“The Effect of Climate Risk on Stock Returns
of European Firms”
Author: Amber Croon (11796154)
Amsterdam June, 2020
Supervisor: Abdulkader Kaakeh
Bachelor Thesis (6013B0520Y) BSc Finance
University of Amsterdam
Business and Business Economics
Statement of Originality
This paper is written by Amber Croon who declared to take full responsibility for the
contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Recently, climate consciousness is growing around the world. This has led to a global movement that aims to reduce climate risk and thus reducing carbon emissions. As a
consequence, these climate risks might be priced, which could lead to different valuations of firm performance. In this study, an empirical analysis on cross-sectional data is performed to test whether climate risk has an impact on monthly returns of European companies in 2019. Furthermore, it is tested whether this impact of climate risk on returns is higher for
companies that emit more carbon and for companies that are located in countries with higher GDP, separately. Data is collected from the Eikon and Factset databases and consist out of financial variables of 335 European firms in 2019.
No evidence was found for the overall effect of climate risk upon monthly stock returns. Moreover, no evidence was found that suggests that the effect of climate risk on monthly returns is higher for companies that emit relatively more carbon. Finally, this study has not found any higher carbon premium for companies that are located in countries with relatively high GDP than for companies located in countries with relatively low GDP.
Table of Contents
1. Introduction ... 5-6 2. Literature review ...7-12
2.1 Climate risk ... 7
2.1.1 What is climate risk? ... 7-8 2.1.2 What is the importance of climate risk?... 8-9 2.1.3 How do we measure climate risk? ... 9-10
2.2 Effect of climate risk on returns ... 10-11
2.2.1 Carbon premium ... 10 2.2.2 Hedging climate risk ... 11
2.3 Hypotheses ... 11-12
3. Data ... 13-16
3.1 Data gathering... 13 3.2 The control variables ... 13-16
4. Methodology ... 17-18
4.1 The regression models ... 17-18
5. Results... 19-21
5.1 Validity of the models ... 19 5.2 Regression models ... 19-21
6. Conclusion ...22 7. Review and limitations ...23 8. References ... 24-26 9. Appendix ...27
1. Introduction
Over the past decades, awareness of global warming and the complications it brings to society is growing. Carbon emissions are increasing exponentially, which leads to a higher occurrence of natural catastrophes. Therefore, climate change mitigation is a popular topic of discussion in the media and politics. In 2015, 189 political leaders signed the Paris
Agreement, committing to reducing carbon emission in order to limit the risks and consequences of climate change. The Paris Agreement also aims to promote a transition towards the use of more renewable sources of energy by corporations.
Research has shown that carbon emission not only brings along environmental risks, but also financial risks. For example, transitioning to more sustainable sources of energy is more expensive for high emission companies than for low emission companies. Therefore, firms that emit more carbon are exposed to higher volatility and as a result expect to deliver higher returns (Fama & French, 1993). This relationship between carbon emission and stock returns has been labelled the “Carbon Premium” by Bolton and Kacperczyk (2020).
In their study, Ceccarelli, Ramelli, and Wagner (2019) introduced a green label called Low Carbon Designation (LCD) and awarded it to companies that met certain environmental criteria. After the awarding, they found that European fund flows into the “green” companies were higher than US fund flows. This study concludes that European investors are more aware of climate change and its consequences than US investors and therefore invest more in environmentally friendly companies. Is this increasing consciousness with respect to climate change impacting financial performance for European firms? The main question of this research is: “Is there a return premium associated with exposure to climate risk for European
firms in 2019?”.
This study tries to contribute to the existing literature by increasing the knowledge about the effect of carbon emissions on stock returns. Moreover, this study aims at expanding knowledge of the impact of climate risk on returns for investors. To investigate this carbon premium, data of firms in the Europe STOXX’s 600 index will be used.
The remainder of this research is structured in the following way: Section 2 includes a review that elaborates on the existing literature about climate risk and the effect of climate risk on returns. In section 3 and 4, an overview of the data employed in the analysis and the methodology will be provided. In section 5, the results from the data analysis will be given, whereas in section 6 those results will be elaborated upon and concluded. Section 7 consists
of a review of the limitations of the analysis and suggestions for further research. In section 8 the reference list is provided and lastly, section 9 include the appendices.
2. Literature Review
To be able to answer the research question, if investors receive a premium for vulnerability to climate risk, this section includes a description of climate risk, the importance of climate risk and how it is measured. Furthermore, we elaborate upon existing evidence on the impact of environmental hazards on returns. Finally, the hypotheses are discussed.
2.1 Climate Risk
2.1.1 What is Climate Risk?
Global warming has been a subject of discussion which receives a great amount of public attention. Global warming is the phenomena of the growing increase of the Earth’s average temperature, leading to several issues. Görgen, Jacob, Nerlinger, Riordan, and Wilkens (2019) constructed an advanced categorization of these environmental hazards, the so-called “TRIP framework”.
Table 1: TRIP framework (Görgen et al., 2019)
Firstly, the authors divide climate risk into two categories: Physical - and transition risk. Physical risk, in turn, is further classified into resource availability and impact of physical damages, and transition risk is split into technology - and policy risk. Resource availability incorporates all climate changes that have a long-term effect on investments. Impact of physical damages comprises all physical damages caused by natural catastrophes such as earthquakes etc. Technology risk covers the hazards associated with transitioning to more renewable resources, for instance, transitioning from carbon to energy inputs. Lastly, political risk includes the measures taken by national and international institutions to mitigate climate change. Because all these risks are associated with specific costs, these
environmental hazards in turn lead to financial risks.
Choi, Gao, and Liang (2019) discuss the potential danger of not accounting for climate risk, which could lead to a so-called “Carbon Bubble”. This carbon bubble is
characterized by carbon intensive firms being overvalued. Therefore, Choi et al. (2019) argue that this environmental risk factor should be priced in. In section 2.2 we discuss the effect of climate risk on returns.
2.1.2 What is the Importance of Climate Risk?
To analyse the importance of climate risk, recent evidence on holdings of socially responsible investments (SRI’s) and social norms in financial markets will be discussed.
Especially after the Paris Agreement in 2015, which was signed by political leaders of 189 different parties, consciousness for global climate change mitigation grew. In their study, Monasterolo and De Angelis (2019) investigated the changes of portfolio composition after the Paris Agreement. It was found that, after the Paris Agreement, the systematic risk of low-carbon indices decreased significantly. Moreover, after the Paris Agreement, the average weight of low-carbon indices within portfolios increased significantly. These results could imply that investors are considering low-carbon indices as an attractive and profitable investment and that the consciousness for climate change mitigation is growing.
In another research, Riedl and Smeets (2017) conducted a survey among holders of socially responsible (SRI) mutual funds. They found that investors expect to receive lower compensation when holding SRI mutual funds in comparison with conventional mutual funds. Therefore, they concluded that investors are willing to sacrifice return in order to hold socially responsible investments.
More research has been done with respect to social preferences. In 2018, the investment research firm, Morningstar, published a new eco-label called the Low Carbon Designation (LCD) (Ceccarelli et al., 2019). Mutual funds are able to receive this label when successfully meeting certain environmental criteria. The authors found that mutual funds that were awarded the LCD label, received significantly more capital inflows than conventional mutual funds. Moreover, fund managers rebalanced their portfolios to more environmental friendly firms in order to be able to receive the LCD label. This evidence shows the current increasing awareness of environmental changes.
Furthermore, a study has been conducted on social norms on markets trading “sin” stocks. Sin stocks are stocks related to firms engaged in the production of tobacco, alcohol and gaming (Hong & Kacperczyk, 2009). The authors found that institutions with higher social norms, such as pension funds, are holding less of these sin stocks in comparison with conventional mutual funds. Hence, Hong, and Kacperczyk (2009) conclude that social norms
affect stock prices and returns. In accordance with these results, Chava (2014) found that environmental unfriendly firms have lower institutional ownership than environmental friendly firms. Dimson, Karakas and Li (2015) also found that enrolling in successful
environmental engagements raises institutional ownership. These findings suggest that social norms are considered when making portfolio composition decisions, especially for
institutions. Another research has been employed on the effect of institutional ownership on carbon emission and economic growth (Bhattacharya, Churchill & Paramati, 2017). It was found that institutional ownership has a negative effect on carbon emissions, and a positive effect on economic growth, both have found to be significant. Moreover, it was found that investors demand a higher expected return from companies that do not make socially responsible investments (Chava, 2014). In section 2.2 we elaborate upon the impact of environmental hazards on returns in more detail.
To evaluate the importance of climate risk, Bansal, Ochoa, and Kiku (2017) constructed a long-run risk model that includes interaction effects of risk, temperature and economic growth. They make use of this model to try to estimate the social cost of carbon (SCC). In this study, it was found that the rising temperature together with the uncertainty on economic development lead to a serious social cost of carbon and a loss of utility. Hence, the authors stress the importance of immediate carbon emission reduction. The current research tries to contribute in raising awareness for mitigating climate change.
2.1.3 How do we Measure Climate Risk?
To be able to quantify the impact of climate risk on returns, a variable for climate risk has to be developed.
Several ratings have been developed over the last decades. Corporate social responsibility rating (CSR) is a commonly used rating, which evaluates the integration of certain values into a firm. These values include a range of topics, such as social, ethical, environmental and supply chain (Liang & Renneboog, 2017). This is a very wide ranged variable that includes more than only environmental risks.
A somewhat better measurement for climate risk is the eco-label called “LCD” (Low Carbon Designation), which was previously mentioned. This label was introduced by
financial services company Morningstar and solely focuses on the carbon intensity level of the company in question and, therefore, is very effective in quantifying climate risk
use carbon footprint data per company to construct indices with minimal exposure to climate risk. Therefore, in this study, carbon emission data is the most appropriate measurement for climate risk and carbon emission data will be used in the analysis.
2.2 Effect of Climate Risk on Returns
In this section existing evidence on the effect of climate risk on returns will be discussed. Moreover, strategies that minimize exposure to environmental hazard will be elaborated upon.
2.2.1 Carbon Premium
When investing in highly volatile assets, investors are expected to receive
compensation for the risk they face. This highly evidenced positive relationship between risk and return is intensively discussed by Fama and French (1993) in their research about the risk factors on stocks and bonds.
As for any type of risk, exposure to carbon risk is expected to incorporate some type of compensation, so-called “carbon premium” (Bolton & Kacperczyk, 2020). Many recent studies aimed to expose this premium for carbon risk. Hsu, Li & Tsou (2019) compared high- and low emission companies and found a significantly lower average return in low emission companies. In another research, Bolton and Kacperczyk (2020) performed an analysis in a sample consisting of 14.400 companies from 77 countries. They also found significantly higher returns in firms with higher carbon emission. From these results could be assumed that investors do indeed earn a compensation for exposure to carbon risk.
Furthermore, Bolton and Kacperczyk (2020) concluded that carbon premiums are present at firm-level and rising in line with the increasing awareness of the need to reduce carbon emissions. In another research, Ceccarelli et al. (2019) found that, after rewarding firms that met certain environmental criteria with the LCD label, European fund flows into the rewarded firms are twice as large as US fund flows. Therefore, the authors concluded that European fund managers reacted more on the desires of their investors than US fund
managers. This study will try to give more insight on the variation of carbon premiums at the firm- and country- level.
2.2.2 Hedging Climate Risk
In order to try to circumnavigate climate risk, several studies have been done on the performance of “green” funds in comparison to conventional funds. Ibikunle and Steffen (2017) perform a comparative analysis of green funds against conventional- and black funds, and their performance. Black funds invest in companies with high carbon intensity
production. Over a time period of 1991-2014 they have found that conventional funds significantly outperform green funds. However, surprisingly, green funds seemed to start outperforming black funds over the sample period of 2012-2014. This could sign a global movement towards more renewable resources and awareness of the importance of carbon emission reduction. Another research has been conducted on the possibility of hedging climate risk. Hedging of climate risk is seen as successful when exposure to carbon risk is low relative to the benchmark (Andersson et al., 2016). In this study, the authors constructed low-carbon indices and compared their performance to benchmark indices. They have shown that low-carbon indices achieve the same compensation as benchmark indices as long as environmental actions are awaiting. However, once the carbon emission data is priced, the low-carbon indices are expected to outperform the benchmark indices. These results could trigger investors to rebalance their portfolios to climate conscious companies.
2.3 Hypotheses
Following from the existing literature, this section describes the hypotheses.
Ceccarelli et al. (2019) concluded from their research that European fund flows were twice as large as US fund flows, after rewarding those companies with the LCD label. Furthermore, according to Schreurs and Tiberghien (2017), the majority of European countries positioned themselves as agenda setters with respect to environmental mitigation policies. Therefore, firms in countries with high climate change consciousness and high movement to more climate-friendly resources are expected to have priced in the carbon risk. Furthermore, Bolton and Kacperczyk (2020) and Hsu et al. (2019) found a significant compensation in return for carbon emission around the world. This leads to the following hypothesis:
𝐻1: There is a carbon premium for European firms.
Moreover, Görgen et al. (2019) state that firms that have higher carbon emission are harmed the most when transitioning to more renewable resources. This also leads to higher
financial risks. Chenet, Thomä and Janci (2015) also state that a large amount of oil, gas and coal will not be burned in a low emission world following from the Paris Agreement. Hence, this will lead to a substantial carbon price, impacting financial performance. According to Fama and French (1993), this increased volatility, due to climate risk, is expected to drive up returns. Hence the following hypothesis is constructed:
𝐻2: The carbon premium is higher for European firms that have high carbon emission than for European firms that have low carbon emission.
Finally, the influence of GDP per country on carbon premium is tested. From their research, Heil and Selden (2001) conclude that high-income countries are responsible for the highest fraction of carbon emission. Acaravci and Ozturk (2010) also exposed this positive relationship between CO2 emissions an economic growth in Europe. It is expected that firms located in countries with high carbon emission have priced in this climate risk factor.
Therefore, the following hypothesis is formulated:
𝐻3: The carbon premium is higher for firms located in European countries with high
3. Data
This section will discuss the method of data collection and gives a description of the data used in the analysis. Furthermore, an explanation of the variables is included.
3.1 Data Gathering
This study focuses on carbon emission data and the impact on monthly returns. In order to answer the research question, if there is a return premium associated with exposure to climate risk, a model is developed with stock returns in 2019 as dependent variable. The stock returns are calculated by means of the change of stock price over 2019 in percentage points. The data used in this analysis is collected from 3 different databases. The carbon emission data essential in this study is gathered from Eikon/Datastream. Data on GDP per capita is collected from the IMF database and the remainder of financial data is collected from Factset. From all databases, data is gathered for the STOXX’s Europe 600 database, which consists of nearly 600 companies originating in 17 European countries with small- mid- and large capitalization in 2019. After removing observations with lacking data, 335 companies remain for the analysis.
3.2 The Control Variables
In line with Fama and French (1993), Bolton and Kacperczyk, (2020) and Hsu et al. (2019), the following determinants of stock returns next to our main independent variable carbon emissions are included: market capitalization, price-to-book ratio, leverage, return-on-equity, investment rate and asset growth. This section will discuss the main explanatory variable, the control variables and their effect on returns according to the existing literature. Table 2 presents the descriptive statistics of the model.
Table 2: Descriptive Statistics
Variable Obs Mean Std.Dev. Min Max Stock return 335 25.172 24.537 -29.878 94.433 Market Capitalization 335 23219.56 30897.27 2439.084 148000 Price-to-book 335 3.404 3.32 .322 17.446 Leverage 335 1.229 1.303 .018 6.842 ROE 335 16.629 12.381 -3.091 67.462 Total Cap Exp % Total
Assets (in 335 3.331 2.936 .009 12.355 Asset Growth 335 10.487 18.319 -17.699 109.405 Total carbon emission 335 3870000 1.24e+07 671 8.20e+07 GDP per capita (in bln) 335 2210000 1460000 36915 4440000
carbon_high 335 .128 .335 0 1
gdp_high 335 .561 .497 0 1
lnmktcap 335 9.471 1.042 7.647 12.567
Carbon Emission
The independent variable of interest in this research is the carbon emission. The dataset includes several measures for carbon emissions. For this analysis, the total carbon emission per company in 2019 is included. This variable consists of all direct and indirect emissions from firm activity. Although this measure of carbon emission is the most reported one, not all firms included in the STOXX’s Europe 600 index have data on this variable. Firms that lack information on any variable are excluded from the analysis, which leaves a total of 335 companies for the regression. To test the first hypothesis, if European firms have a significant carbon premium, carbon emissions will be regressed on returns including the control variables. In line with the first hypothesis, it is expected that the coefficient of carbon emission will be positive.
In order to test the second hypothesis, if carbon-intensive firms have higher carbon premium than low-carbon firms, one dummy variable will be created where carbon_high equals 0 represents total emissions < average emissions and high equals 1 represents total emissions > average emission. Hsu et al. (2019) found that high emission firms deliver a higher average return than low emission firms. Therefore, the expected outcome of this regression of carbon emission on returns including control variables is that the carbon emission coefficient is higher for high-carbon firms than for low-carbon firms. Table 2A represents the distribution of the dummy variable carbon_high.
Market Capitalization
Market capitalization is defined as the number of shares outstanding multiplied by the share price. According to Banz (1981) and Fama and French (1992), market capitalization has a negative effect on returns. It was found that smaller companies on average have a higher risk adjusted return. Market capitalization is winsorized at 1st and 99th percentiles to
minimize the effect of outliers.
Price-to-Book Ratio
Studies conducted by Bolton and Kacperczyk (2020) and Hsu et al. (2019), found that price-to-book ratio has a positive impact on stock return. The reason for this is that an
increasing market price leads to a higher price-to-book-ratio which in return increases the stock returns. For this reason, the price-to-book ratio is included as a control variable in the current model. Price-to-book ratio is winsorized at 1st and 99th percentiles to reduce the
impact of outliers.
Leverage
Leverage is defined by the value of debt divided by the book value of the assets. Previous research showed that leverage has a negative effect on stock returns (Bolton and Kacperczyk, 2020; Hsu et al., 2019). Therefore, leverage is included as control variable in the analysis. Furthermore, leverage is winsorized at 1st and 99th percentiles to mitigate the effect
of outliers.
Return-on-Equity
Return on equity (ROE) is calculated by dividing net yearly income by the value of the equity. Both Bolton and Kacperczyk (2020) and Hsu et al. (2019) found that ROE has a positive effect on stock returns. A possible explanation for this result is that higher earnings result in a higher ROE, which in turn leads to higher stock returns. As a result, ROE is included in the analysis as control variable, and winsorized at 1st and 99th percentiles to
reduce the effect of outliers.
Investment Rate
Investment rate is defined as capital expenditures divided by total assets. In their research, Bolton and Kacperczyk (2020) as well as Hsu et al. (2019) have shown that
regression analysis. Furthermore, investment rate is winsorized at 1st and 99th percentiles to
minimize the impact of outliers.
Asset Growth
Asset growth is calculated by the percentage change in assets compared to previous year. Hsu et al. (2019) found that asset growth has a positive impact on stock returns.
Therefore, asset growth is added as a control variable in the analysis and winsorized at 1st and
99th percentile to reduce the effect of outliers.
GDP
Gross Domestic Product (GDP) measures the value of all finished goods in a specific country (Acaravci & Ozturk, 2010). To be able to test the third hypothesis, if firms located in countries with higher GDP have higher carbon premium than firms located in countries with lower GDP, GDP for every European country is included. Assignment of countries to firms is performed based on the location of primary exchange. A dummy variable is created where gdp_high equals 0 for firms located in countries of primary exchange where GDP < average GDP and gdp_high equals 1 for firms located in countries of primary exchange where GDP > average GDP. In line with the hypothesis, the expected outcome of the regression is that the coefficient of carbon emission is higher for high-GDP countries than for low-GDP countries. Table 2B graphs the distribution of the gdp_high dummy variable.
4. Methodology
In this section, the various regression models employed in the analysis will be discussed.
4.1 The Regression Models
To test the hypotheses, an OLS regression analysis on cross-sectional data is
performed in Stata. This model is constructed based on existing literature on the determinants of stock returns. After checking for multicollinearity, the variables discussed in the previous section are included in the regression model as control variables. The correlation test is represented in Table 3.
Table 3: Multicollinearity Test
Furthermore, White’s test is used to test for heteroscedasticity (Williams, 2020). Table 4 represents the outcome of the White’s test. The test statistic is significant, hence there is heteroscedasticity. For the following regression analysis robust standard errors will be used.
Table 4: White’s Test for Heteroscedasticity
Moreover, the natural logarithm of the independent variables is used in the regression to limit heteroscedasticity and for interpretation reasons.
VIF 1/VIF Ln(PTB) 3.604 .278 Ln(totalemission) 2.721 .367 Ln(investment) 2.457 .407 Ln(ROE) 2.204 .454 Ln(mktcap) 1.553 .644 Ln(assetgrowth) 1.219 .821 Ln(leverage) 1.123 .891 Mean VIF 2.126 .
In this research, the following regression model will be tested:
𝛾𝑖 = 𝛽0 + 𝛽1𝑇𝑜𝑡𝑎𝑙𝑒𝑚𝑖𝑠𝑠𝑖𝑜𝑛 + 𝛽2𝐿𝑜𝑔(𝑀𝑘𝑡𝑐𝑎𝑝) + 𝛽3𝑃𝑟𝑖𝑐𝑒𝑡𝑜𝑏𝑜𝑜𝑘 + 𝛽4𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒
+ 𝛽5𝐸𝑅𝑂𝐸 + 𝛽6𝐼𝑛𝑣𝑒𝑠𝑡𝑚𝑒𝑛𝑡 + 𝛽7𝐴𝑠𝑠𝑒𝑡𝑔𝑟𝑜𝑤𝑡ℎ + 𝜀𝑖
Where 𝛾𝑖 = 𝑠𝑡𝑜𝑐𝑘 𝑟𝑒𝑡𝑢𝑟𝑛 𝑖𝑛 2019 (𝑖𝑛 𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑝𝑜𝑖𝑛𝑡𝑠)
The coefficient of interest in this is study is 𝛽1.
In order to test the hypotheses, the followings regression models will be run:
The first model tests the overall existence of a carbon premium for European firms in 2019. Therefore, total emission and all the control variables are regressed on monthly returns.
The second and third model estimate the existence of a carbon premium in high carbon firms and low carbon firms respectively. Therefore, two regressions will be performed, each including total emissions and all control variables, the former including only companies with high carbon emission and the latter only including companies with low carbon emission.
The third and fourth model test the existence of a carbon premium in high GDP countries and low GDP countries respectively. Therefore, two regressions will be run, each including total emissions and all control variables, the former including only companies with primary exchange in high GDP countries and the latter only including companies with primary exchange in low GDP countries.
5. Results
In this section, the validity of the models and the results of the OLS regressions will be discussed.
5.1 Validity of the Models
Firstly, to evaluate the performance of the overall models, the 𝑅2 will be discussed.
The 𝑅2 measures the goodness of fit of the model. In other words, the value of R-squared
describes the fraction of variation in the dependent variable that is explained by the independent variables. In the regression models in this research the average R-squared is 0.2388. This means that the variation in monthly returns is explained for 23.88% by the included independent variables, which is considered relatively low.
5.2 Regression Models
The results of the OLS regressions are represented in Table 5.
Table 5: Regression Results
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
(1) (2) (3) (4) (5)
VARIABLES C02premium C02_high C02_low gdp_low gdp_high Ln(totalemission) 0.00614 -0.122 0.0302 -0.0251 0.0306 (0.0403) (0.287) (0.0477) (0.0420) (0.0699) Ln(mktcap) -0.0988 -0.384 -0.0693 -0.105 -0.0520 (0.0640) (0.227) (0.0647) (0.0768) (0.117) Ln(PTB) 0.357** 0.241 0.323** 0.355** 0.349 (0.143) (0.518) (0.147) (0.137) (0.257) Ln(leverage) -0.201*** -0.0211 -0.234*** -0.149** -0.263*** (0.0513) (0.284) (0.0497) (0.0710) (0.0765) Ln(ROE) -0.0997 0.914 -0.107 -0.117 -0.119 (0.154) (1.110) (0.161) (0.143) (0.252) Ln(investment) -0.0916** -0.892 -0.0892** -0.100** -0.0732 (0.0435) (0.545) (0.0428) (0.0481) (0.0686) Ln(assetgrowth) 0.108** 0.166 0.0912* 0.0506 0.188 (0.0537) (0.234) (0.0490) (0.0554) (0.128) Constant 3.753*** 7.456 3.297*** 4.420*** 2.808*** (0.646) (5.569) (0.630) (0.825) (1.003) Observations 228 27 201 129 99 R-squared 0.214 0.297 0.210 0.234 0.239
The first model tests the existence of a carbon premium for the whole sample and therefore this regression model tests H1: There is a carbon premium for European firms. Firstly, the inclusion of total carbon emission does not lead to a significant effect on monthly returns. A 1% increase in total emission leads to a 0.006% increase in the monthly returns. However, as mentioned earlier, this result is insignificant. Therefore, the carbon premium Bolton and Kacperczyk (2020) and Hsu et al. (2019) found in their studies cannot be verified in the current study. For this regression model the inclusion of price-to-book ratio, leverage, investment rate and asset growth lead to a significant effect on monthly returns. The signs of the coefficients are in line with the expectations following from previous research (Hsu et al., 2019). A 1% increase in the price-to-book ratio leads to a 0.357% increase of monthly
returns. A 1% increase in leverage leads to a 0.201% decrease of monthly returns. A 1% increase of investment rate leads to a 0.092% decrease in monthly returns and a 1% increase in asset growth leads to a 0.108% increase in monthly returns. In conclusion, we fail to reject H0 and the first hypothesis that there is a carbon premium for European firms cannot be proven.
The second and third model test the existence of a carbon premium for high and low carbon emission firms respectively. Therefore, those regression models test H2: The carbon
premium is higher for firms that have high carbon emission than for firms that have low carbon emission.
For both models, the inclusion of total carbon emission does not lead to a significant effect on monthly returns. Against expectations, a 1% increase in total emission leads to a 0.122% decrease in monthly returns for high emission companies and to a 0.03% increase in monthly returns for low emission companies. Although it was expected that the inclusion of total emission led to a higher effect on returns for high carbon emission companies than for low carbon emission, these coefficients are insignificant so no reliable conclusions can be drawn from these results. Hence, the existence of a carbon premium Bolton and Kacperczyk (2020) and Hsu et al. (2019) found cannot be verified in the current study. For the second regression model that tests a sample of high carbon emission companies no inclusion in any of the variables leads to a significant effect on monthly returns. In the third regression model that analyses a sample of low carbon emission companies, evidence is found of a significant effect on returns for the following variables: price-to-book ratio, leverage and investment rate. A 1% increase in the price-to-book ratio leads to a 0.323% increase in monthly returns. A 1% increase in leverage leads to a 0.234% decrease in monthly returns and a 1% increase in the investment rate leads to a 0.089% decrease in monthly returns. To conclude, we fail to
reject H0 and the second hypothesis that the carbon premium is higher for firms that have high carbon emission than for firms that have low carbon emission cannot be proven.
The fourth and fifth model test the existence of a carbon premium for firms located in high GDP countries and low GDP countries respectively. The results of these regression models test H3: The carbon premium is higher for firms located in countries with high GDP
than for firms located in countries with low GDP.
No evidence of a significant effect of total emission on monthly returns for high- and low GDP countries is found. Against expectations, a 1% increase in total emission leads to a 0.025% decrease for companies located in high GDP countries and to a 0.031% increase for companies located in low GDP countries. However, these coefficients are insignificant. Hence, no reliable conclusions can be drawn from these results and the existence of a carbon premium found by Bolton and Kacperczyk (2020) and Hsu et al. (2019) cannot be verified in the current study. For the model that analyses a sample of firms located in high GDP
countries, the inclusion of price-to-book, leverage and investment rate lead to significant effects on monthly returns, with a significance level,
p < .05. As expected, a 1% increase in price-to-book ratio leads to a 0.355% increase of monthly returns. A 1% increase in leverage leads to a 0.149% decrease in returns and a 1% increase in investment rate leads to a 0.1% decrease in monthly returns. For the regression model that tests a sample consisting of companies located in low GDP countries, the only variable that gives a significant effect on monthly returns is leverage. As expected, a 1% increase in leverage leads to a 0.263% decrease in monthly returns, with a significance level,
p < .001. In conclusion, we fail to reject H0 and the third hypothesis that the carbon premium
is higher for firms located in countries with high GDP than for firms located in countries with low GDP cannot be proven.
6. Conclusion
To conclude, this section will discuss the main findings of this study and give an
interpretation of the results. The research question of this study is the following: Is there a
return premium associated with exposure to climate risk for European firms in 2019?
The goal of this study is to investigate whether a carbon premium was present in a sample of European firms in 2019. However, no significant effect of total emission on monthly returns was found. Hence, no evidence was found that investors receive any extra compensation in returns with respect to exposure to carbon emission, and therefore, climate risk. These results are in contrast with the findings of Bolton et al. (2020) and Hsu et al. (2019). In these studies, the authors found that a carbon premium was present in all sectors over three continents, including Europe.
Furthermore, previous research showed that firms with higher carbon emission have significantly higher stock returns than firms with lower carbon emission (Bolton &
Kacperczyk, 2020; Hsu et al., 2019). This study aims at finding whether this higher return is the effect of higher carbon emission for European firms in 2019. No evidence is found that companies with high carbon emission have higher compensation for carbon emission than companies with low carbon emission. Therefore, it can be concluded that there is no evidence that the higher risk, implied by higher carbon emission companies according to Görgen et al. (2019), is priced in.
Finally, the theory that the carbon premium is highest for firms originating from countries with the highest GDP is tested. According to Heil and Selden (2001), high GDP countries are responsible for the highest fraction of carbon emission. Hence, it is expected that firms located in countries with the highest GDP have priced in this climate risk factor. However, no evidence is found that firms in those countries have a higher premium for exposure to climate risk.
This study is relevant, because it contributes to the existing knowledge about the impact of carbon emission on stock returns. The subject of this research is significant, because awareness of environmental change is growing and the belief of a necessary
7. Review and Limitations
This section will provide a review of the research and explain its limitations
The first and biggest limitation of this research is the quality and lack of data with respect to carbon emissions. According to a recent study of Krueger, Sautner, and Starks (2020), there is a widespread view that improvement on environmental risk disclosure is necessary, because of the lack of reporting standards regarding to carbon emissions. In the database employed in this research, data on carbon emissions was available on about two-third in the sample. To give reliable interpretations with regards to investment/corporate data, a larger sample is desirable. In addition to this, from the survey Krueger et al. (2020)
conducted it seems that investors use all kinds of carbon emission data, all measured
differently, and some even don not use any C02 emission data. This might give biased results from carbon emission analysis. Therefore, the implementation of a more general and accurate way of calculating carbon emissions is necessary to give reliable inferences from investment data with respect to C02 emission. In conclusion, further research on the measurement, quality and improvement of carbon emission disclosure is desirable.
The second limitation of this study is that GDP per country based on country of primary exchange might give unreliable results. To test if carbon premium is higher for firms located in high GDP countries than for firms located in low GDP countries, the country based on primary exchange of the firm is used. To reflect carbon emission per firm resulting from production, it might be more reliable to use country of major production of the firm instead of country of primary exchange in further research.
Thirdly, this research lacks results for the regression model regarding the sample with high carbon emission companies. The reason for this is that the sample size for firms with high carbon emission is only 35, and therefore no reliable inferences can be made from the analysis of such a small sample of firms.
The final drawback of this research is the inclusion of too little determinants of the dependent variable. On average, the R-squared of the regression models was 0.2388, which means that only 23,88% of the total variance in monthly returns was explained by the independent variables. To make the inferences about the models more reliable, the inclusion of more determinants of monthly returns is necessary.
8. References
Acaravci, A., & Ozturk, I. (2010). On the relationship between energy consumption, CO2 emissions and economic growth in Europe. Energy, 35(12), 5412–5420.
https://doi.org/10.1016/j.energy.2010.07.009
Andersson, M., Bolton, P., & Samama, F. (2016). Hedging climate risk. Financial Analysts
Journal, 72(3), 13–32. https://doi.org/10.2469/faj.v72.n3.4
Bansal, R., Ochoa, M., & Kiku, D. (2016). Climate Change and Growth Risks. Working Paper, Duke University. https://doi.org/10.3386/w23009
Banz, R. W. (1981). The relationship between return and market value of common stocks. Journal of Financial Economics, 9(1), 3–18. https://doi.org/10.1016/0304- 405X(81)90018-0
Bhattacharya, M., Churchill, S. A., & Paramati, S. R. (2017). The dynamic impact of renewable energy and institutions on economic output and CO 2 emissions across regions. Renewable Energy, 111, 157–167.
https://doi.org/10.1016/j.renene.2017.03.102
Bolton, P., & Kacperczyk, M. T. (2020). Carbon Premium Around the World. CEPR Discussion Paper No. DP14567. Retrieved from https://ssrn.com/abstract=3594188
Ceccarelli, M., Ramelli, S., & Wagner, A. F., Low-carbon mutual funds (2020). Swiss
Finance Institute Research Paper, 19-13. http://dx.doi.org/10.2139/ssrn.335323
Chava, S. (2014). Environmental externalities and cost of capital. Management
Science, 60(9), 2223–2247. https://doi.org/10.1287/mnsc.2013.1863
Chenet, H., Thomä, J., Janci, D. (2015). Financial Risk and the Transition to a Low-Carbon
Economy. Retrieved from
http://unepinquiry.org/wp-content/uploads/2015/10/2dii_risk_ transition_low-carbon_workingpaper_jul2015.pdf
Financial Studies, 33(3), 1112–1145. https://doi.org/10.1093/rfs/hhz086
Dimson, E., Karakaş, O., & Li, X. (2015). Active Ownership. Review of Financial
Studies, 28(12), 3225–3268. https://doi.org/10.1093/rfs/hhv044
Fama, E. F., & French, K. R. (1992). The Cross-Section of Expected Stock Returns. The
Journal of Finance, 47(2), 427–465.
https://doi.org/10.1111/j.1540-6261.1992.tb04398.x
Fama, E. F., & French, K. R. (1993). Common Risk Factors in the Returns on Bonds and Stocks. Journal of Financial Economics, 33, 3–56. Retrieved from
http://fir.nes.ru/~agoriaev/Papers/Fama-French%20APT%20three-factor%20model%20JFE93m.pdf
Görgen, M., Jacob, A., Nerlinger, M., Riordan, R., Rohleder, M. & Wilkens, M. (2017). Carbon Risk. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.2930897
Heil, M. T., & Selden, T. M. (2001). Carbon emissions and economic development: future trajectories based on historical experience. Environment and Development
Economics, 6(1), 63–83. https://doi.org/10.1017/s1355770x01000043
Hong, H., & Kacperczyk, M. (2009). The price of sin: The effects of social norms on markets. Journal of Financial Economics, 93(1), 15–36.
https://doi.org/10.1016/j.jfineco.2008.09.001
Hsu, P., Li, K., & Tsou, C. (2020). The Pollution Premium. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3578215
Ibikunle, G., & Steffen, T. (2015). European Green Mutual Fund Performance: A
Comparative Analysis with their Conventional and Black Peers. Journal of Business
Ethics, 145(2), 337–355. https://doi.org/10.1007/s10551-015-2850-7
Institutional Investors. The Review of Financial Studies, 33(3), 1067–1111. https://doi.org/10.1093/rfs/hhz137
Liang, H., & Renneboog, L. (2017). On the Foundations of Corporate Social Responsibility. The Journal of Finance, 72(2), 853–910.
https://doi.org/10.1111/jofi.12487
Monasterolo, I., & de Angelis, L. (2020). Blind to carbon risk? An analysis of stock market reaction to the Paris Agreement. Ecological Economics, 170.
https://doi.org/10.1016/j.ecolecon.2019.106571
Riedl, A., & Smeets, P. (2017). Why Do Investors Hold Socially Responsible Mutual
Funds? The Journal of Finance, 72(6), 2505–2550. https://doi.org/10.1111/jofi.12547
Schreurs, M. A., and Tieberghien, Y. (2007). Multi-Level Reinforcement: Explaining European Union Leadership in Climate Change Mitigation. Global Environmental
Politics, 7(4), 19–46. https://doi.org/10.1162/glep.2007.7.4.19
Williams, R. (2020). Heteroscedasticity. University of Notre Dame. Retrieved from https://www3.nd.edu/~rwilliam/stats2/l25.pdf
9. Appendix
Appendix 1: Figures of dummy variable distribution
