University of Groningen Faculty of Economics and Business
MSc Finance
The impact of emission reductions on the cost of
capital
David Jaarsma S2751976 June 6th, 2018
Abstract
Firms, investors, and other stakeholders are increasingly concerned about climate change risks. In this study, I focus on the relationship between changes in a firm’s emissions level and changes in its cost of capital. I use the S&P 500 index and the Nikkei 225 index in separate panel regressions to test this relationship. I find that for US firms changes in lagged total emission levels is positively and significantly associated with changes in the WACC. I could not find such a relationship for Japanese firms. This may indicate that cross-country differences in regulation and society could affect the way emission reductions are valued in the market.
Keywords
1. Introduction
Global carbon emissions have significantly increased over the last century. If emissions will continue to increase at the current rate, it could have serious consequences on global biodiversity and human societies (Mora et al., 2013). As a result, investors, regulators, firms, and society show an increasing interest in environmental issues. For example, a rising number of investors are applying environmental screens when allocating their capital to stocks (Kempf and Osthoff, 2007). Similarly, commitments to divest from the fossil fuel industry have increased among institutions and individuals, representing a total divested value of approximately $6.09 trillion as of May 2018 (Trinks et al., 2017; www.gofossilfree.org/divestment/commitment). Moreover, the Paris Agreement was sealed by the United Nations to combat climate change by reducing greenhouse gas (GHG) emissions, and was signed in 2016 by 195 signatories (https://unfccc.int/process#:a0659cbd-3b30-4c05-a4f9-268f16e5dd6b).
The increased attention to the environment has led to a growing body of empirical literature investigating the relationship between a firm’s environmental performance (EP) and its financial performance (Dowell and Muthulingham, 2017). Although there seems to be a positive relationship between environmental and financial performance, mixed results imply that there are other factors influencing this relationship (Lewandowski, 2017; Jacobs, 2014). Fatemi et al. (2015) report that some researchers have found a positive relationship between CSR and a firm’s performance, whereas others find a negative or even an inverse U-relationship. These mixed results stem from the various competing theoretical views on the relationship between environmental and financial performance (El Ghoul et al., 2011).
Several researchers (e.g., Ng and Rezaee, 2015; El Ghoul et al., 2011; Chava, 2014; Connors and Silva-Gao, 2008; Goss and Roberts, 2011; Oikonomou et al., 2014) have empirically investigated the relationship between CSR performance and the cost of capital, where they rely on binary ratings from the database KLD STATS to measure CSR performance. However, binary ratings give little insight into a firm’s actual performance relative to others and are therefore hard to compare. By looking at reported emissions, I examine the impact of actual environmental performance on the cost of capital.
Only a few studies examine the relationship between firm’s carbon emissions and its cost of capital. Kim et al. (2015) study Korean firms and find that carbon intensity (carbon emissions divided by sales) and the cost of equity capital are positively related. Li et al. (2014), with a sample of Australian firms, do not find a relationship between carbon intensity and the cost of equity. They do, however, find a positive relationship between carbon intensity and the cost of debt: firms with lower levels of carbon intensity face a lower cost of debt. This positive relationship between emission levels and the cost of capital is then generally interpreted as a motivation for firms to reduce emissions so that they can benefit from a lower cost of capital (Lewandowski, 2017). “However, a positive association between a firm’s annual reported carbon emissions and financial performance does not necessarily reflect a positive relationship between improvements in carbon performance over time and financial performance” (Lewandowski, 2017, p.1197). For example, firms might spend too much resources to reduce their emission level or their efforts to reduce emissions might be ineffective, resulting in reduced profits and increased risk. So far, no study examines the impact of changes in emission levels on changes in the cost of capital. Hence, there is no conclusive evidence that actual reductions in a firm’s emission level leads to a lower cost of capital. I try to fill this gap by considering how changes in a firm’s carbon emissions impact its cost of capital. I address the following research question:
What is the impact of changes in a firm’s carbon emissions on its cost of capital?
imply that there is a positive relationship between a change in emissions and the cost of capital. I contribute to the existing literature by examining whether decreases in emissions will result in a decrease in a firm’s cost of capital. Second, by looking at emissions rather than aggregate scores of environmental performance, I might provide insight into the impact of a firm’s actual behavior on the cost of capital. Finally, I use two indices in my analysis: the S&P 500 and the Nikkei 225. These indices are from different geographical regions (i.e., the US and Japan, respectively) and may provide evidence of cultural and regulatory factors influencing the relationship between emissions and cost of capital. Using Hofstede’s (1980; 1991) five dimensions of national culture, the US is considered more individualistic and short-term oriented, while Japan is considered more collectivistic and long-term oriented. These differences might influence the way investors and firms value environmental performance and, more specifically, emission reductions. Moreover, Japanese firms may also face more regulatory pressures to reduce their emissions. I argue that markets would value emission reductions by US firms more than Japanese firms, because the US has less governmental and societal monitoring (Gupta, 2018). Although this is not the main focus of my paper, it might give an indication of cultural differences in environmental performance.
While prior research generally finds a negative relationship between EP and the cost of capital, it does not necessarily imply that a reduction in a firm’s emission levels leads to a lower cost of capital (Lewandowski, 2017). To the best of my knowledge, this paper is the first to examine the impact of changes in emission levels on the cost of capital. The findings will have implications for management, as a reduction in emission levels might lead to a lower cost of capital. In that case, lower cost of capital may serve as an incentive for management to engage in efforts of emission reduction. Moreover, this paper might give insight into the role of country differences.
2. Literature
The literature has produced mixed findings on the relationship between environmental performance and financial performance, reflecting different theoretical views on this relationship (Jiao, 2010; El Ghoul et al, 2011). Broadly, there are two contrasting theories regarding environmental performance: the shareholder theory and the stakeholder theory (Ge and Liu, 2015). Shareholder theory suggests that a firm’s only responsibility is to maximize profits for shareholders (Friedman, 1962). Therefore, efforts by firms to enhance their environmental performance can be regarded as using corporate resources and management effort in a way which is not in the best interest of shareholders and reduces profits (Walley and Whitehead, 1994; Preston and O’Bannon, 1997). Goss and Roberts (2011) call this the overinvestment view. Drawing on agency theory, they argue that management, by investing in EP, might gain private benefits at the expense of shareholder value. By contrast, the stakeholder theory suggests that a firm should create shareholder value while protecting the interests of all stakeholders (Freeman, 1984). Building on stakeholder theory, Ng and Rezaee (2015) suggest that EP might enhance the long-term value of the firm by satisfying its social responsibilities, meeting their environmental obligations and improving their reputation. Moreover, a firm with high EP ‘is more likely to improve customer loyalty, increase employee attraction, retention rates and productivity, effectively lobby for tax breaks and have a superior access to capital’ (Oikonomou et al., 2014, p.54). Early studies mainly adopted the shareholder theory and assumed that environmental performance provide no or little financial/economic benefits (Gallego-Álvarez et al., 2015; Waddock and Graves, 1997). Porter and van der Linde (1995), however, argue that firms can be competitive and green at the same time. Therefore, firms can serve shareholder and stakeholder needs at the same time (Fatemi et al., 2015). For example, Smith and Rönnegard (2016) point out that as shareholder preferences move away from maximizing shareholder value to encompass broader stakeholder concerns, management can engage in efforts to increase environmental performance without interfering with shareholder needs.
Another perspective on the EP–financial performance relationship is the risk mitigation view, which states that environmental performance can increase firm value by mitigating the exposure to several risks (Goss and Roberts, 2011; Trinks et al., 2017).
literature points to several reasons why reducing emissions could decrease a firm’s cost of capital. First, irresponsible firms may face uncertain future claims or other negative events, such as product boycotts, fines, penalties, government sanctions, and litigation costs (Waddock and Graves, 1997; Oikonomou et al., 2014). Moreover, Sharfman and Fernando (2008) point out that environmental performance may decrease the possibility that firms face extreme environmental events (e.g., oil spills, waste dumping) and, as a result, financial claims. Consequently, firms with good environmental performance are less likely to be in financial distress and investors will therefore require a lower risk premium, resulting in a lower cost of equity (Goss and Roberts, 2011). Similarly, lenders of debt will assess these “greener” firms as having a lower default risk due to these uncertain future negative events, which will reduce the cost of debt.
Second, irresponsible firms may face a higher cost of capital because of the crowding out of moral-constrained investors and lenders. Initially, Heinkel et al. (2001) develop a model in which they focus on investors in equity markets. Building on the capital market equilibrium model of Merton (1987), Heinkel et al. (2001) argue that the presence of green investors drives down the share price of polluting firms, since they do not want to hold shares in these firms. Therefore, neutral investors, who hold shares in both polluting and green firms, require a higher expected return. Indeed, Hong and Kacperczyk (2009) find that norm-constrained investors hold less stocks in controversial firms and, consequently, these “sin” stocks require a higher expected return and face a greater litigation risk. Moreover, Heinkel et al. (2001) conclude that a large enough fraction of green investors can “force” polluting firms to turn into green firms, because the benefits of being green (i.e., a higher share price) will exceed the costs of reforming. Menz (2010) hypothesizes that this mechanism might be even more important for the debt market. Menz (2010) argues that the amount of debt a firm bears is much higher than the amount of equity. In addition, firms need to revert to the debt market more often to refinance their loans. Finally, lenders may be liable by law for environmental damages caused by borrowers and may face a reputation risk by lending to environmentally irresponsible firms (Chava, 2014). Concluding, Menz (2010, p.121) states that lenders “have the quantitative potential as well as the incentive to convince firms to behave more responsible”, which would reduce the cost of debt.
in which they show that EP information may signal a responsible image to investors. Second, responsible firms may receive more coverage by media and analysts (Hong and Kacperczyk, 2009). Finally, information related to responsible firms is likely to be better received by socially conscious investors than from irresponsible firms. Furthermore, Oikonomou et al. (2014) argue that management, by showing good environmental performance, can signal its competence and trustworthiness to stakeholders, thereby reducing potential agency risks and lowering the cost of capital.
Several previous studies have examined the effect of environmental performance on financial performance and the cost of capital. While not being exhaustive, Appendix 1 gives an overview of the current literature. In general, most studies find a significantly negative relationship between some measure of environmental performance and cost of capital, and a positive relationship between environmental performance and financial performance measures such as return on equity. Most research covers the period between 2002 and 2007, while Lewandowski (2017) and Trinks et al. (2017) use a more recent sample, with periods up until 2015 and 2016, respectively. Studies mainly focus on US firms when investigating the relationship between environmental performance and financial performance, followed by an international sample. Moreover, panel regressions with fixed effects is a commonly used method to estimate this relationship. There is more variation in the way environmental performance and the cost of capital is measured. For example, El Ghoul et al. (2011), Chava (2014), and Ng and Rezaee (2015) rely on KLD data to construct a measure of environmental performance, while Gupta (2018) constructs an environmental sustainability index of KLD data. By contrast, Li et al. (2014), Kim et al. (2015), and Trinks et al. (2017) measure GHG emission intensity; Alvarez (2012) measure first differences in emissions; and Gallego-Álvarez et al. (2015) measure percentage changes in emissions. Cost of equity is generally calculated using some implied cost of equity method. Sharfman and Fernando (2008) and Trinks et al. (2017), however, use the CAPM to calculate the cost of equity. Goss and Roberts (2011), Chava (2014), and Oikonomou et al. (2014) measure the cost of debt as the (log of the) yield spread, while Li et al. (2014) divide the interest expenses over the long-term debt to arrive at the cost of debt.
announcements are viewed negatively by markets. Similarly, Fisher-Vanden and Thorburn (2011) find that firms that announce to participate in voluntary corporate environmental initiatives face significant losses in market value. It is interesting to note the similarities between these two studies, and the differences with the other studies described in Appendix 1. First, Jacobs et al. (2010) and Fisher-Vanden and Thorburn (2011) both use an event study to examine the effect of announcements on subsequent abnormal returns. Second, announcements to participate in voluntary environmental programs shows only the intention of firms to reduce their emissions, but does not necessarily lead to actual environmental performance.
The literature points to several mechanisms in which environmental performance could linearly reduce the cost of capital (i.e., the win logic). Others point to the so-called win-lose reasoning and view emission reduction efforts as costs that reduces firms’ competitiveness (Boiral et al., 2012). Still others take on a ‘too-little-of-a-good-thing’ (TLGT) view, which implies that after reaching a certain minimum level of environmental performance, the relationship between EP and FP becomes positive (Trumpp and Guenther, 2017). Trumpp and Guenther (2017) and Lewandowski (2017) find evidence for this U-shaped relationship. However, Boiral et al. (2012) argue that the win-win logic is the dominant view in the literature. Moreover, most literature discussed in Appendix 1 show a positive and linear relationship between EP and FP, reflecting the win-win logic. Hence, I will test the following hypotheses in this paper:
Hypothesis 1: There is a positive relationship between changes in a firm’s emission levels and changes in its cost of equity.
Hypothesis 2: There is a positive relationship between changes in a firm’s emission levels and changes in its cost of debt.
Hypothesis 3: There is a positive relationship between changes in a firm’s emission levels and changes in its cost of capital.
results will be similar. Consistent with Sharfman and Fernando’s (2008) prediction regarding cost of equity, El Ghoul et al. (2018) find no cross-country differences when investigating the impact of EP on cost of equity worldwide. By contrast, Gupta (2018) does find cross-country differences and show that the relationship between EP and cost of equity is stronger for countries where country-level governance is weak. Busch and Hoffmann (2011) find country differences for Japan, Europe, and North America when examining the EP-FP relationship.
Japan and the US can be characterized using Hofstede’s (1980; 1991) five dimensions of national culture. They differ on two dimensions which may affect their assessment of climate change and efforts to reduce emissions. While Japan can be viewed as more collectivistic and long-term oriented, the US is considered more individualistic and short-term oriented. Hence, Japanese society may pressure firms more to reduce their emissions than US society would, because Japan is more concerned with the long-term effects of emissions and climate change on society.
The regulatory environment regarding climate change also significantly differs between the US and Japan. Thus far, the US lacks dedicated climate change legislation, mainly due to difficulties to secure support for comprehensive climate change legislation in the Senate. As a result, the US government is managing GHG emissions mainly using regulation under the Clean Air Act of 1963. Former president Obama announced in 2013 the Climate Action Plan, which sets out actions to accomplish the GHG emission reduction (http://www.lse.ac.uk/GranthamInstitute/country-profiles/united-states-of-america). However, the current Trump administration has announced to withdraw from the Paris Agreement and weakened the impact of the Climate Action Plan, which increases the uncertainty regarding US climate change legislation. Meanwhile, Japan has two key climate-dedicated laws in place. The Energy Conservation Law of 1979 promotes the effective and rational use of energy. Moreover, the Act on Promotion of Global Warming Countermeasures was enacted in 1998 and aims to reduce GHG emissions. This law obliges the government to produce and implement a plan containing reduction targets and detailed actions that government entities, business sectors and citizens shall take to achieve the targets (http://www.lse.ac.uk/GranthamInstitute/country-profiles/japan/).
3. Methodology
This section describes the methodology employed to test my hypotheses. I use a panel regression model with industry and year-fixed effects to estimate the impact of emission reduction on the change of the cost of capital. An advantage of panel data is “the ability to control for possibly correlated, time-invariant heterogeneity without observing it” (Arellano, 2003, p.8). Moreover, “panel data facilitates making reasonable causal inferences based on economic theory” (Trinks et al., 2017, p.8). Hence, I want to estimate the following fixed effects models, where I control for year fixed effects.
Δ𝐶𝑜𝐸𝑖,𝑡 = 𝛽0+ 𝛽1Δ𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝑠𝑖,𝑡−1+ 𝛽2Δ𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝛽3Δ𝑀𝑇𝐵𝑖,𝑡+ 𝛽4Δ𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖,𝑡 + 𝛽5𝑅𝑂𝐴𝑖,𝑡+ 𝛽6𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 + 𝑢𝑖,𝑡 (1) Δ𝐶𝑜𝐷𝑖,𝑡 = 𝛽0+ 𝛽1Δ𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝑠𝑖,𝑡−1+ 𝛽2Δ𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝛽3Δ𝑀𝑇𝐵𝑖,𝑡+ 𝛽4Δ𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖,𝑡 + 𝛽5𝑅𝑂𝐴𝑖,𝑡+ 𝛽6𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 + 𝑢𝑖,𝑡 (2) Δ𝐶𝑜𝐶𝑖,𝑡 = 𝛽0+ 𝛽1Δ𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝑠𝑖,𝑡−1+ 𝛽2Δ𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝛽3Δ𝑀𝑇𝐵𝑖,𝑡+ 𝛽4Δ𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖,𝑡 + 𝛽5𝑅𝑂𝐴𝑖,𝑡+ 𝛽6𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 + 𝑢𝑖,𝑡 (3)
Δ𝐶𝑜𝐸𝑖,𝑡, Δ𝐶𝑜𝐷𝑖,𝑡, and Δ𝐶𝑜𝐶𝑖,𝑡 are my estimates of the annual percentage change in the cost of equity, cost of debt, and the cost of capital for firm i at time t, respectively. Δ𝐸𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝑠𝑖,𝑡−1 is the percentage change in firm’s i carbon emissions at time t – 1. I lag emissions by one year, because I hypothesize that environmental performance precedes financial performance. In addition, by lagging emissions, I ensure that information on emissions have been fully processed and valued by investors and lenders. Size, market to book ratio (MTB), leverage ratio and return on assets (ROA) are control variables. To remain consistent, these variables are also defined as the annual percentage change. Industry is a dummy variable equal to 1 if a firm is active in a particular industry, and 0 otherwise.
variable (i.e., cost of capital) is measured as the percentage change. By looking at changes in the cost of capital rather than absolute values, one can better test what the impact of emission reductions is on the cost of capital. The variables are now explained in more detail.
Cost of capital
I use the Capital Asset Pricing Model (CAPM) to estimate the cost of equity. The CAPM is a widely-used measure for the expected rate of return (Elbannan, 2015). However, the CAPM is shown to yield imprecise estimates of the cost of equity (Fama and French, 1997). Prior studies have mainly used an implied cost of equity, which requires analyst forecasts to estimate the cost of equity (e.g., El Ghoul et al. [2011], Chava [2014], Kim et al., [2015]). Unfortunately, I do not have access to the I/B/E/S database which provides analyst forecast data. Hence, I follow Sharfman and Fernando (2008) and Trinks et al. (2017) who use the CAPM to test the EP-FP relationship. I estimate the cost of equity based on the prior five-year arithmetic average monthly returns. Then, I average the monthly cost of equities to arrive at a yearly cost of equity. I assume the MSCI World Index as the market index. Furthermore, I use the yearly arithmetic averages of the yield on ten-year Treasury bonds and the ten-year Japanese government bond yield as the risk-free rate for the S&P 500 and Nikkei 225, respectively. The yield on Japanese government bonds is low, which might, as a result, underestimate the cost of equity. However, since the yield is low for more than a decade, the cost of equity is consistently underestimated in my sample. Since I look at percentage changes in the cost of equity, I expect that the low interest rate will not pose a serious threat. Finally, I use a market risk premium of 5% which is in line with estimates of Koller et al. (2015).
I calculate the cost of debt as interest expenses divided by long- and short-term debt, which is a straightforward way to measure the cost of debt and is in line with Li et al. (2014). This is opposed to Oikonomou et al. (2014), Sharfman and Fernando (2008), Chava (2014) and Goss and Roberts (2011), who use the credit spread as a proxy for the cost of debt. However, I was not able to collect data on spreads from Datastream.
Finally, the firm’s after-tax weighted average cost of capital (WACC) is calculated using the following formula (Modigliani and Miller, 1958):
𝑊𝐴𝐶𝐶 = ( 𝐸
𝐷 + 𝐸) 𝑟𝐸 + ( 𝐷
where E is the market value of the firm’s equity, D is the market value of the firm’s debt, 𝑟𝐸 is the firm’s cost of equity capital, 𝑟𝐷 is the firm’s cost of debt capital, and T is the firm’s effective tax rate.
Emission reduction
I measure emission reduction as the annual percentage change in a firm’s total reported emission levels. By looking at relative changes in a firm’s emission levels, I account for the large differences in firm’s emission levels. This definition of emission reduction is consistent with, amongst others, Hart and Ahuja (1996) and Gallego-Álvarez et al. (2015), but differs from Alvarez (2012), who looks at first differences in emissions, and from Nishitani and Kokubu (2012), who use net sales divided by carbon emissions as a proxy for emission reductions
Control variables
I control for several variables which might affect the risk and, consequently, the cost of capital of a firm. Size, measured as the natural logarithm of total assets, is argued to be negatively related with risk: larger firms are considered less risky (Fama and French, 1993). In addition, size may also proxy for increased coverage and reputation, which might reduce information asymmetry and the cost of capital (Goss and Roberts, 2011; Trinks et al., 2017). Fama and French (1992) state that a low MTB ratio is an indicator of financial distress and increases the cost of capital, while leverage, defined as debt over equity, is argued to have a positive relationship with firm risk. ROA is a profitability measure which controls for the possibility that less profitable firms might reduce their efforts to reduce emissions, because of a lack of available funds (Goss and Roberts, 2011; Trinks et al., 2017). Finally, I control for industry effects by including industry dummy variables. Different industries may have completely different costs of capital and emission levels, which would affect the results of my models. Industry classification is based on the Industry Classification Benchmark (ICB), which distinguishes ten different industries: oil and gas, basic materials, industrials, consumer goods, health care, consumer services, telecommunications, utilities, financials, and technology.
Table 1: Summary of variables used in regression
Variable Exp. sign Why? Empirical evidence
Emissions + Riskier, lower investor base, higher information asymmetry
Li et al. (2014), Kim et al. (2015), Trinks et al. (2017), Connors and Silva-Gao (2008)
Size - Less risky (“too-big-to-fail”) Fama and French (1993)
Market-to-book ratio
- Smaller possibility of financial distress
Fama and French (1992)
Leverage + More debt increases default risk Fama and French (1992) Return on
assets
- Smaller possibility of financial distress
Goss and Roberts (2011)
4. Data
I obtain data on emissions from the Thomson Reuters Datastream database, which gathers emission data for each fiscal year from public sources such as annual and CSR reports. My focus is on total emissions, which includes direct (or Scope 1) and indirect emissions. Since reporting on Scope 3 emissions is currently poor and not yet widespread, I assume that reported indirect emissions only include Scope 2 emissions (Trinks et al., 2017, p.12). Additionally, I follow Trinks et al. (2017) by only considering emissions from 2008 and onwards. Reporting on emissions is of poorer quality in the years before 2008. So, my sample consists of data from 2008-2017, and the first data point is in 2009, since I’m looking at percentage changes in emissions.
Data on emissions are self-reported by firms, which entails some drawbacks as identified by Trinks et al. (2017). There is no standardized way of reporting on emissions, and it is also not regulated. Consequently, firms may report on different emissions or not include all emissions in their report. This reduces the quality and reliability of the data. However, despite these concerns, reported emissions is the best indication of a firm’s emissions investors and lenders have. Therefore, if they use emission levels in their evaluation of a firm’s riskiness, I expect that they will use the information reported by firms.
The financial variables are also obtained from Datastream. Both the financial and the emission variables are winsorized at the 1st and 99th percentile to mitigate the impact of extreme outliers on the regression results.
I ended up with a sample of 321 US firms. Similarly, 187 firms included in the Nikkei 225 report data about emissions. The final sample included 183 Japanese firms. The distribution of years and industries are shown in Appendix 2.
Table 2 shows descriptive statistics of the financial and emission variables, in terms of absolute values. Descriptive statistics in terms of percentage changes can be found in Appendix 3. Emissions are measured in terms of CO2-equivalent metric tons. The mean emissions for
companies included in the S&P is much higher than that of Japanese firms (7.28 million metric tons compared to 4.05 million). Both medians differ a lot from their respective mean. This indicates that, as expected, there are a few big emitters while a large group has much lower emission levels. This huge variation can also be seen in the large standard deviations. Moreover, the medians of the US and Japanese sample lie much closer together than their means, indicating that there are more big US than Japanese emitters, and that the majority of lower-emitting firms have more similar levels of emissions. The mean total emissions of Japanese firms are comparable with the sample of Trinks et al. (2017), while the mean for US firms is much higher. The mean cost of equity for US firms is 7% and for Japanese firms 5%. These numbers are comparable with Trinks et al. (2017), but higher than the estimates of El Ghoul et al. (2011) and Ng and Rezaee (2015) and lower than Gupta (2018) and Sharfman and Fernando (2008). The mean cost of debt is much lower for Japanese firms compared with US firms. The mean cost of debt for US firms, however, is comparable with the results of Sharfman and Fernando (2008), but lower than Li et al. (2014). US firms appear to be more profitable than Japanese firms, as the mean measures for MTB and ROA are higher. Finally, both leverage ratios are similar.
Table 2: Summary statistics of key EP variables and financial variables for S&P 500 and Nikkei 225 (in absolute values)
S&P 500
Emission variables
N Mean Median SD Min Max Skewness Kurtosis Total emissions (tCO2e) 2291 7,286,965 784,785 18,488,008 8459 148,200,000 4.18 22.30 Direct emissions (tCO2e) 1997 5,505,033 197,975 15,581,126 137 148,200,000 4.25 22.78 Indirect emissions (tCO2e) 1911 1,024,748 318,864 1,998,130 4816 12,467,875 3.78 18.75
Financial variables
N Mean Median SD Min Max Skewness Kurtosis CoD 3249 0.04 0.04 0.02 0.001 0.16 1.53 8.46 CoE 3286 0.07 0.07 0.02 0.03 0.16 0.79 3.88 WACC 2790 0.06 0.06 0.02 0.02 0.12 0.35 2.77 MTB 3308 3.48 2.42 4.62 -11.41 31.99 3.09 20.27 Size (ln, USD thousands) 3354 16.92 16.80 1.33 13.70 21.66 0.70 3.81 Leverage 3354 0.27 0.26 0.16 0.00 0.74 0.48 2.94 ROA 3354 0.06 0.05 0.06 -0.17 0.23 0.12 4.91
Nikkei 225
Emission variables
N Mean Median SD Min Max Skewness Kurtosis Total emissions (tCO2e) 1603 4,050,500 607,079 12,054,573 9826.18 81,619,688 4.95 28.17 Direct emissions (tCO2e) 832 1,687,096 209,913 3,943,925 737.66 20,957,989 3.31 13.92 Indirect emissions (tCO2e) 795 735,001 283,000 1,046,079 1640,95 4,959,000 2.23 7.71
Financial variables
N Mean Median SD Min Max Skewness Kurtosis CoD 1817 0.01 0.01 0.01 0.00 0.10 3.25 17.05 CoE 1786 0.05 0.05 0.02 0.01 0.09 -0.12 2.63 WACC 1533 0.03 0.03 0.01 0.00 0.08 0.48 3.02 MTB 1839 1.28 1.11 .067 0.43 3.94 1.68 6.36 Size (ln, JPY thousands) 1848 21.27 21.07 1.33 18.79 26.43 0.98 4.25 Leverage 1847 0.26 0.24 0.15 0.00 0.61 0.22 2.28 ROA 1848 0.02 0.02 0.03 -0.12 0.12 -0.72 6.43
Table 3: correlations matrix of emission and financial variables S&P 500 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Total emissions (1) 1.00 Direct emissions (2) 0.60 1.00 Indirect emissions (3) 0.66 0.23 1.00 CoD (4) 0.05 0.05 0.03 1.00 CoE(5) 0.02 -0.001 0.07 -0.01 1.00 WACC (6) -0.004 0.02 0.02 0.38 0.47 1.00 MTB (7) -0.03 0.003 -0.05 0.02 0.02 0.11 1.00 Size (8) 0.20 0.09 0.20 -0.16 0.01 -0.18 -0.22 1.00 Leverage (9) 0.05 0.09 0.03 -0.30 0.03 -0.26 0.05 0.21 1.00 ROA (10) -0.07 -0.005 -0.04 0.10 0.007 0.24 0.11 -0.18 -0.07 1.00 Nikkei 225 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Total emissions (1) 1.00 Direct emissions (2) 0.54 1.00 Indirect emissions (3) 0.47 0.09 1.00 CoD (4) 0.11 0.12 0.09 1.00 CoE(5) -0.03 -0.02 -0.02 -0.01 1.00 WACC (6) 0.04 0.04 0.06 0.18 0.32 1.00 MTB (7) -0.01 -0.07 0.09 -0.01 -0.02 0.14 1.00 Size (8) 0.09 0.004 0.13 -0.17 -0.004 0.04 0.05 1.00 Leverage (9) -0.003 0.007 0.02 -0.34 -0.01 -0.13 -0.06 0.14 1.00 ROA (10) 0.05 0.02 0.05 0.08 0.02 0.40 0.04 0.10 -0.09 1.00
Table 4: Test of means
Univariate tests
To compare the impact of changes in emissions on changes in the cost of capital, I first divide the sample into two groups depending on whether total emissions increased or decreased compared with the previous year. Subsequently, I run an ANOVA test to test for differences in the mean change in cost of capital. The results are presented in table 4. The mean change in cost of capital does not significantly differ between firms whose emissions decline and those whose emissions rise. In the next section, I will examine and test my hypotheses through multivariate panel regressions.
5. Results
Before I discuss the results, I first analyse the statistical assumptions of the panel regression. EViews is not able to test for heteroskedasticity in panel data. Nevertheless, I will use White period standard errors to account for the possibility of heteroskedasticity. Concerning autocorrelation, the Durbin-Watson statistic gives values ranging from 1.245 to 1.817, indicating a possible presence of autocorrelation. White period standard errors, however, estimate standard errors that are robust to heteroskedasticity and (within cross-section) autocorrelation (Arellano, 2003, p.18). Table 5 shows the correlation between variables. None of the correlation coefficients are higher than 0.4 (except for correlations between emissions), suggesting that multicollinearity is not an issue. Finally, I test whether a fixed effects model is necessary by running a redundant fixed effects test. This test is highly significant for all models, indicating that using a fixed effects model is justified.
Table 5 shows the results of the panel regression with year fixed effects. After controlling for factors that could affect the risk of a firm (see Table 1), I find a positive association between the cost of capital measures and lagged total emissions for US firms. This relationship is only significant at the 5% level for the WACC. The CoE and CoD models are not significant. A 1% increase in this year’s emissions level will, ceteris paribus, result in, an increase of 2.63% in the WACC next year.
who found that Japanese firms reducing their emissions are more likely to enhance their firm value.
Table 5: Lagged total emissions: regression results S&P: CoE S&P: CoD S&P:
WACC Nikkei: CoE Nikkei: CoD
Nikkei: WACC Constant -0.0024 (0.0069) 0.2608*** (0.0517) 0.0117* (0.0071) -0.0295** (0.0115) 0.0242 (0.0437) -0.0870*** (0.0261) Total emissions 0.0063 (0.0094) 0.0862 (0.0547) 0.0263** (0.0128) -0.0098 (0.0115) -0.0057 (0.0223) -0.0192 (0.0318) MTB -0.0127 (0.0078) -0.0093 (0.0248) 0.0041 (0.0091) -0.0970*** (0.0272) 0.0296 (0.0359) 0.1430** (0.0631) Size -0.0844 (0.3354) -7.8348*** (1.1904) -2.4563*** (0.4687) -1.6958 (1.5007) -13.5658*** (2.6715) -4.0875 (2.9151) Leverage 0.0067 (0.0060) -0.3645*** (0.0413) -0.0512*** (0.0108) 0.00199 (0.0237) -0.4572*** (0.0352) -0.1202*** (0.0320) ROA -0.0006 (0.0016) 0.0072 (0.0057) 0.0073 (0.0060) 0.0034 (0.0212) 0.0009 (0.0040) 0.0394*** (0.0100) Year effect Yes Yes Yes Yes Yes Yes Industry Yes Yes Yes Yes Yes Yes Adj. R-squared 0.4006 0.1389 0.3869 0.1296 0.1997 0.1664 F-statistic 59.21 15.02 46.72 9.74 15.79 10.23 N 1830 1827 1522 1233 1246 972
This table shows the fixed-effects regressions of cost of capital measures on lagged total emission reductions and the control variables. CoE is measured using the CAPM. CoD is interest expense divided by long and short-term debt. Total emissions is the one-year lagged annual percentage change in total emissions. MTB is market value of equity to book value of equity. Size is the log of total assets. Leverage is total debt over total assets. ROA is return on assets. Robust standard errors clustered at the firm-level (cross-section) are in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
with my hypotheses, and in line with studies such as Chava (2014), El Ghoul et al. (2011) and Sharfman and Fernando (2008), who also find that US firms with better EP face a lower cost of equity or debt. The finding that the impact of EP on the cost of capital seems to be stronger for US firms than for Japanese firms is supported by Gupta (2018), who argues that this relationship is stronger for countries where climate regulation and governance is weaker.
As a robustness check, I repeat the regression model with direct and indirect emissions replacing total emissions. The results can be found in Appendix 4. Looking at US firms, we find results that are consistent with our earlier findings in table 5 and 6. There seems to be a positive relationship between changes in emissions and changes in cost of capital, regardless of Scope 1, Scope 2 or Scope 1+2 emissions. Hence, efforts of US firms to reduce emissions are rewarded by the market with a lower cost of capital.
Regarding Japanese firms, lagged direct emissions again only yield negative signs. On the other hand, cost of equity and the WACC show a positive sign for lagged indirect emissions. Interestingly, direct and indirect emissions with no lag show a positive relationship with the cost of debt, while it is negatively related with the cost of equity.
Table 6: Total emissions with no lag: regression results S&P: CoE S&P: CoD S&P:
WACC Nikkei: CoE Nikkei: CoD
Nikkei: WACC Constant -0.0207*** (0.0071) 0.2894*** (0.0520) -0.0054 (0.0079) -0.0347*** (0.0108) 0.0304 (0.0374) -0.0791** (0.0252) Total emissions 0.0049 (0.0100) 0.0646 (0.0579) 0.0158 (0.0142) -0.0094 (0.0153) 0.0387 (0.0370) -0.0018 (0.0237) MTB -0.0141 (0.0091) 0.0190 (0.0310) 0.0041 (0.0086) -0.0946*** (0.0237) 0.0474 (0.0372) 0.1932*** (0.0619) Size -0.2705 (0.3413) -5.6737*** (1.6642) -2.0966*** (0.4581) -4.4982*** (1.7296) -13.1113*** (2.4994) -5.0273* (3.0510) Leverage -0.0007 (0.0061) -0.3694*** (0.0422) -0.0549*** (0.0107) 0.0183 (0.0215) -0.4014*** (0.0265) -0.1087*** (0.0256) ROA -0.0031** (0.0015) 0.0168* (0.0089) 0.0118** (0.0060) 0.0017 (0.0023) 0.0002 (0.0040) 0.0419*** (0.0098) Year effect Yes Yes Yes Yes Yes Yes
Industry Yes Yes Yes Yes Yes Yes Adj. R-squared 0.3378 0.1317 0.3326 0.0878 0.2020 0.1746
This table shows the fixed-effects regressions of cost of capital measures on total emission reductions and the control variables. CoE is measured using the CAPM. CoD is interest expense divided by long and short-term debt. Total emissions is the annual percentage change in total emissions. MTB is market value of equity to book value of equity. Size is the log of total assets. Leverage is total debt over total assets. ROA is return on assets. Robust standard errors clustered at the firm-level (cross-section) are in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
So far, I have looked at the percentage change in a firm’s emission levels, but I have not yet considered the amount of emissions a firm emits. As several authors have suggested a curvilinear relationship (see Lewandowski, 2017) between environmental and financial performance, it might be that low and high emitters may face different changes in the cost of capital as their emission level changes. To test this, I run separate regressions with the 20% highest and lowest total emitters. The results are shown in table 7. I find that the cost of equity of both US and Japanese firms change in different directions depending on whether the firm is a high or low emitter in absolute terms. Low emitting firms face a negative (but insignificant) relationship between changes in total emissions and changes in cost of equity. By contrast, for firms with high emission levels, this association is positive. The results in table 7 indicate that investors value reductions in emissions more for high emitting firms than they do for low emitting firms. This might imply that efforts undertaken by low emitting firms to reduce their emission levels are inefficient (or, in other words, too-little-of-a-good-thing), in that the marginal benefits of emission reduction do not outweigh the additional costs. I did not find a similar association for the cost of debt.
This table shows the fixed-effects regressions of cost of capital measures on total emission reductions and the control variables. The regression was separately performed with the 20% highest and 20% lowest observations in terms of total emissions. CoE is measured using the CAPM. CoD is interest expense divided by long and short-term debt. Total emissions is the annual percentage change in total emissions. MTB is market value of equity to book value of equity. Size is the log of total assets. Leverage is total debt over total assets. ROA is return on assets. Robust standard errors clustered at the firm-level (cross-section) are in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
Year effect Yes Yes Yes Yes Industry Yes Yes Yes Yes Adj. R-squared 0.5672 0.5872 0.0731 0.1344 F-statistic 5.10 7.59 1.24 1.72 N 261 321 260 322 Nikkei 225 CoE CoD 20% lowest emitters 20% highest emitters 20% lowest emitters 20% highest emitters Constant -0.0210 (0.0103) 0.0030 (0.0063) 0.0887*** (0.0188) -0.0313*** (0.0058) Total emissions -0.0386 (0.0426) 0.0125 (0.0237) -0.0497 (0.0640) -0.0211 (0.0242) MTB 0.0007 (0.0826) -0.0212 (0.0605) -0.0454 (0.1188) -0.0094 (0.0495) Size -0.3343 (4.2805) -4.9194 (4.3360) -29.0706*** (9.8613) -15.8602*** (5.2709) Leverage 0.0460 (0.0820) -0.0329 (0.0737) -0.5188*** (0.0974) -0.6237*** (0.1648) ROA 0.0141 (0.0104) 0.0048 (0.0041) 0.0174 (0.0182) -0.0051 (0.0045) Year effect Yes Yes Yes Yes Industry Yes Yes Yes Yes Adj. R-squared -0.0498 0.1928 0.2825 0.2109 F-statistic 0.82 2.16 2.45 2.29
6. Conclusion
Firms, investors, and other stakeholders are increasingly concerned about climate change risks. In this study, I focus on the relationship between changes in a firm’s emissions level and changes in its cost of capital. While a growing number of studies examine the relationship between environmental performance and a firm’s cost of capital, so far none have examined the impact of changes in emission levels on changes in cost of capital. Yet, given the contrasting theories on the EP-FP relationship, it is an important question for management whether emission reductions are rewarded by the market with a lower cost of capital. I use the S&P 500 index and the Nikkei 225 index in separate panel regressions to test this relationship. I find that for US firms changes in lagged total emission levels is positively and significantly associated with changes in the WACC, which is consistent with my hypotheses. Interestingly, I could not find any evidence of this relationship for Japanese firms. In fact, most models yield a negative sign for emissions. Even though these coefficients are not significant, it may indicate that cross-country differences in regulation and society could affect the way EP is valued in the market. Specifically, I argue that there is more regulatory and social pressure in Japan to reduce emissions than in the US. Hence, the impact of emission reductions on the cost of capital is stronger for US firms (Gupta, 2018). An alternative explanation might be that Japanese firms are less efficient in their efforts to reduce emissions, which could result in increased riskiness or reduced profitability.
Finally, I perform regressions to test whether the hypothesized relationship differs for low and high emitters. I do not find evidence for different relationships between high and low emitters, but it is interesting to note that – in both samples – the sign of the relationship for the cost of equity is negative for low emitters and positive for high emitters. This implies that firms with relatively high emission levels can benefit from emission reductions, while firms with relatively low emission levels can be “punished” with a higher cost of capital after emission reductions. These findings may support the “too-little-of-a-good-thing” effect (Lewandowski, 2017; Trumpp and Guenther, 2017), because firms with low emission levels have less potential to reduce their emissions than high emitters have. However, this assumes that environmental performance is defined as emission reductions.
Ghoul et al. (2011), Chava (2014), Sharfman and Fernando (2008), Gupta (2018) and Ng and Rezaee (2015). Concerning Japanese firms, however, I could not find such a relationship. This contradict the conclusion of Kim et al. (2015) that efforts undertaken by firms to reduce their emissions is rewarded by the market with a lower cost of equity. As is illustrated with the Japanese sample, this is not necessarily the case. Instead, I argue that country differences in regulation and social pressure matter and should be considered when measuring the impact of emission reductions and EP.
My study contributes to the existing literature by examining the impact of reducing emissions on the cost of capital. In the literature, environmental performance is linked with a lower cost of equity. However, the impact of emission reductions on the cost of capital has not yet been considered. This is an important question, because, given the increased pressure on firms to reduce emissions, a decrease in the cost of capital might give managers an incentive to engage in efforts to reduce emissions. The results have implications for managers as the results imply that, at least for US firms, reducing emissions is compensated with a lower cost of capital. Hence, managers should consider the impact of emission reductions on the firm’s value. A major implication for researchers is that differences in corporate governance, regulation, and culture regarding climate change might explain country differences. Hence, future research should consider these country-specific factors when examining the EP-FP relationship.
References
Alvarez, I.G. 2012. Impact of CO2 emission variation on firm performance. Business
Strategy and the Environment. 21, 435–454.
Arellano, M. 2003. Panel data econometrics. Oxford: Oxford University Press.
Boiral, O., Henri, J-F., & Talbot, D. 2012. Modeling the impacts of corporate commitment on climate change. Business Strategy and the Environment. 21, 495–516.
Busch, T., & Hoffmann, V.H. 2011. How hot is your bottom line? Linking carbon and financial performance. Business & Society. 50, 233–265.
Chava, S. 2014. Environmental externalities and cost of capital. Management Science. 60, 2223–2247.
Connors, E., & Silva-Gao, L. 2008. The impact of environmental risk on the cost of equity capital: evidence from the toxic release inventory. Working paper.
Dhaliwal, D., Li, O.Z., Tsang, A.H, & Yang, Y.G. 2011. Voluntary non-financial disclosure and the cost of equity capital: the initiation of corporate social responsibility reporting. The Accounting Review. 86, 59–100.
Dowell, G.W.S., & Muthulingham, S. 2017. Will firms go green if it pays? The impact of disruption, cost, and external factors on the adoption of environmental initiatives. Strategic Management Journal. 38, 1287–1304.
El Ghoul, S., Guedhami, O., Kwok, C.C.Y. & Mishra, D.R. 2011. Does corporate social responsibility affect the cost of capital? Journal of Banking and Finance. 35, 2388– 2406.
Elbannan, M.A. 2014. The capital asset pricing model: An overview of the theory. International Journal of Economics and Finance. 7, 216–228.
Fama, E.F., & French, K.R. 1992. The cross-section of expected stock returns. The Journal of Finance. 47, 427–465.
Fama, E.F., & French, K.R. 1993. Common risk factors in the returns on stocks and bonds. Journal of Financial Economics. 33, 3–56.
Fama, E.F., & French, K.R. 1997. Industry costs of equity. Journal of Financial Economics. 43, 153–193.
Fatemi, A., Fooladi, I., & Tehranian, H. 2015. Valuation effects of corporate social responsibility. Journal of Banking & Finance. 59, 182–192.
Fisher-Vanden, K., & Thorburn, K.S. 2011. Voluntary corporate environmental initiatives and shareholder wealth. Journal of Environmental Economics and Management. 62, 430–445.
Freeman, R. 1984. Strategic management: A stakeholder perspective. Prentice-Hall, New Jersey.
Friedman, M. 1962. Capitalism and Freedom. University of Chicago Press, Chicago.
Gallego-Álvarez, I., Segura, L., & Martínez-Ferrero, J. 2015. Carbon emission reduction: the impact on the financial and operational performance of international companies. Journal of Cleaner Production. 103, 149–159.
Ge, W., & Liu, M. 2015. Corporate social responsibility and the cost of corporate bonds. Journal of Accounting and Public Policy. 34, 597–624.
Gupta, K. 2018. Environmental sustainability and implied cost of equity: International evidence. Journal of Business Ethics. 147, 343–365.
Hart, S.L., & Ahuja, G. 1996. Does it pay to be green? An empirical examination of the relationship between emission reduction and firm performance. Business Strategy and the Environment. 5, 30 – 37.
Heinkel, R., Kraus, A., & Zechner, J. 2001. The effect of green investment on corporate behavior. Journal of Financial and Quantitative Analysis. 36, 431–449.
Hofstede, G. 1980. Culture’s consequences: international differences in work-related values. Beverly Hills, CA: Sage.
Hofstede, G. 1991. Cultures and organizations: software of the mind. London: McGraw-Hill.
Hong, H., & Kacperczyk, M. 2009. The price of sin: the effects of social norms on markets. Journal of Financial Economics. 93, 15–36.
Horváthová, E. 2010. Does environmental performance affect financial performance? A meta-analysis. Ecological Economics. 70, 52–59.
Jacobs, B.W. 2014. Shareholder value effects of voluntary emissions reduction. Production and Operations Management. 23, 1859–1874.
Jacobs, B.W., Singhal, V.R., & Subramaniam, R. 2010. An empirical investigation of environmental performance and the market value of the firm. Journal of Operations Management. 28, 430–441.
Kempf, A., & Osthoff, P. 2007. The effect of socially responsible investing on portfolio performance. European Financial Management. 13, 908–922.
Kim, Y-B., An, H.T., & Kim, J.D. 2015. The effect of carbon risk on the cost of equity capital. Journal of Cleaner Production. 93, 279–287.
Koller, T., Goedhart, M., & Wessels, D. 2015. Valuation, measuring and managing the value of companies. Sixth university edition. John Wiley & Sons.
Lewandowski, S. 2017. Corporate carbon and financial performance: the role of emission reductions. Business Strategy and the Environment. 26, 1196–1211.
Li, Y., Eddie, I., & Liu, J. 2014. Carbon emissions and the cost of capital: Australian evidence. Review of Accounting and Finance. 13, 400–420.
Menz, K.M. 2010. Corporate social responsibility: is it rewarded by the corporate bond market? A critical note. Journal of Business Ethics. 96, 117–134.
Merton, R.C. 1987. A simple model of capital market equilibrium with incomplete information. The Journal of Finance. 42, 483–510.
Modigliani, F., & Miller, M.H. 1958. The cost of capital, corporation taxes and the theory of investment. American Economic Review. 58, 261–297.
Mora, C., Frazier, A.G., Longman, R.J, Dacks, R.S. Walton, M.M., Tong, E.J., Sanchez, J.J., Kaiser, L.R., Stender, Y.O., Anderson, J.M., Ambrosino, C.M., Fernandez-Silva, I., Giuseffi, L.M., & Giambelluca, T.W. 2013. The projected timing of climate departure from recent variability. Nature. 502, 183–187.
Nishitani, K., & Kokubu, K. 2012. Why does the reduction of greenhouse gas emissions enhance firm value? The case of Japanese manufacturing firms. Business Strategy and the Environment. 21, 517–529.
Oikonomou, I., Brooks, C., & Pavelin, S. 2014. The effects of corporate social performance on the cost of corporate debt and credit ratings. The Financial Review. 49, 49–75.
Porter, M.E., & Van der Linde, C. 1995. Toward a new conception of the environment-competitiveness relationship. Journal of Economic Perspective. 9, 97–118.
Preston, L.E., & O’Bannon, D.P. 1997. The corporate social-financial performance relationship: a typology and analysis. Business & Society. 36, 419–429.
Schaltegger, S., & Synnestvedt, T. 2002. The link between ‘green’ and economic success: Environmental management as the crucial trigger between environmental and economic performance. Journal of Environmental Management. 65, 339–346.
Sharfman, M.P., & Fernando, C.S. 2008. Environmental risk management and the cost of capital. Strategic Management Journal. 29, 569–592.
Smith, N.C., & Rönnegard, D. 2016. Shareholder primacy, corporate social responsibility, and the role of business schools. Journal of Business Ethics. 134, 463–478.
Trinks, A., Ibikunle, G., Mulder, M., & Scholtens, B. 2017. Greenhouse gas emission intensity and the cost of capital. SOM Research Reports, no.17017-EEF, University of Groningen, SOM research school, Groningen.
Trumpp, C., & Guenther, T. 2017. Too little or too much? Exploring U-shaped relationships between corporate environmental performance and corporate financial performance. Business Strategy and the Environment. 26, 49–68.
Walley, N., & Whitehead, B. 1994. It’s not easy being green. Harvard Business Review. 72, 46–51.
Appendix 1: Overview of literature
Table A.1: Literature review
Authors Period Region Number of observations
Dep. Variable Ind. Variable Method Results
Alvarez (2012) 2006-2010 International sample.
89 firms ROA and ROE Variation in emissions (first differences)
Panel regression Did not found a relationship between emission variation from 2006-07 and ROE in subsequent years. One-year lagged ROA was significantly negatively associated with emission variation (but not two-, three- and four-year lags). May be due to the financial crisis.
Chava (2014) 1992-2007 Firms included in the KLD database (i.e.,
mainly US)
Not discussed CoE is defined as the implied cost of capital estimated using analyst estimates. CoD is measured as the log of the loan
spread.
Total number of environmental strengths
and concerns, and a net measure defined as
concerns minus strengths.
Panel regression Both investors and lenders expect significantly higher returns for
firms that have higher net environmental concerns.
Connors & Silva-Gao (2008)
2001-2007 US 326 observations
The implied CoE-models of Gebhardt et al., Gode and
Mohanran, and Easton.
Annual TRI emissions divided by sales.
Panel regression For chemical companies, there is a positive relationship between risk
premium and TRI emissions; for electric companies, no significant
relationship could be found.
El Ghoul et al. (2011)
1992-2007 US 12,915 observations
Implied cost of equity using four different models: Environmental strengths minus environmental concerns (KLD) Pooled cross-sectional time-series regressions
with year and
industry fixed effects Fisher-Vanden & Thorburn (2011) 1993-2008 117 announcements
Abnormal return Announcements to participate in voluntary corporate environmental
initiatives, which aims to improve environmental
performance
Event study Firms announcing that they join a voluntary environmental reduction
program face significant losses in market value.
Gallego-Álvarez, Segura
& Martínez-Ferrero (2015)
2006-2009 21 countries 89 firms ROA and ROE Emission variation (annual percentage change in emissions)
Panel regression with random
effects
Emission reduction positively affects financial performance
(ROE), but do not affect operational performance (ROA).
Goss & Roberts (2011)
1991-2006 US 3996 loans CoD is measured as the log of the initial all-in drawn spread over
the LIBOR rate.
Separate scores for total strengths and total concerns (data from
KLD)
OLS regression Higher concerns are associated with higher spreads, while the coefficient for strengths is not
significant.
Gupta (2018) 2002-2012 43 countries; US, UK, Japan
and Canada dominate the
sample.
23,301 observations
CoE is measured as the average of four implied CoE models:
Claus and Thomas, Easton, Gebhardt et al., and Ohlson and
Juettner-Nauroth
Self-constructed environmental sustainability index,
where all positive indicators are first added
and then divided by the total number of
attributes.
Panel regression with firm- and year-fixed effects.
Gupta finds a negative relationship between environment-friendly
practices and the CoE.
Jacobs, Singhal, & Subramanian
(2010)
2004-2006 US and Europe 780 announcements
Abnormal returns Announcements of voluntary emission
reductions
Event study Announcements of voluntary emission reductions are followed
leads to significantly negative abnormal returns.
Kim, An, & Kim (2015)
2007-2011 South Korea 379 observations
Implied CoE calculated using Ohlson and Juettner-Nauroth’s
model and the Easton model
Carbon intensity (GHG emissions divided by
sales)
Regression There is a positive association between carbon intensity and CoE,
and this relationship is more prominent in industries with lower
GHG emissions. Lewandowski (2017) 2003-2015 International sample. 7,625 observations
Financial performance is a vector containing ROA, ROE, ROS,
ROIC, and Tobin’s q.
Carbon performance is a vector containing annual
emissions and improvements in emissions Panel regression with firm-fixed effects.
Lewandowski finds a curvilinear, U-shaped relationship between carbon and financial performance.
Moreover, changes in carbon performance is significantly negative associated with Tobin’s q.
Li, Eddie & Liu (2014)
2006-2010 Australia 1050 observations
CoD is calculated by dividing the total pre-tax interest expenses over the total long-term debt. CoE is measured as the implied
CoE using Easton’s model.
Emission intensity (carbon emissions divided by sales)
Regression There is a significant and positive relationship between emission intensity and CoD. For CoE, this
relationship is not significant.
Ng & Rezaee (2015) 1991-2013 All firms included in the KLD database (i.e., mainly US) 13,745 observations
CoE is measured as the industry-adjusted EP ratio and as the implied CoE following Gordon
and Gordon.
Environmental strengths minus environmental
concerns (data from KLD)
Lead-lag regression with industry and
year-fixed effects.
Net environmental score has a significantly negative relationship
with cost of equity.
Nishitani & Kokubu (2012) 2006-2008 Japan 1,888 observations Tobin’s q GHG emission reduction is proxied as
net sales divided by emissions.
The random effect instrumental variable model
Oikonomou, Brooks & Pavelin (2014)
1991-2008 US 3,240 bonds CoD is measured by corporate spreads and corporate bond
ratings
The authors add the scores of environmental
strengths/concerns and divide by the number of
relevant indicators to arrive at a measure of environmental strengths/concerns. Fixed approach panel regression.
Environmental concerns are linked with higher yield spreads and bond ratings; environmental strengths are linked with lower bond ratings,
but higher yield spreads. All these results are not significant.
Sharfman & Fernando (2008)
2001-2002 US Not discussed WACC, where CoE is measured using the CAPM and CoD is measured as the firm’s marginal cost of borrowing and is based on
estimates from the Bloomberg Financial Dataset.
KLD’s environmental strengths and concerns are averaged separately.
Hierarchical regression
The authors find a significantly positive relationship between environmental risk management
(EVM) and the CoD, but a negative relationship between EVM and the CoE and WACC.
Trinks et al. (2017) 2002/2008-2016 International sample; US, UK and Japanese firms comprise roughly a half of the sample. 1,920 firms and 32,592 observations
CoE is measured using the CAPM
GHG intensity is measured as total GHG emissions divided by net
sales.
Panel fixed-effects regression
Appendix 2: Distributions of years and industries
Table A.2: Distribution of years
Year S&P 500 Nikkei 225
2009 161 145 2010 219 152 2011 245 158 2012 269 159 2013 251 167 2014 245 172 2015 249 167 2016 238 161 2017 34 119 Total 1911 1400
Table A.3: Distribution of industries
Industry S&P 500 Nikkei 225
Oil & Gas 134 20
Basic Materials 95 199 Industrials 302 420 Consumer Goods 280 290 Health Care 163 81 Consumer Services 221 92 Telecommunications 20 31 Utilities 196 44 Financials 272 118 Technology 228 105 Total 1911 1400
Appendix 3: Descriptive statistics
Table A.4: Summary statistics of key EP variables and financial variables for S&P 500 and Nikkei 225 (in percentage changes)
S&P 500
Emission variables
N Mean Median SD Min Max Skewness Kurtosis Total emissions 1911 0.0087 -0.0120 0.2243 -0.6205 1.3503 2.54 16.95 Direct emissions 1624 0.0385 -0.0081 0.3382 -0.6575 2.0526 3.19 18.32 Indirect emissions 1546 0.0074 -0.0156 0.2565 -0.7708 1.4469 2.07 14.85
Financial variables
N Mean Median SD Min Max Skewness Kurtosis CoD 2908 0.0497 -0.0248 0.4922 -0.7565 3.3301 4.12 26.03 CoE 2949 -0.0081 -0.0260 0.1385 -0.3044 0.5271 0.99 5.33 WACC 2308 -0.0157 -0.0220 0.1775 -0.8244 0.5979 -0.48 8.35 MTB 2971 0.0703 0.0522 0.3912 -1.0362 2.1524 1.83 12.22 Size 3017 0.0036 0.0024 0.0092 -0.020 0.0509 2.09 11.85 Leverage 2935 0.0644 -0.0027 0.3856 -0.5657 2.7399 4.47 29.26 ROA 3017 -0.0487 -0.0187 1.4651 -7.3696 6.7593 -0.24 14.97 Nikkei 225 Emission variables
N Mean Median SD Min Max Skewness Kurtosis Total emissions 1400 0.0468 0.0001 0.31 -0.6013 2.0882 3.88 24.79 Direct emissions 639 0.0383 -0.0157 0.4437 -0.7716 2.9667 4.47 28.48 Indirect emissions 600 0.0479 0.0028 0.3071 -0.6285 2.3959 5.04 38.89
Financial variables
N Mean Median SD Min Max Skewness Kurtosis CoD 1623 -0.0266 -0.0598 0.3296 -0.7440 1.7788 2.29 13.24 CoE 1602 -0.0156 -0.0398 0.1986 -0.4792 0.8965 1.63 8.81 WACC 1212 -0.0644 -0.0376 0.4147 -2.6668 0.9355 -3.28 21.90 MTB 1652 0.0278 -0.005 0.2485 -0.4497 0.9106 0.8886 4.3667 Size 1661 0.0011 0.0011 0.0039 -0.0103 0.0134 -0.05 4.12 Leverage 1645 0.0339 -0.0265 0.3324 -0.5297 2.1588 3.78 22.82 ROA 1661 -0.1867 -0.0741 2.1914 -9.5659 10.0791 0.19 12.77
Appendix 4: Regression results with direct and indirect emissions
Table A.5: Lagged direct emissions: regression results S&P: CoE S&P: CoD S&P:
WACC Nikkei: CoE Nikkei: CoD
Nikkei: WACC Constant -0.0037 (0.0078) 0.2246*** (0.0537) 0.0129 (0.0083) -0.0377** (0.0160) -0.0219 (0.0267) -0.0547*** (0.0182) Direct emissions 0.0009 (0.0093) 0.0077 (0.0299) 0.0072 (0.0110) -0.0016 (0.0079) -0.0195 (0.0247) -0.0027 (0.0173) MTB -0.0101 (0.0080) 0.0090 (0.0267) 0.0076 (0.0094) -0.1582*** (0.0434) -0.0297 (0.0542) 0.0997 (0.0908) Size 0.1541 (0.3761) -7.4187*** (1.2890) -2.2985*** (0.5681) 0.2325 (2.4887) -18.3912*** (4.1067) -3.9510 (4.1208) Leverage 0.0030 (0.0072) -0.3452*** (0.0498) -0.0602*** (0.0124) -0.0206 (0.0225) -0.3867*** (0.0550) -0.1314*** (0.0406) ROA -0.0014 (0.0018) 0.0051 (0.0075) 0.0075 (0.0064) 0.0012 (0.0036) 0.0022 (0.0082) 0.0247** (0.0127) Year effect Yes Yes Yes Yes Yes Yes Industry Yes Yes Yes Yes Yes Yes Adj. R-squared 0.4126 0.1191 0.3911 0.1492 0.1906 0.1490 F-statistic 52.92 10.97 40.62 5.62 7.16 4.7873 N 1533 1549 1296 555 551 455
This table shows the fixed-effects regressions of cost of capital measures on lagged direct emission reductions and the control variables. CoE is measured using the CAPM. CoD is interest expense divided by long and short-term debt. Direct emissions are the one-year lagged annual percentage change in total emissions. MTB is market value of equity to book value of equity. Size is the log of total assets. Leverage is total debt over total assets. ROA is return on assets. Robust standard errors clustered at the firm-level (cross-section) are in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
Table A.6: Direct emissions with no lag: regression results S&P: CoE S&P: CoD S&P:
WACC Nikkei: CoE Nikkei: CoD
ROA -0.0023 (0.0016) 0.0169 (0.0108) 0.0139** (0.0064) 0.0008 (0.0033) 0.0043 (0.0066) 0.0317** (0.0124) Year effect Yes Yes Yes Yes Yes Yes Industry Yes Yes Yes Yes Yes Yes Adj. R-squared 0.3347 0.1307 0.3405 0.1143 0.1885 0.1209 F-statistic 36.84 11.65 31.53 4.63 7.55 4.07 N 1568 1569 1302 620 622 493
This table shows the fixed-effects regressions of cost of capital measures on direct emission reductions with no lag and the control variables. CoE is measured using the CAPM. CoD is interest expense divided by long and short-term debt. Direct emissions are the contemporaneous annual percentage change in total emissions. MTB is market value of equity to book value of equity. Size is the log of total assets. Leverage is total debt over total assets. ROA is return on assets. Robust standard errors clustered at the firm-level (cross-section) are in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
Table A.7: Lagged indirect emissions: regression results S&P: CoE S&P: CoD S&P:
WACC Nikkei: CoE Nikkei: CoD
Nikkei: WACC Constant -0.0071 (0.0070) 0.2342*** (0.0517) 0.0084 (0.0074) -0.0328** (0.0150) -0.0109 (0.0246) -0.0403** (0.0187) Indirect emissions 0.0094 (0.0098) 0.1069* (0.0579) 0.0355** (0.0141) 0.0009 (0.0150) -0.0070 (0.0396) 0.0162 (0.0176) MTB -0.0095 (0.0082) -0.0023 (0.0275) 0.0047 (0.0094) -0.1570*** (0.0452) -0.0135 (0.0582) 0.0876 (0.0943) Size 0.3099 (0.3791) -7.4803*** (1.3422) -2.3477*** (0.5864) 0.0274 (2.6759) -18.7962*** (3.9247) -5.2755 (4.2778) Leverage -0.0002 (0.0070) -0.3517*** (0.0499) -0.0599*** (0.0122) -0.0201 (0.0228) -0.3738*** (0.0563) -0.1296*** (0.0403) ROA -0.0023 (0.0018) 0.0081 (0.0066) 0.0086 (0.0074) 0.0029 (0.0038) 0.0013 (0.0090) 0.0305** (0.0127) Year effect Yes Yes Yes Yes Yes Yes Industry Yes Yes Yes Yes Yes Yes Adj. R-squared 0.4016 0.1268 0.3963 0.1548 0.1900 0.1697 F-statistic 48.18 11.16 39.33 5.49 6.72 5.20 N 1477 1470 1227 516 513 433
Table A.8: indirect emissions with no lag: regression results S&P: CoE S&P: CoD S&P:
WACC Nikkei: CoE Nikkei: CoD
Nikkei: WACC Constant -0.0239*** (0.0072) 0.2988*** (0.0558) -0.0091 (0.0077) -0.0300* (0.0158) 0.0418 (0.0479) -0.0461** (0.0223) Indirect emissions 0.0282** (0.0111) 0.0841 (0.0601) 0.0245 (0.0152) -0.0377* (0.0207) 0.1049 (0.0821) 0.0402 (0.0348) MTB -0.0135* (0.0075) 0.0228 (0.0337) 0.0107 (0.0088) -0.0903** (0.0363) -0.0292 (0.0560) 0.1406 (0.0893) Size -0.3214 (0.3957) -6.9216*** (1.3752) -1.8683*** (0.5702) 0.3670 (2.2600) -18.1440*** (3.5845) -1.7128 (4.7757) Leverage -0.0036 (0.0068) -0.3574*** (0.0490) -0.0592*** (0.0115) -0.0422* (0.0239) -0.4005*** (0.0533) -0.1846*** (0.0492) ROA -0.0030* (0.0017) 0.0213** (0.0110) 0.0161** (0.0072) 0.0011 (0.0032) 0.0061 (0.0067) 0.0371** (0.0135) Year effect Yes Yes Yes Yes Yes Yes Industry Yes Yes Yes Yes Yes Yes Adj. R-squared 0.3371 0.1300 0.3449 0.1169 0.1926 0.1097 F-statistic 35.46 11.12 30.36 4.49 7.32 3.63 N 1492 1491 1228 582 584 472
