impact on CFP. However, if accounting-based measurements, such as ROA and ROE, are used to test this relationship, the results will be different (Delmas and Nairn-Birch, 2011). In more detail, the results prove that when ROA is used to measure the GHG-CFP relationship, firms, that reduce their GHG emissions, will be on the losing side. The results will be similar, if ROE is used to measure this relationship.

Griffin et al. (2012) use carbon emission data obtained from Carbon Disclosure Project (CDP) to study the GHG-CFP relationship. The results prove that GHG emission levels are negatively correlated to stock prices. Furthermore, Griffin et al. (2012) posit that this negative relationship is stronger for carbon intensive firms. In a similar research, Matsumura et al. (2011), using CDP data, find the relationship between carbon emissions and firm value to be negatively correlated. In more detail, the authors find that the penalty equals $202 per ton of emissions, which far exceeds the spot price of carbon.

Saka and Oshika’s (2010) research has the objective to investigate several things: (1) is the impact of CO2 emissions on market value negative; (2) is the negative impact on market

value removed after the disclosure of CO2-related information; and, (3) alleviate is the negative

impact on market value decreasing, when firms participate in ETSs. The results of the research prove that: (1) CO2 emissions and market value are negatively correlated; (2) CO2-related

10

2.4 Hypotheses Development.

In their study, Wang et al. (2013) use GHG emissions to investigate the effect they have on CFP. Boiral et al. (2012) posit that there are two main approaches around which the analysis of the relationship between GHG emissions and CFP is structured. The first one, namely win-lose reasoning, seems to dominate in the debate about commitments to reduce GHG emissions over the second approach, namely win-win reasoning. The win-lose reasoning postulates that firm’s efforts to reduce GHG emissions lead to additional costs that could damage firm’s competitiveness. The win-win reasoning, on the other hand, suggests that corporate competitiveness can be improved by the efforts to reduce GHG emissions (Boiral et al., 2012). The second reasoning justifies this research’s objective.

The purpose of this research is to study the relationship between GHG emissions and financial performance of Swedish companies. This relationship as showed above is researched by many scholars. Ziegler et al. (2007) find evidence of a positive relationship between positive environmental performance and stock returns. Konar and Cohen (2001) prove that companies that perform better environmentally have higher market value, as measured by Tobin’s Q. Wahba (2008) continues to build on the existing literature and finds that environmental conscious companies are rewarded by the market, thus having higher market value, measured as well by Tobin’s Q. Another study that researches the GHG-CFP relationship is that of Griffin et al. (2012). The authors find that GHG emissions are negatively correlated to stock price. In earlier researches, Saka and Oshika (2010) and Matsumura et al. (2011) confirm that CO2

emissions have a significant negative impact on firm value. Of particular interest are the results proved by Delmas and Nairn-Birch (2011), which show another side of the GHG-CFP relationship, i.e. the effect that increase in annual amounts of carbon emissions has on financial performance. The results show that increase in carbon emissions has a significant negative impact on CFP. Based on this, the following research hypotheses are formed and tested:

H1: Financial performance will be negatively affected by GHG emissions.

H2: Increase in GHG emissions will affect financial performance negative.

11

Furthermore, increase in annual amounts of GHGs affect in a negative way financial performance. This gives Boiral et al.’s (2012) win-win reasoning certain advantage, because companies have obvious incentives to invest in emission reduction, i.e. better financial performance and increased competitiveness. However, most of the performed studies so far, such as that of Chapple et al. (2013), have researched this relationship in countries that are emission intensive. Thus, this research aims to study the GHG-CFP relationship in a country that is environmentally responsible and is one of the leaders in carbon reduction over the years, i.e. Sweden. Furthermore, Sweden is a country that has adopted a single stringent environmental standard, the EU ETS, which can be profitable and can lead to better firm performance (Dowell et al., 2000).

3. Data and Methodology.

3.1. Data description.

In order to measure the environmental performance of companies, GHG emission data are needed. This kind of data is reported by the Carbon Disclosure Project (CDP) on a yearly basis. The CDP is an organization that collects climate change information from firms and cities around the world (CDP, 2012). In their research of Australian companies, Wang et al. (2013) and Chapple et al. (2013) use GHG emissions data that are voluntarily disclosed by firms and published in the annual CDP reports.

In this study data on GHG emissions are obtained from the CDP as well, because the CDP is acknowledged as the most distinguished source of information on GHG emissions (Chapple et al., 2013). Data for both scope 1 and scope 2 emissions are being used. Scope 1 are direct emissions, i.e. they arise from GHG sources owned or controlled by a firm. Scope 2 GHG emissions are indirect emissions, i.e. consumption of electricity, heat and cooling rather than physically arising from a company (CDP, 2012). Scope 2 emissions are also called purchased electricity. Both scope 1 and scope 2 emissions reflect firms’ efficiency of resource use (Wang et al., 2013). In this study scope 1 and scope 2 emissions reflect the environmental performance of firms.

12

GHG emission data for the years 2010, 2011, and 2012 are obtained. For this research period 44 Swedish publicly listed firms disclose their GHG emissions. As in the study of Wang et al. (2013), in this study the Jarque–Bera test shows that emissions are not normally distributed. Therefore they are logarithmically transformed in order to get normality. According to Osborne (2010), this transformation is considered to benefit significantly correlation and regression analysis. As for the CFP data, share data are collected from the Nordic Exchange. Accounting data are collected from Datastream.

3.2 Methodology.

3.2.1 Dependent variable.

The dependent variable in this research is CFP. As mentioned in the Literature review section above the different scholars measure it using different variables. Some of them use accounting measures, such as ROA and ROS (Russo and Fouts, 1997) and ROE (Hart and Ahuja, 1996; Alvarez, 2012). However, a downside of accounting data is that it is based on historical information and can be manipulated (Wang et al., 2013). On the other hand, market measures, such as stock returns, are used as well to measure company’s financial performance. The advantage of market-based variables is that they account for market expectations and financial risk.

In this study financial performance is measured by Tobin’s Q. Tobin’s Q is measured as the ratio of the market value of the firm to invested capital at replacement cost (Tobin and Brainard, 1977). It looks at how efficiently do firms utilize their assets and whether further investments are required (Wang et al., 2013). In this study, the approach to measure the Q ratio by Chung and Pruitt (1994) and Wang et al. (2013) is used.

Tobin’s Q is calculated as:

13

sum of the firm’s net current liability and the book value of its long-term debt, and TA is the book value of the firm’s total assets.

3.2.2 Independent variables.

The studies, that use Tobin’s Q as their dependent variable, control for several firm specific factors, such as firm size, capital intensity, growth, leverage, and research and development (R&D) intensity (Hart and Ahuja, 1996). Some studies, as that of Dowell et al. (2000), use two additional variables: advertising expenditures and multinationality. Later on, Wang et al. (2013) use: (1) firm size, which is calculated as the log of the companies’ assets, (2) sales, (3) capital intensity, measured as the ratio of capital expenditures to sales, (4) growth, measured as the percentage change in sales in different years, (5) leverage, measured as the ratio of companies’ debt to assets, (6) beta, which controls for systematic risk, (7) industry, and (8) emissions, as their control variables in studying the GHG emissions-CFP relationship of Australian companies.

In this study, the research approach of Wang et al. (2013) is followed, because of several reasons. First, this research is studying the relationship between GHG emissions and CFP of Swedish companies, which means that this relationship is studied on national level. Thus, the multinationality variable is not needed. Second, advertising expenditures are difficult to find, thereby it is being excluded as a control variable. In the following paragraphs, the control variables are presented.

14

measured by the annual negative percentage change in 2011 and 2012. It takes the negative value of change in emissions in 2011 and 2012, and zero otherwise.

In this study an industry dummy variable is included to show, if the results of Griffin et al. (2012), reporting a significant negative correlation between carbon intensive firms and financial performance, are going to be confirmed. This dummy variable’s objective is to indicate, whether a particular company is in an emission-intensive industry or in a non-emission-intensive industry (Wang et al., 2013). Fig. 1 presents industry average emission from the acquired GHG emission data. It is found that the top two polluters in Sweden are pulp and paper, and transport industry with 4.64 and 3.91 million metric tonnes, respectively (for 2012). Three companies (Holmen AB, Svenska Cellulosa AB, and SAS AB) in these two industries account for 54.51% of the GHG emission in the sample for 2012. Thus, these two industries are classified as emission-intensive and other as non-emission-intensive. In the first case the dummy variable takes the value of one, and zero otherwise.

Fig. 1 Industry emissions

This figure represents the average industry emissions in Sweden in 2012. Emissions are calculated as the sum of Scope 1 and Scope 2 emission, and are measured in million metric tonnes.

Margolis et al. (2008) add that in the CSP-CFP relationship studies the most common control variables are firm size and risk. Firm size is calculated in many ways by different scholars. For example, Delmas and Nairn-Birch (2011) measure it as total assets, Alvarez (2012)

15

as total revenues, and Busch and Hoffman (2011) as market value. In this study firm size is measured as the firm’s market capitalization, because all of the firms in the sample are publicly listed. This variable is transformed logarithmically for the sake of normality (Wang et al., 2013).

Previous studies, such as these of Hillman and Keim (2001) and Wang et al. (2013), use beta to control for systematic risk. Thus, its impact on financial performance is recognized by scholars. In this study, the systematic risk is calculated using historical stock prices and the OMX GES Ethical Sweden Index-20, and for later purposes the percentage change in 2011 and 2012 is calculated.

In addition, sales growth is included as a control variable. Studies made by Wang et al. (2013) and King and Lenox (2001) confirm that growth is a significant variable. Both studies show that the correlation between growth and Tobin’s Q is positive.

Leverage is another variable that is considered to be significant for financial performance. Alvarez (2012) and Wang et al. (2013) find in their respective studies that the correlation between leverage and financial performance is negative. Following agency theory leads to the same negative relationship, indicating that higher leverage decreases the chances for increase in cash levels (Jensen, 1986). In this study, leverage is measured as the ratio of total debt to total assets, and the annual change in 2011 and 2012. The reason for this negative correlation is that highly leveraged companies might be less able to afford the required investments in order to implement more stringent standards (Dowell et al., 2000).

Capital intensity is measured as the ratio of capital expenditures to sales. King and Lenox (2001) find it is significant and the correlation between it and financial performance to be negative.

Finally, two year dummy variables are constructed to capture for time effects. First one takes the value of one, if the observations are for 2010, and zero otherwise. Second one takes the value of one, if the observations are for 2011, and zero otherwise.

3.2 Regressions.

16

the overall amount of emissions has on financial performance during the years. Model 1 presents the results, when the dummy variables are not included; Model 2 includes the industry variable, and not the year dummies; Model 3 includes year variables, and not the industry variable; Model 4 includes the industry and year variables; and Model 5 presents the results, when all dummy variables and an additional interaction variable are included. Thus, the following general regression is formed:

To obtain more detailed knowledge about the GHG-CFP relationship, regressions 6 to 19 are built. They have the purpose of showing the effect that the overall annual change in emissions, as well as the annual increase and decrease in emissions have on financial performance. In order to find out, if the annual change in GHG emissions affects Tobin’s Q, a new variable (EP), which measures the annual change in emissions in 2011 and 2012 is introduced in the regression, taking the place of EM:

Next, the relationship is tested with Tobin’s Q, leverage and risk, measured by annual changes in 2011 and 2012, EP, and rest being the same:

In order to find out, if the positive or negative change in GHG emissions during 2011 and 2012 affect Tobin’s Q, two new variables, that measure the positive and negative percentage change in emissions in 2011 and 2012, are introduced in the regression, as well as two interaction variables. Hence, the following general regression is formed:

17

Finally, Tobin’s Q, as well as leverage and risk are measured by annual changes in 2011 and 2012. Positive and negative change in emissions in 2011 and 2012, and their respective interaction variables, are added as well to test the impact they have on financial performance. Thus, the last general regression is formed:

SZ – logarithm of a firm’s market capitalization; CI – ratio of capital expenditures to sales; GR – growth in sales;

LV – leverage;

LVP – annual change in leverage in 2011 and 2012; RI – risk;

PRI – annual change in risk in 2011 and 2012; EM - logarithm of total emissions;

IND – industry dummy variable;

EI – interaction variable, measured as Emission times Industry; EP – annual change in Emission in 2011 and 2012;

P – positive change in emissions; N – negative change in emission;

PI – interaction variable, measured as Positive change in emissions times Industry; NI – interaction variable, measured as Negative change in emissions times Industry; Y2010 – year dummy variable;

18 4. Results and Discussion.

4.1 Descriptive statistics.

In order to improve the representativeness of the results, winsorization is used for all financial variables at 1% and 99%. Thus, possible outliers are taken out. Table 1 below presents descriptive statistics of the variables that are used in this research. The main control variables, emission, emission (annual), positive change and negative change have a mean of 4.66, 1.05, 1.12 and -0.07, respectively. The median is 4.77 for emission, 0.02 for emission (annual), 0.02 for positive change, and 0.00 for negative change. The number of observations is 132 for emission, and 88 for emission (annual), positive change and negative change.

Table 1 Descriptive statistics

This table presents descriptive statistics of the used variables. Tobin’s Q is measured as the ratio of the sum of market capitalization, liquidating value of the firm’s preferred stock and sum of the net current liability and book value of the firm’s long term debt to the book value of total assets of the firm. Tobin’s Q (annual) represents the annual change in Tobin’s Q. Size is calculated as the logarithm of the firm’s market capitalization. Capital intensity is the ratio of capital expenditures to sales. Growth is the percentage change in sales. Leverage is the ratio of total debt to total assets. Leverage (annual) represents the annual change in leverage. Risk is calculated using historical stock prices and OMX GES Ethical Sweden Index-20. Risk (annual) is the percentage change in risk. Emission is calculated as the logarithm of total emissions. Emission (annual) is the percentage change in emissions. Emission*Industry, Positive*Industry and Negative*Industry are interaction variables, calculated as industry times emissions, positive change in emissions, and negative change in emissions, respectively. Positive change is calculated as the annual positive change in emission. Negative change is calculated as the annual negative change in emissions. The variables are winsorized at the 1st and 99th percentiles.

Mean Median Maximum Minimum Std. Dev. Observations

19

4.2 Regression results and discussion.

Some of the results obtained in the research are similar to the ones proposed by existing theory. However, the main control variable that is of interest to this study proves not to have a significant effect on the dependent variable. This has to be explained in detail in this section. The findings of the study are discussed and economic explanation is found for them.

In this research 19 regression models are run. Table 2 presents the results for the first five models, thus showing the effect that the overall amount of emissions has on financial performance during the years 2010, 2011 and 2012. In all models size is statistically significant at the 1% level, while growth and leverage are significant at the 5% level. Out of these variables size and growth have positive signs, and leverage has a negative sign. Although, the emission variable has a negative sign as expected, it is insignificant. Hence, Hypothesis 1 is not supported by the findings in these models. The industry variable in Model 2, 3 and 5 does not affect significantly Tobin’s Q, which is in accordance with the results obtained by Wang et al. (2013).

20

Table 2 Model 1 to Model 5 regression coefficients

This table presents results for panel regressions. Tobin’s Q is the dependent variable and is measured as the ratio of the sum of market capitalization, liquidating value of the firm’s preferred stock and sum of the net current liability and book value of the firm’s long term debt to the book value of total assets of the firm. Size is calculated as the logarithm of the firm’s market capitalization. Capital intensity is the ratio of capital expenditures to sales. Growth is the percentage change in sales. Leverage is the ratio of total debt to total assets. Risk is calculated using historical stock prices and OMX GES Ethical Sweden Index-20. Emission is calculated as the logarithm of total emissions. Industry is a dummy variable, taking the value of 1, if a company is in emission-intensive industry and zero otherwise. Emission*Industry is interaction variable, calculated as industry times emissions. Year2010 and Year2011 are year dummy variables taking the value of 1, if an observation corresponds to year 2010 and year 2011, respectively, and zero otherwise.

Model 1 Model 2 Model 3 Model 4 Model 5

Variable Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error

C -8.9729 1.1269 -8.8485 1.1167 -8.8092 1.1717 -8.6790 1.1603 -8.7818 1.1746 SIZE 0.6303*** 0.0640 0.6230*** 0.0635 0.6166*** 0.0665 0.6088*** 0.0659 0.6139*** 0.0666 CAPITAL INTENSITY -0.1021 0.3233 -0.0962 0.3221 -0.0454 0.3247 -0.0385 0.3235 -0.0403 0.3256 GROWTH 0.1492** 0.0745 0.1488** 0.0745 0.1584** 0.0772 0.1588** 0.0772 0.1580** 0.0776 LEVERAGE -0.8671** 0.3634 -0.8800** 0.3621 -0.7974** 0.3637 -0.8097** 0.3623 -0.8011** 0.3651 RISK 0.0047 0.1468 -0.0115 0.1469 -0.1310 0.1579 -0.1534 0.1581 -0.1542 0.1591 EMISSION -0.0654 0.0671 -0.0544 0.0675 -0.0390 0.0677 -0.0260 0.0683 -0.0233 0.0689 INDUSTRY -0.7426 0.6896 -0.8261 0.6904 3.8979 7.8191 EMISSION*INDUSTRY -0.7654 1.2619 Year2010 0.0855 0.0374 0.0889 0.0374 0.0908 0.0378 Year2011 0.0326 0.0366 0.0334 0.0366 0.0358 0.0370 R-squared 0.4114 0.4075 0.4328 0.4304 0.4335 Adjusted R-squared 0.3831 0.3741 0.3959 0.3884 0.3866 F-statistic 14.5619*** 12.1870*** 11.7344*** 10.2438*** 9.2597*** Observation 132 132 132 132 132

21

Table 3 Model 6 and Model 7 regression coefficients

This table presents results for panel regressions. Tobin’s Q is the dependent variable for Model 6 and is measured as the ratio of the sum of market capitalization, liquidating value of the firm’s preferred stock and sum of the net current liability and book value of the firm’s long term debt to the book value of total assets of the firm. Tobin’s Q (annual) is the dependent variable for Model 7 and represents the annual change in Tobin’s Q. Size is calculated as the logarithm of the firm’s market capitalization. Capital intensity is the ratio of capital expenditures to sales. Growth is the percentage change in sales. Leverage is the ratio of total debt to total assets. Leverage (annual) represents the annual change in leverage. Risk is calculated using historical stock prices and OMX GES Ethical Sweden Index-20. Risk (annual) is the percentage change in risk. Emission (annual) is the percentage change in emissions. Emission (annual)*Industry is interaction variable, calculated as industry times emissions (annual). Industry is a dummy variable, taking the value of 1, if a company is in emission-intensive industry and zero otherwise. Year2011 is an year dummy variable taking the value of 1, if an observation corresponds to year 2011 and zero otherwise.

Model 6 Model 7

Variable Coefficient Std. Error Coefficient Std. Error

C -5.6825 1.5102 -0.3153 0.3253 SIZE 0.4159*** 0.0903 0.0225 0.0188 CAPITAL INTENSITY -0.1918 0.3262 0.0350 0.1398 GROWTH 0.0089 0.0881 0.2867** 0.1146 LEVERAGE -0.9783** 0.3904 LEVERAGE (ANNUAL) -0.0417 0.0349 RISK 0.1262 0.1925 RISK (ANNUAL) -0.0189 0.1162 EMISSION (ANNUAL) -0.0002 0.0023 -0.0017 0.0031 INDUSTRY -0.7817 0.7271 0.1699 0.1746

EMISSION (ANNUAL )*INDUSTRY -0.7013 1.4673 1.5996 2.1160

Year2011 0.0158 0.0311 -0.2174 0.0512

R-squared 0.2433 0.2765

Adjusted R-squared 0.1560 0.1931

F-statistic 2.7869*** 3.3127***

Observations 88 88

* 10% significance level, **5% significance level, ***1% significance level

22

Table 4 Model 8 to Model 13 regression coefficients

This table presents results for panel regressions. Tobin’s Q is the dependent variable and is measured as the ratio of the sum of market capitalization, liquidating value of the firm’s preferred stock and sum of the net current liability and book value of the firm’s long term debt to the book value of total assets of the firm. Size is calculated as the logarithm of the firm’s market capitalization. Capital intensity is the ratio of capital expenditures to sales. Growth is the percentage change in sales. Leverage is the ratio of total debt to total assets. Risk is calculated using historical stock prices and OMX GES Ethical Sweden Index-20. Positive change is calculated as the annual positive change in emission. Negative change is calculated as the annual negative change in emissions. Positive*Industry and Negative*Industry are interaction variables, calculated as industry times positive change in emissions, and negative change in emissions, respectively. Industry is a dummy variable, taking the value of 1, if a company is in emission-intensive industry and zero otherwise. Year2011 is an year dummy variable taking the value of 1, if an observation corresponds to year 2011 and zero otherwise.

Model 8 Model 9 Model 10 Model 11 Model 12 Model 13

Variable Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error C -5.6731 1.4912 -5.6511 1.5216 -5.6504 1.4847 -5.6774 1.5034 -5.6399 1.5157 -5.6507 1.5516 SIZE 0.4159*** 0.0891 0.4148*** 0.0908 0.4141*** 0.0888 0.4153*** 0.0899 0.4134*** 0.0908 0.4138*** 0.0926 CAPITAL INTENSITY -0.1974 0.3222 -0.1966 0.3246 -0.2031 0.3237 -0.1963 0.3277 -0.2029 0.3274 -0.1959 0.3315 GROWTH 0.0130 0.0867 0.0070 0.0902 0.0215 0.0933 0.0157 0.0953 0.0213 0.0945 0.0136 0.1001 LEVERAGE -0.9831** 0.3857 -0.9785** 0.3897 -0.9887** 0.3872 -0.9834** 0.3919 -0.9887** 0.3917 -0.9812** 0.3976 RISK 0.1180 0.1896 0.1109 0.1935 0.1251 0.1906 0.1313 0.1931 0.1263 0.1954 0.1291 0.2037 POSITIVE CHANGE -0.0002 0.0022 -0.0001 0.0022 -0.0000 0.0023 -0.0000 0.0023 NEGATIVE CHANGE -0.0492 0.1964 -0.0354 0.2010 -0.0482 0.2001 -0.0325 0.2062 INDUSTRY -0.7437 0.7131 -0.7336 0.7236 -0.7397 0.7146 -0.7782 0.7295 -0.7394 0.7226 -0.7722 0.7419 POSITIVE* INDUSTRY -7.9645 28.236 -2.3727 31.881 NEGATIVE* INDUSTRY -0.6711 1.5223 -0.6212 1.6825 Year2011 0.0131 0.0303 0.0151 0.0313 0.0119 0.0303 0.0148 0.0313 0.0117 0.0312 0.0150 0.0328 R-squared 0.2409 0.2433 0.2412 0.2435 0.2411 0.2435 Adjusted R-squared 0.1640 0.1560 0.1644 0.1562 0.1536 0.1340 F-statistic 3.1343*** 2.7871*** 3.1405*** 2.7906*** 2.7548*** 2.2248** Observations 88 88 88 88 88 88

23

Regression coefficients for models 14 to 19 are presented in Table 5. These models test the GHG-CFP relationship using annual changes in leverage, risk, positive and negative annual change in emissions in 2011 and 2012. The results show that in these models only growth is significant. While in every model growth has a positive sign, in models 14, 15, 16 and 18, and in models 17 and 19 it is significant at the 5% and 10% levels, respectively. Although positive annual change in emission has the expected negative sign in all models, it is statistically not significant, which means that Hypothesis 2 does not hold.

When each individual variable is run separately in regression, the results are the following (Appendix A). Only three variables (size, growth and leverage) are statistically significant, when the variables are tested with the ratio of Tobin’s Q in 2010, 2011 and 2012. Size and growth are significant at the 1% and 10% levels, respectively, and have positive signs, whereas leverage is statistically significant at the 10% level, but has a negative sign. These three variables have the highest goodness of fit measure. It is equal to 0.37, 0.023 and 0.026 for size, growth and leverage, respectively. When the individual variables are regressed with annual change in Tobin’s Q for 2011 and 2012, the obtained results are somewhat different (Appendix B). Growth is the only significant variable. It is significant at the 10% level, and has a positive sign. Similarly with the result in Appendix A, it has the largest goodness of fit measure, which equals 0.12.

Finally, because of the possibility of multicollinearity among the variables, an additional test, which has the purpose to serve for robustness, is performed. The Variance Inflation Factor (VIF) test, the results of which are not reported here, shows that all values are smaller than 10. This suggests that none of the variables has to be taken out of the regressions due to multicollinearity (Kutner et al., 2004).

24

Table 5 Model 14 to Model 19 regression coefficients

This table presents results for panel regressions. Tobin’s Q (annual) is the dependent variable and represents the annual change in Tobin’s Q. Size is calculated as the logarithm of the firm’s market capitalization. Capital intensity is the ratio of capital expenditures to sales. Growth is the percentage change in sales. Leverage (annual) represents the annual change in leverage. Risk (annual) is the percentage change in risk. Positive change is calculated as the annual positive change in emission. Negative change is calculated as the annual negative change in emissions. Positive*Industry and Negative*Industry are interaction variables, calculated as industry times positive change in emissions, and negative change in emissions, respectively. Industry is a dummy variable, taking the value of 1, if a company is in emission-intensive industry and zero otherwise. Year2011 is a year dummy variable taking the value of 1, if an observation corresponds to year 2011 and zero otherwise.

Model 14 Model 15 Model 16 Model 17 Model 18 Model 19

Variable Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error Coefficient Std. Error C -0.3422 0.3191 -0.3110 0.3218 -0.3682 0.3188 -0.3406 0.3246 -0.3524 0.3235 -0.3159 0.3254 SIZE 0.0239 0.0184 0.0224 0.0186 0.0246 0.0184 0.0231 0.0188 0.0239 0.0187 0.0220 0.0188 CAPITAL INTENSITY 0.0359 0.1379 0.0349 0.1379 0.0184 0.1402 0.0161 0.1420 0.0148 0.1417 0.0115 0.1414 GROWTH 0.2753** 0.1122 0.2841** 0.1128 0.3007** 0.1162 0.3153*** 0.1190 0.3016** 0.1173 0.3175*** 0.1183 LEVERAGE (ANNUAL) -0.0404 0.0344 -0.0412 0.0344 -0.0393 0.0345 -0.0407 0.0349 -0.0401 0.0348 -0.0414 0.0348 RISK (ANNUAL) -0.0056 0.1134 0.0198 0.1184 -0.0058 0.1136 -0.0224 0.1168 -0.0023 0.1149 0.0045 0.1357 POSITIVE CHANGE -0.0016 0.0031 -0.0017 0.0031 -0.0014 0.0031 -0.0015 0.0031 NEGATIVE CHANGE -0.1636 0.1737 -0.1781 0.1769 -0.1589 0.1757 -0.1764 0.1763 INDUSTRY 0.0738 0.1182 0.0181 0.1401 0.0807 0.1186 0.1947 0.1822 0.0782 0.1198 0.1103 0.3049 POSITIVE*INDUSTRY 32.858 44.278 21.490 63.952 NEGATIVE*INDUSTRY 1.8382 2.2086 1.1012 3.1197 Year2011 -0.2106 0.0497 -0.2175 0.0506 -0.2105 0.0497 -0.2181 0.0512 -0.2128 0.0504 -0.2221 0.0513 R-squared 0.2696 0.2759 0.2765 0.2847 0.2792 0.2888 Adjusted R-squared 0.1956 0.1923 0.2033 0.2022 0.1961 0.1858 F-statistic 3.6453*** 3.3025*** 3.7756*** 3.4509*** 3.3577*** 2.8060*** Observations 88 88 88 88 88 88

25

leverage and Tobin’s Q is found in the studies, made by Dowell et al. (2000) and Wang et al. (2013). The negative sign of leverage means that the larger the leverage of a company the worse it will perform financially.

Although regression coefficients of emission, the main control variable, are negative as anticipated, the variable is not significant neither when measured by the overall amount of emissions in 2010, 2011 and 2012, nor when measured by the annual change in emissions in 2011 and 2012. This means that an increase in emissions does not have significant time-series effects on financial performance. This is in accordance, though, with the findings of Saka and Oshika (2010). Their analysis for firms participating in ETSs shows that carbon emissions do not have a significant negative effect on market value. This might explain this study’s results, since Sweden is a member of an ETS and by far the most renowned country in emission reduction. Indeed, country-specific factors seem to affect this relationship. Hence, it is the win-lose reasoning that dominates in Sweden, i.e. firms that invest in emission reduction won’t be as competitive as companies that neglect such investments. It seems that the incentives for engaging in CSP practices are absent or seem to be small (Margolis et al., 2007). Since the industry variable is, as well, insignificant the same can be concluded for companies in emission-intensive industries.

Overall it can be stated that the results obtained are not the ones that were anticipated, but they also expand the knowledge on the relationship between GHG emissions and CFP, although further research is needed, testing the current elaborations on the reasons for certain results.

5. Conclusion.

26

industries in Sweden is not affected by GHG emissions. Furthermore, regressions run for measuring the effect that increase or decrease in GHG emissions show that firms’ financial performance is not impacted. A possible reason for these findings, are the country-specific factors. In the end it can be concluded that Swedish firms are not incentivized to reduce their GHG emissions.

This research contributes to the growing body of literature by studying the effect that GHG emissions have on firms’ financial performance in Sweden. The findings in this study add to the literature about this relationship. Thus, new evidence is added to the “pays to be green" debate in ETS member states, which will bring further understanding of the commitment that companies make to reduce GHG emissions and its impact.

These findings have important implications as they highlight the non-generalizability of the results found in other countries, which is critical both for academics researching the CSP-CFP relationship and for managers deciding on the emission reduction investments. This study implies that, when an ETS is enforced, those companies with high GHG emissions won’t perform worse financially than the companies that are trying to reduce their GHG emissions.

27 References

Alvarez, I. G. (2012). Impact of CO2 Emission Variation on Firm Performance.Business Strategy and the Environment, 21(7), 435-454.

Boiral, O., Henri, J. F., & Talbot, D. (2012). Modeling the impacts of corporate commitment on climate change. Business Strategy and the Environment, 21(8), 495-516.

Busch, T., & Hoffmann, V. H. (2011). How hot is your bottom line? Linking carbon and financial performance. Business & Society, 50(2), 233-265.

Capon, N., Farley, J. U., & Hoenig, S. (1990). Determinants of financial performance: a meta-analysis. Management Science, 36(10), 1143-1159.

CDP (Carbon Disclosure Project) (2012). Climate Change Programs.

https://www.cdproject.net/en-US/Programmes/Pages/climate-changeprograms. Aspx

Chapple, L., Clarkson, P. M., & Gold, D. L. (2013). The cost of carbon: Capital market effects of the proposed emission trading scheme (ETS). Abacus.

Chung, K. H., & Pruitt, S. W. (1994). A simple approximation of Tobin's q. Financial

management, 23(3).

Cohen, M. A., Fenn, S., & Naimon, J. S. (1995). Environmental and financial performance: are they related?. Investor Responsibility Research Center, Environmental Information Service. Connolly, R. A., & Hirschey, M. (2005). Firm size and the effect of R&D on Tobin's q. R&D Management, 35(2), 217-223.

Delmas, M. A., & Nairn-Birch, N. S. (2011). Is the tail wagging the dog? An empirical analysis of corporate carbon footprints and financial performance.

Derwall, J., Guenster, N., Bauer, R., & Koedijk, K. (2005). The eco-efficiency premium puzzle. Financial Analysts Journal, 51-63.

Dowell, G., Hart, S., & Yeung, B. (2000). Do corporate global environmental standards create or destroy market value?. Management Science, 46(8), 1059-1074.

28

Engels, A. (2009). The European Emissions Trading Scheme: An exploratory study of how companies learn to account for carbon. Accounting, Organizations and Society, 34(3), 488-498. European Commission. (2003). Directive 2003/87/EC of the European Parliament and of the Council of 13 October 2003 establishing a scheme for greenhouse gas emission allowance trading within the Community and amending Council Directive 96/61. Oficial Journal of the European Union, No: L, 275(25.10).

European Commission (2008). Questions and Answers on the Commission’s Proposal to Revise the EU ETS, press release

Feldman, S. J., Soyka, P. A., & Ameer, P. G. (1997). Does improving a firm's environmental management system and environmental performance result in a higher stock price?. The Journal of Investing, 6(4), 87-97.

Garnaut, R. (2008). The Garnaut climate change review. Global Environmental change, 13, 1-5. Griffin, P., Lont, D., & Sun, Y. (2012). The relevance to investors of greenhouse gas emission disclosures. UC Davis Graduate School of Management Research Paper, (01-11).

Guenster, N., Bauer, R., Derwall, J., & Koedijk, K. (2011). The Economic Value of Corporate Eco‐Efficiency. European Financial Management, 17(4), 679-704.

Gusbin D, Kouvaritakis N. (2000). Kyoto Protocol and Emission Trading: Potential Cost Savings and Emission Reduction. Economic Evaluation of Quantitative Objectives for Climate Change.

Final Report, Comision Europea, Bruselas

Hamilton, J. T. (1995). Pollution as news: media and stock market reactions to the toxics release inventory data. Journal of environmental economics and management, 28(1), 98-113.

Hart S. (1997). Beyond greening: strategies for a sustainable world. Harvard Business Review, 75(1), 66–76.

29

Hayami, H., & Nakamura, M. (2007). Greenhouse gas emissions in Canada and Japan: Sector-specific estimates and managerial and economic implications. Journal of Environmental Management, 85(2), 371-392.

Hillman, A. J., & Keim, G. D. (2001). Shareholder value, stakeholder management, and social issues: what's the bottom line?. Strategic management journal, 22(2), 125-139.

Horváthová, E. (2010). Does environmental performance affect financial performance? A meta-analysis. Ecological Economics, 70(1), 52-59.

Jamet, S. (2011). Enhancing the Cost-Effectiveness of Climate Change Mitigation Policies in Sweden. OECD Economics Department Working Papers, No. 841, OECD Publishing

Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The

American Economic Review, 76(2), 323-329.

Karpoff, J. M., Lott Jr, J. R., & Wehrly, E. W. (2005). The Reputational Penalties for Environmental Violations: Empirical Evidence*. Journal of Law and Economics, 48(2), 653-675.

King, A. A., & Lenox, M. J. (2001). Does it really pay to be green? An empirical study of firm environmental and financial performance: An empirical study of firm environmental and financial performance. Journal of Industrial Ecology, 5(1), 105-116.

King, A., & Lenox, M. (2002). Exploring the locus of profitable pollution reduction. Management Science, 48(2), 289-299.

Klassen, R. D., & McLaughlin, C. P. (1996). The impact of environmental management on firm performance. Management science, 42(8), 1199-1214.

Konar, S., & Cohen, M. A. (1997). Information as regulation: The effect of community right to know laws on toxic emissions. Journal of environmental Economics and Management, 32(1), 109-124.

Konar, S., & Cohen, M. A. (2001). Does the market value environmental performance?. Review of Economics and Statistics, 83(2), 281-289.

30

Margolis, J. D., Elfenbein, H. A., & Walsh, J. P. (2007). Does it pay to be good? A meta-analysis and redirection of research on the relationship between corporate social and financial performance. Ann Arbor, 48, 1109-1234.

Margolis, J. D., & Elfenbein, H. A. Walsh., JP (2008). Do Well by Doing Good? Don't Count on It. Harvard Business Review, 86(1).

Matsumura, E. M., Prakash, R., & Vera-Muñoz, S. C. (2011). Voluntary disclosures and the firm-value effects of carbon emissions. Manuscript submitted for publication.

Nehrt, C. (1996). Timing and intensity effects of environmental investments.Strategic Management Journal, 17(7), 535-547.

Orlitzky, M., Schmidt, F. L., & Rynes, S. L. (2003). Corporate social and financial performance: A meta-analysis. Organization studies, 24(3), 403-441.

Osborne, J. W. (2010). Correlation and Other Measures of Association. The Reviewer’s Guide to

Quantitative Methods in the Social Sciences, 55.

Porter M, Van der Linde C. (1995). Green and competitive. Harvard Business Review September–October, 121–134.

Rosenzweig, R., Forrister, D., & Youngman, R. (2005). A Paper on European Greenhouse Gas Performance, the European Union Emissions Trading Scheme, and Technology Policy, prepared for the TransAtlantic Conference on Energy Policy. NATSOURCE LLC, March, 28.

Russo, M. V., & Fouts, P. A. (1997). A resource-based perspective on corporate environmental performance and profitability. Academy of management Journal, 40(3), 534-559.

Saka, C., & Oshika, T. (2010). Market Valuation of Corporate CO₂ Emissions, Disclosure and Emissions Trading. Disclosure and Emissions Trading (June 1, 2010).

Smale, R., Hartley, M., Hepburn, C., Ward, J., & Grubb, M. (2006). The impact of CO2 emissions trading on firm profits and market prices. Climate Policy, 6(1), 31-48.

31

Stanwick, S. D., & Stanwick, P. A. (2000). The relationship between environmental disclosures and financial performance: an empirical study of US firms. Eco‐Management and Auditing, 7(4), 155-164.

Telle, K. (2006). “It Pays to be Green”–A Premature Conclusion?. Environmental and Resource Economics, 35(3), 195-220.

Thomas, A. (2001). Corporate environmental policy and abnormal stock price returns: an empirical investigation. Business strategy and the Environment, 10(3), 125-134.

Tobin, J., & Brainard, W. C. (1977). Asset markets and the cost of capital. Economic Progress,

Private Values, and Public Economic Journal, 96, 120-138.

Wahba, H. (2008). Does the market value corporate environmental responsibility? An empirical examination. Corporate Social Responsibility and Environmental Management, 15(2), 89-99. Davidson, W. N., & Worrel, D. L. (1988). The impact of announcements of corporate illegalities on shareholder returns. Academy of management journal, 31(1), 195-200.

Wang, L., Li, S., & Gao, S. (2013). Do Greenhouse Gas Emissions Affect Financial Performance?–an Empirical Examination of Australian Public Firms.Business Strategy and the Environment.

Yang, X. and Yao, Y. (2012). Environmental Compliance and Firm Performance: Evidence from China. Oxford Bulletin of Economics and Statistics, 74, 397–424.

Ziegler, A., Schröder, M., & Rennings, K. (2007). The effect of environmental and social performance on the stock performance of European corporations. Environmental and Resource Economics, 37(4), 661-680.

32

Appendix A

This table presents results for single variable panel regressions. Tobin’s Q is the dependent variable and is measured as the ratio of the sum of market capitalization, liquidating value of the firm’s preferred stock and sum of the net current liability and book value of the firm’s long term debt to the book value of total assets of the firm. Size is calculated as the logarithm of the firm’s market capitalization. Capital intensity is the ratio of capital expenditures to sales. Growth is the percentage change in sales. Leverage is the ratio of total debt to total assets. Risk is calculated using historical stock prices and OMX GES Ethical Sweden Index-20. Emission is calculated as the logarithm of total emissions. Emission (annual) is the percentage change in emissions. Industry is a dummy variable, taking the value of 1, if a company is in emission-intensive industry and zero otherwise. Emission*Industry is interaction variable, calculated as industry times emissions. Year2010 and Year2011 are year dummy variables taking the value of 1, if an observation corresponds to year 2010 and year 2011, respectively, and zero otherwise.

Variable Coefficient (Std. Error)

33 Year2011 -0.0731 (0.0457) R-squared 0.3710 0.0003 0.0226 0.0263 0.0037 0.0000 0.0006 0.0068 0.0066 0.0612 0.0192 Adjusted R-squared 0.3662 -0.0074 0.0249 0.0189 -0.0039 -0.0076 -0.0109 -0.0007 -0.0010 0.0540 0.0116 F-statistic 76.6980*** 0.0341 3.3515* 3.524718* 0.4875 0.0005 0.0566 0.8997 0.8699 0.4777 0.5487

* 10% significance level, **5% significance level, ***1% significance level

Appendix B

This table presents results for single variable panel regressions. Tobin’s Q (annual) is the dependent variable and represents the annual change in Tobin’s Q. Size is calculated as the logarithm of the firm’s market capitalization. Capital intensity is the ratio of capital expenditures to sales. Growth is the percentage change in sales. Leverage (annual) represents the annual change in leverage. Risk (annual) is the percentage change in risk. Emission (annual) is the percentage change in emissions. Positive change is calculated as the annual positive change in emission. Negative change is calculated as the annual negative change in emissions. Positive*Industry and Negative*Industry are interaction variables, calculated as industry times positive change in emissions, and negative change in emissions, respectively. Industry is a dummy variable, taking the value of 1, if a company is in emission-intensive industry and zero otherwise. Year2011 is a year dummy variable taking the value of 1, if an observation corresponds to year 2011 and zero otherwise.

Variable Coefficient (Std. Error)

34 CHANGE (0.0037) NEGATIVE CHANGE -0.0661 (0.2030) INDUSTRY 0.0537 (0.1387) POSITIVE* INDUSTRY 6.6074 (44.1679) NEGATIVE*I NDUSTRY -0.4235 (1.7375) Year2011 -0.1767 (0.0518) R-squared 0.0329 0.0039 0.1221 0.0021 0.0014 0.0006 0.0006 0.0018 0.0025 0.0003 0.0010 0.0895 Adjusted R-squared 0.0216 -0.0076 0.1107 -0.0094 -0.0101 -0.0109 -0.0110 -0.0097 -0.0090 -0.0112 -0.0105 0.0691 F-statistic 2.9276* 0.3391 1.9433 0.1841 0.1289 0.0566 0.0533 0.1591 0.2199 0.0331 0.0878 0.05392