The environmental and financial performance of European green energy investments

Julho de 2020 Universidade do Minho Escola de Economia e Gestão Nuno Miguel Santos Andrade - PG36871 The Environmental and Financial Performance of European Green Energy Investments Dissertação de Mestrado Mestrado em Finanças Trabalho efetuado sob a orientação da Professora Doutora Maria do Céu Ribeiro Cortez 2 DIREITOS DE AUTOR E CONDIÇÕES DE UTILIZAÇÃO DO TRABALHO POR TERCEIROS Este é um trabalho académico que pode ser utilizado por terceiros desde que respeitadas as regras e boas práticas internacionalmente aceites, no que concerne aos direitos de autor e direitos conexos. Assim, o presente trabalho pode ser utilizado nos termos previstos na licença abaixo indicada. Caso o utilizador necessite de permissão para poder fazer um uso do trabalho em condições não previstas no licenciamento indicado, deverá contactar o autor, através do RepositóriUM da Universidade do Minho. Atribuição-NãoComercial-SemDerivações CC BY-NC-ND https://creativecommons.org/licenses/by-nc-nd/4.0/ 3 Acknowledgments First of all, I would like to give a special thanks to my supervisor, Prof. Doutora Maria do Céu Cortez, for all the support, motivation and patience towards the difficulties found during the dissertation. I really appreciate all the availability, competence and professionalism that was invaluable to this project. I would like also to thank my friends and colleagues, especially Jorge, for the motivation in finishing this dissertation. A special thanks to my family, especially my parents, for unconditionally supporting and accompanying me through this degree. I would also like to thank my girlfriend, Anabela, for all the company. 4 STATEMENT OF INTEGRITY I hereby declare having conducted this academic work with integrity. I confirm that I have not used plagiarism or any form of undue use of information or falsification of results along the process leading to its elaboration. I further declare that I have fully acknowledged the Code of Ethical Conduct of the University of Minho. 5 Abstract This paper addresses the relationship between environmental and financial performance of energy companies. For this purpose, we compare the performance of green energy portfolios of European stocks compared to their non-green counterparts from 2009 to 2018. Furthermore, we form portfolios based on the dimensions of the Environmental ASSET4 ESG pillar, namely the Environmental Pillar, Emission Reduction, Resource Reduction and Product Innovation dimensions and compare the performance of high-rated portfolios against low-rated portfolios. Our results show that, for the most part, green energy portfolios are very similar to non-green portfolios in terms of performance, with most of our results not showing any statistical difference between green and non-green portfolios. However, when analyzing performance across different sub-periods, namely from 2009 to 2013 and 2014 to 2018, we observe an improvement in abnormal returns. These results suggest that, over time, the performance of green stocks improved but these improvements are not enough to outperform that of non-green portfolios. Regarding the ranked-based portfolios, the results show that, overall, high-rated and low-rated portfolios perform similarly but, when considering a 50% cut-off and when using a sector benchmark (the MSCI Energy EU), as well as in the 2014 to 2018 subperiod, the high-rated portfolio formed on the Resource Reduction dimension significantly outperforms the low-rated portfolio. Overall, our results suggest that investors will not suffer financial penalties by forming portfolios based on green energy screening. Keywords: Environmental performance; Green energy; Green finance; Sustainable investments; ESG Investing. 6 Resumo Este estudo investiga a relação entre o desempenho financeiro e ambiental de empresas de energia na Europa. Com este objetivo, comparamos o desempenho de carteiras de energia verde com carteiras de energia não verde no período de 2009 a 2018. De forma a comparar o desempenho ambiental com o desempenho financeiro, criámos carteiras baseadas nas dimensões do “Environmental ASSET4 ESG pillar”, nomeadamente as dimensões Ambiental, Redução de emissões, Redução de recursos e Inovação de produto. Os nossos resultados mostram que, na maioria dos casos, as carteiras de energia verde são muito semelhantes às carteiras de energia não verde em termos de desempenho financeiro, não havendo na maioria dos casos diferença estatisticamente significativas. No entanto, ao analisar o desempenho dos subperíodos (de 2009 a 2013 e 2014 a 2018), observamos uma melhoria do desempenho. Estes resultados sugerem que, ao longo do tempo, o desempenho financeiro das ações verdes melhoraram, embora não o suficiente para superar de forma significativa as carteiras de ações não verdes. Considerando as carteiras baseadas na classificação ESG, na sua maioria, os nossos resultados mostram que as carteiras “high-rated” e “low-rated” têm um desempenho semelhante, mas quando consideramos um “cut-off” de 50% e o índice do sector de energia (MSCI Energy EU), e também o subperíodo de 2014 a 2018, a carteira “highrated” formada na dimensão de Redução de recursos supera em termos financeiros a carteira “low-rated”. No geral, os nossos resultados sugerem que os investidores não sofrem qualquer tipo de penalizações financeiras ao formar carteiras baseadas em critérios de energia verde. 7 Contents 1. Introduction .................................................................................................... 11 2. Literature Review ............................................................................................ 12 3. Methodology ................................................................................................... 16 4. Data ................................................................................................................ 18 5. Empirical Results ............................................................................................. 22 5.1. Performance of green vs non-green portfolios ....................................................... 22 5.2. Performance of green vs non-green portfolios with ESG Scores ............................ 28 5.3. Performance of environmentally ranked portfolios ............................................... 39 5.4. Environmental performance of green and non-green energy portfolios ............... 64 6. Conclusions ................................................................................................................ 68 References ...................................................................................................................... 71 8 List of tables Table 1 - Descriptive statistics of the green and non-green portfolios considering all firms ................................................................................................................................. 19 Table 2 - Descriptive Statistics of the green and non-green portfolios considering the ESG rated firms. .............................................................................................................. 19 Table 3 - Number of stocks by country and industry: non-green portfolio .................. 20 Table 4 - Number of stocks by country and industry: green portfolio. ........................ 21 Table 5 - Performance of green and non-green portfolios including all firms – unconditional model. ...................................................................................................... 23 Table 6 - Performance of green and non-green portfolios including all firms - conditional model. .......................................................................................................... 26 Table 7 - Performance of green and non-green portfolios including only ESG rated firms - unconditional model ........................................................................................... 29 Table 8 - Performance of green and non-green portfolios including only ESG rated firms - Conditional model. .............................................................................................. 31 Table 9 - Performance of a green portfolio containing all the initial firms and a green portfolio containing only the ESG-rated firms – unconditional model ........................ 33 Table 10 - Performance of a green portfolio containing all the initial firms and a green portfolio containing only the ESG-rated firms - conditional model ............................. 34 Table 10 - Performance of a green portfolio containing all the initial firms and a green portfolio containing only the ESG-rated firms - conditional model (Continued) ......... 35 Table 11 - Performance analysis of a non-green portfolio containing all the initial firms and a non-green portfolio containing only the ESG-rated firms – Unconditional model ............................................................................................................................... 36 Table 12 - Performance analysis of a non-green portfolio containing all the initial firms and a non-green portfolio containing only the ESG-rated firms – Conditional model ............................................................................................................................... 37 Table 12 - Performance analysis of a non-green portfolio containing all the initial firms and a non-green portfolio containing only the ESG-rated firms – Conditional model (Continued) .......................................................................................................... 38 Table 13 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Unconditional model ............................................................................................................................... 42 Table 13 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Unconditional model (Continued) .......................................................................................................... 43 9 Table 13 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Unconditional model (Continued) .......................................................................................................... 44 Table 14 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Conditional model ............................................................................................................................... 48 Table 14 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Conditional model (Continued) .......................................................................................................... 49 Table 14 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Conditional model (Continued) .......................................................................................................... 50 Table 14 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Conditional model (Continued) .......................................................................................................... 51 Table 15 – Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Unconditional model ............................................................................................................................... 54 Table 15 – Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Unconditional model (Continued) .......................................................................................................... 55 Table 15 – Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Unconditional model (Continued) .......................................................................................................... 56 Table 16 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Conditional model ............................................................................................................................... 60 Table 16 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Conditional model (Continued) .......................................................................................................... 61 Table 16 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Conditional model (Continued) .......................................................................................................... 62 Table 16 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Conditional model (Continued) .......................................................................................................... 63 Table 17 – Mean Environmental, Emission Reduction, Resource Reduction and Product Innovation scores .............................................................................................. 65 16 tending towards underperformance or neutrality, it is also worth noting that investors are paying a premium for their socially responsible decisions when investing in alternative energy funds (Reboredo et al., 2017, Martí-Ballester, 2019b). To the best of my knowledge, there are only two studies – those of Anderloni & Tanda (2017) and Ng & Zheng (2018) that deal with green energy stocks. Anderloni & Tanda (2017) focus on the performance of initial price offerings of European green and non-green energy stocks up to 36-months after the initial price offering and find that the financial performance of green energy companies does not differ from the nongreen energy ones. In turn, Ng and Zheng (2018) show that synthetically formed green energy portfolios perform similarly or even better compared to its matching non-green portfolio and the S&P 500 Energy benchmark when considering the whole period and the economic boom from 2000 to 2009. 3. Methodology Within the energy sector, we start by distinguishing green energy companies from the non-green energy companies. Then, we form a value-weighted portfolio of green energy companies and a value-weighted portfolio of non-green energy companies as well as a differences portfolio that reflects a strategy of a long position in the green portfolio and a short position in the non-green portfolio. In order to evaluate the financial performance of the portfolios, we use the Carhart (1997) four-factor model (Eq. 1) that accounts for the market, size, growth, and momentum factors. Previous findings in socially responsible and green funds suggest that the size and growth factors play a significant role in explaining performance in socially responsible and green funds (e.g., Ng & Zheng, 2018, Bauer et al., 2005, and Silva & Cortez, 2016). The four-factor model is represented as follows: 𝑟𝑝,𝑡 = 𝛼𝑝+ 𝛽𝑝𝑟𝑚,𝑡 + 𝛽𝑠𝑆𝑀𝐵𝑡+ 𝛽ℎ𝐻𝑀𝐿𝑡+ 𝛽𝑚𝑀𝑂𝑀𝑡+ 𝜀𝑝,𝑡 (Eq. 1) where 𝑟𝑝,𝑡 is the excess return of portfolio p in period t; 𝑟𝑚,𝑡is the market excess return in period t, 𝑆𝑀𝐵𝑡 is the difference in return between small-cap and large-cap portfolios 17 in period t, 𝐻𝑀𝐿𝑡 is the difference in return between high book-to-market stocks and low book-to-market stocks in period t, 𝑀𝑂𝑀𝑡 is the difference in returns of a portfolio of past winning stocks and a portfolio of past losing stocks (in the past 12 months) and 𝜀𝑝,𝑡 is the error term. Considering that using an unconditional model might lead to biased estimates of performance (Ferson & Schadt, 1996), we follow Christopherson et al. (1998) and apply the four-factor model in a conditional setting that allows for alphas and betas to vary linearly over time as a function of a vector of conditioning information 𝑍𝑡−1 which represents the public information available at time t − 1 that is relevant for predicting returns at time t, as follows: 𝑟𝑝,𝑡 = 𝛼0𝑝 + 𝐴′𝑝𝑧𝑡−1 + 𝛽0𝑝𝑟𝑚,𝑡 + 𝛽′0𝑝(𝑧𝑡−1𝑟𝑚,𝑡) + 𝛽1𝑝𝑆𝑀𝐵𝑡+ 𝛽′1𝑝(𝑧𝑡−1𝑆𝑀𝐵,𝑡) + 𝛽2𝑝𝐻𝑀𝐿𝑡+ 𝛽′2𝑝(𝑧𝑡−1𝐻𝑀𝐿𝑡) +𝛽3𝑝𝑀𝑂𝑀𝑡+ 𝛽′3𝑝(𝑧𝑡−1𝑀𝑂𝑀𝑡)+ 𝜀𝑝,𝑡 (Eq. 2) where 𝑧𝑡−1 = 𝑍𝑡−1 − 𝐸(𝑍) represents a vector of deviations of 𝑍𝑡−1 from their unconditional average values, 𝛽0𝑝, 𝛽1𝑝, 𝛽2𝑝, 𝛽3𝑝 are the average betas, 𝛽′0𝑝, 𝛽′1𝑝, 𝛽′2𝑝, 𝛽′3𝑝 are vectors that measure the sensitivity of conditional betas to the information variables 𝑍𝑡−1, 𝐴′𝑝 is a vector that measures the response of the conditional alpha to the information variables, 𝛼0𝑝 is the average conditional alpha, and 𝜀𝑝,𝑡 is the error term. Complementing the study, we identify the stocks that are rated by ASSET4 ESG 3 and form portfolios based on the environmental dimensions of the ASSET4 ESG Scores. The procedure to form portfolios is inspired by Kempf & Osthoff, (2007) and Statman & Glushkov, (2009): Each year we rank companies according to their scores in the Environmental Pillar of the ASSET4 ESG database as well as in its three categories: Emission Reduction, Resource Reduction and Product Innovation. The portfolios are based on the highest and lowest rated companies for each dimension, considering a 30% and 50% cut-off. The portfolios are rebalanced annually and scores from period t-1 are 3 Asset4ESG is a source of social data. It is described in more detail in the next section. 18 used to form the portfolios in the period t. The performance of these portfolios is also evaluated with the models presented in equations (1) and (2). 4. Data Green stocks are defined as firms that produce energy directly from environmentally friendly sources (e.g., solar, wind, biodiesel, etc.) and firms closely related to energy efficiency (e.g., electricity storage, smart grid, clean transportation). In order to identify the European green and non-green energy firms to be included in the dataset, we use the Eikon database together with the altenergystocks.com website 4 . In the Eikon database we used the code "INDUS" and then selected the Energy Sector to retrieve all the constituents from the "Energy - Fossil Fuel" and "Renewable Energy" tabs for European Markets. The altenergystocks.com website was used to complement the green portfolio (including only European firms), excluding trust funds and ETF's. After retrieving all the firms for both portfolios, we started with a pool of 387 firms for the non-green portfolio and 122 for the green portfolio. We then collected the monthly market capitalization and return index data, in USD, of these firms on Eikon DataStream over the 2009-2018 period. As for some firms the search for these variables returned errors, we ended up with 321 firms for the non-green portfolio and 115 firms for the green portfolio. Table (1) presents the descriptive statistics for the green and non-green portfolios including all firms, with the non-green portfolio showing a lower minimum and higher maximum, a higher standard deviation, a higher mean and a lower median when compared to the green portfolio. As mentioned in the methodology section, we also check whether the green and non-green firms considered in the initial dataset are rated by ASSET4. Asset 4ESG is a database that provides information on companies’ ratings in terms of several dimensions of corporate social responsibility, namely on the Environment, Social and Governance pillars. The overall rating of the Environmental pillar provided by ASSET ESG results from the aggregation of the scores in three of its categories: Emission Reduction, Resource Reduction and Product Innovation. We identify the companies in our initial dataset that are rated by ASSET4 ESG and form 4 The altenergystocks.com website was also used by Ng & Zheng, (2018) to identify green energy stocks. 19 portfolios of green and non-green rated firms (44 and 85 firms, respectively). Table (2) presents the descriptive statistics for these portfolios. We observe that the non-green portfolio exhibits wider swings in returns, and similar to when all firms are included, it presents a higher mean but a lower median. As for the normality test, using the JarqueBera test the p-value shows that, in any case, we cannot reject the hypothesis of the returns being normally distributed. Table 1 - Descriptive statistics of the green and non-green portfolios considering all firms. This table presents descriptive statistics of monthly returns of the green and non-green portfolios. Portfolios are value-weighted. The dataset includes 436 companies from 2009 to 2018. P-value is the probability of the Jarque-Bera normality test. Descriptive Statistics Non-Green Green Nº Observations 120 120 Minimum -15.01% -14.17% Maximum 21.28% 18.59% Mean 1.01% 0.93% Median 0.83% 1.25% Variance 0.40% 0.40% Stand. Deviation 6.32% 6.29% Skewness 0.320 -0.099 Kurtosis 0.326 -0.062 Jarque-Bera test p-value 0.243 0.905 Table 2 - Descriptive Statistics of the green and non-green portfolios considering the ESG rated firms. This table presents descriptive statistics of monthly returns of the green and non-green portfolios with ESG ratings from ASSET4. Portfolios are value-weighted. The dataset includes 129 companies from 2009 to 2018 (85 and 44 companies for the non-green and green portfolios, respectively). P-value is the probability of the Jarque-Bera normality test. Descriptive Statistics Non-Green Green Nº Observations 120 120 Minimum -15.26% -14.11% Maximum 21.14% 18.58% Mean 0.94% 0.90% Median 0.73% 1.29% Variance 0.41% 0.40% Stand. Deviation 6.40% 6.30% Skewness 0.312 -0.091 Kurtosis 0.294 -0.069 Jarque-Bera test p-value 0.271 0.917 20 As for information regarding the industry and location of the firms, Tables (3) and (4) show the number of stocks by country as well as industry that are considered for the green and non-green portfolios, respectively. The stocks in the green portfolio are mainly located in 6 countries - Germany, France, Sweden, United Kingdom, Italy and Poland, with the main industries for the portfolio being Renewable Energy Equipment, Alternative Fuels, Electrical Components and Alternative Energy. The non-green portfolio stocks are mostly located in the United Kingdom, Russia, Norway, Romania, France and Poland, with their main industries relating to Oil, Oil Equipment, Oil Refining, Coal, Offshore Drilling and Marine Transportation. Table 3 - Number of stocks by country and industry: non-green portfolio This table presents the number of stocks of the green portfolio, by country and industry, from 2009 to 2018. 21 Table 4 - Number of stocks by country and industry: green portfolio. This table presents the number of stocks of the non-green portfolio in the dataset, by country and industry, from 2009 to 2018. Regarding the portfolios of companies with ASSET4 ESG scores, we merged the initial pool of firms from both green and non-green portfolios and for each company we collected the scores for the Environmental Score as well as for its three categories: Emission Reduction, Resource Reduction and Product Innovation. For each dimension we created high and low-rated portfolios with 50% and 30% cut-offs. Again, due to DataStream not returning the respective scores from each dimension for some firms, the high and low-rated portfolios are taken from a pool of 129 firms. As benchmarks, we use a specialized and a general benchmark. The specialized benchmark was proxied by the MSCI Europe Energy Index. The return index (RI) data of this index was collected from DataStream. To compute the excess returns of the specialized index, we used the risk-free rate from Professor Kenneth French’s data library. 5 As the general market benchmark, we use the European market excess returns from Professor Kenneth French’s data library. The relevant risk factors associated with European markets that are necessary to the Carhart (1997) 4-factor model regression 5 http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html. 22 analysis were also collected from the Kenneth French website under the "Developed Markets Factors and Returns" section for European markets. As conditioning information, we chose two public information variables: the term spread and the dividend yield. These variables were also used by Leite & Cortez (2018) and Leite et al. (2018).The term spread is computed as the difference between the yield of a long term (proxied by the 10 Year EMU Benchmark) and a short-term bond (represented by the 3 Month Euribor). The dividend yield is based on the STOXX Europe 600 index. This data was collected from DataStream. To mitigate the possibility of spurious regressions, we followed the suggestion of Ferson et al. (2003) and stochastically detrended these series by subtracting the 12-month moving average. Also, these variables were used in the mean-zero form. 5. Empirical Results 5.1. Performance of green vs non-green portfolios Table (5) presents the regression estimates of the unconditional multifactor model using all the firms initially included in the dataset. We perform the regressions using the MSCI Energy EU and the Kenneth French (KF) market factors as benchmarks. Besides evaluating performance for the overall period 2009-2018, we also evaluate portfolio performance for two subperiods of five years (2009-2013 and 2014-2018). In order to identify any differences between the green and non-green portfolios we also present the estimates of the difference between the green and non-green portfolios, consisting of a strategy of a long position in the green portfolio and a short position in the nongreen portfolio. This approach is followed for all models implemented. Also, in some tables that consider the conditional model, some columns are omitted since those do not have any statistical significance. 23 Table 5 - Performance of green and non-green portfolios including all firms – unconditional model. This table presents estimates of monthly abnormal returns, factor loadings, and the adjusted 𝑅2 obtained from regressing equation (1). Mkt corresponds to the excess returns of the benchmark, proxied by market returns in Professor Kenneth French’s website (Kenneth French) or the returns of the MSCI EU Energy Index in excess of the risk-free rate (MSCI EU Energy); SMB is the difference in return between small-cap and large-cap portfolios; HML is the difference in return between of high book-to-market and low bookto-market portfolios; and MOM is the difference in return between portfolios of stocks considered to have strong momentum and portfolios of stocks considered to have weak momentum. The portfolios are valueweighted and rebalanced annually. The observation period is from 2009 to 2018. ***, **, and * indicate significance at the 1%, 5% and 10% level, respectively. Standard errors are corrected for heteroskedasticity and autocorrelation using the Newey-West (1987) method. As shown in the table, both portfolios show a neutral performance when using the general market index as benchmark. When using the MSCI Energy EU, both portfolios yield positive and significant abnormal returns. As for the subperiods analysis, performance tend to be neutral. The exception is the non-green portfolio, which shows positive and statistically significant returns when using the MSCI Energy EU benchmark. In either case, the difference between the performance of both portfolios is not 24 statistically significant, meaning that there are no abnormal gains resulting from a strategy of going long in green portfolios and short in non-green portfolios. Taking a closer look to the risk factors estimates, when considering the whole period and using all KF factors, we observe that the non-green portfolio is not significantly exposed to any risk factor besides the market and the green portfolio is exposed to value stocks, although the difference between both portfolios is not statistically significant. But, when we use the MSCI Energy EU as the benchmark, the market risk factor on the green portfolio decreases from around 1 to around 0.6 and the exposure to value stocks also increases. As for the non-green portfolio, the market factor remains similar, but it shows exposure to a small firm effect. Comparing portfolios, all these differences are statistically significant, with the SMB factor significant at 5% level and the HML and market risk factors significant at the 1% level. Considering now the period division with the KF market factor, we observe some interesting risk factor estimates. In the 2009 to 2013 period, the green portfolio is significantly exposed, at the 1% level, to value stocks. The results of the long-short portfolio show that the difference of the exposure to the SMB and HML factors is statistically significant at the 10% and 1% level, respectively, indicating that the green portfolios is more exposed to large and value stocks than the non-green portfolio. In the 2014 to 2018 period, these results change: the non-green portfolio is now significantly exposed to value stocks (at the 1% level), whereas the green portfolio is now negatively exposed to the size factor (at the 10% level) and positively exposed to the momentum factor (at the 5% level), although the momentum and size coefficients are not significantly different from those of the non-green portfolio. The subperiod analysis with the MSCI Energy EU benchmark shows that in the first period, similarly to when using the KF market factor, the green portfolio is exposed to value stocks (at the 1% level). The green portfolios also exhibit a statistically significant lower market risk, at 1% level, compared to the non-green portfolio. The non-green portfolio shows positive and significant coefficient, at 10% level, for the size factor as well as a significant negative exposure, at 1% level, to momentum factor, although the latter does not result in a significant difference between portfolios’ exposure to momentum. In the 2014 to 2018 period, the R2 for the green portfolio declines and the risk exposures to the additional risk factors are now statistically insignificant. The non- 25 green portfolio shows some exposure to small cap and value stocks (although only at the 10% level). Yet, except for the market factor, none of these coefficients are statistically different between both portfolios. In the case of the former, the difference might be due to the benchmark used not capturing all relevant market effects in the second period. As for the alphas, although the non-green portfolio shows a positive and statistically significant coefficient at the 5% and 10% levels, respectively, in the first and second period, there are no statistical difference between the performance of the two portfolios. Table (6) presents the regression estimates for the conditional multifactor models with time-varying alphas and betas, and with the term spread and dividend yield used as public information variables. We can observe that in some cases the variables associated with the public information variables are statistically significant, supporting the use of the conditional models in performance evaluation. We performed Wald tests for the conditional alphas and betas and these are in every case statistically significant, 6 meaning that we reject the hypothesis of the conditional alphas and conditional betas being jointly equal to zero. 6 The results of the Wald test are not reported for the sake of space. 32 We further explore the performance of green and non-green portfolios by comparing the performance of the green and non-green portfolios containing all the firms versus green and non-green portfolios of companies that are rated by ASSET4 ESG. Tables 9 to 12 present the results. We can see that the risk exposures are almost identical in every case. The same can be said for the abnormal performance estimates, that are nearly identical in every case. The major difference in performance is observed when comparing the performance of all non-green firms to non-green firms that are rated (Tables 11 and 12). In this case, we observe some outperformance of the former compared to the latter. In table 12, the non-green portfolio of all firms, when considering the first and full periods and for both benchmarks, shows a statistically significant outperformance at 1% level compared to the non-green portfolio with ESG rated firms only. These results indicate that the portfolio of non-green firms with Environmental ratings underperform the non-green portfolio that includes all firms. When using the conditional model, the results of the first and full periods with the MSCI Europe Energy benchmark still show a statistically significant difference in performance, at the 5% level, but those obtained with the KF benchmark show a statistically significant difference, at the 1% level, only in the 2009 to 2013 period. 33 Table 9 - Performance of a green portfolio containing all the initial firms and a green portfolio containing only the ESG-rated firms – unconditional model This table presents estimates of monthly abnormal returns, factor loadings, and the adjusted R^2 obtained from regressing equation (1). Mkt corresponds to the excess returns of the benchmark, proxied by market returns in Professor Kenneth French’s website (Kenneth French) or the returns of the MSCI EU Energy Index in excess of the risk-free rate (MSCI EU Energy); SMB is the difference in return between small-cap and large-cap portfolios; HML is the difference in return between of high book-to-market and low book-to-market portfolios; and MOM is the difference in return between portfolios of stocks considered to have strong momentum and portfolios of stocks considered to have weak momentum. The portfolios are value-weighted and rebalanced annually. The observation period is from 2009 to 2018. ***, **, and * indicate significance at the 1%, 5% and 10% level, respectively. Standard errors are corrected for heteroskedasticity and autocorrelation using the Newey-West (1987) method. 34 Table 10 - Performance of a green portfolio containing all the initial firms and a green portfolio containing only the ESG-rated firms - conditional model This table presents estimates of monthly abnormal returns, factor loadings, and the adjusted R^2 obtained from regressing equation (2). Mkt corresponds to the excess returns of the benchmark, proxied by market returns in Professor Kenneth French’s website (Kenneth French) or the returns of the MSCI EU Energy Index in excess of the risk-free rate (MSCI EU Energy); SMB is the difference in return between small-cap and large-cap portfolios; HML is the difference in return between of high book-to-market and low book-to-market portfolios; and MOM is the difference in return between portfolios of stocks considered to have strong momentum and portfolios of stocks considered to have weak momentum. TS corresponds to the term spread, computed as the difference between the yield of a long-term and short-term bond (proxied by the 10 year EMU benchmark and 3-month Euribor, respectively); DY corresponds to the dividend yield and is based on the STOXX Europe 600 Index. The portfolios are valueweighted and rebalanced annually. . Panel A presents the results for the 2009 to 2018 period. Panel B present the results for the subperiods using the KF and MSCI EU Energy Index, respectively. ***, **, and * indicate significance at the 1%, 5% and 10% level, respectively. Standard errors are corrected for heteroskedasticity and autocorrelation using the Newey-West (1987) method. 35 Table 10 - Performance of a green portfolio containing all the initial firms and a green portfolio containing only the ESG-rated firms - conditional model (Continued) 36 Table 11 - Performance analysis of a non-green portfolio containing all the initial firms and a non-green portfolio containing only the ESG-rated firms – Unconditional model This table presents estimates of monthly abnormal returns, factor loadings, and the adjusted R^2 obtained from regressing equation (1). Mkt corresponds to the excess returns of the benchmark, proxied by market returns in Professor Kenneth French’s website (Kenneth French) or the returns of the MSCI EU Energy Index in excess of the risk-free rate (MSCI EU Energy); SMB is the difference in return between small-cap and large-cap portfolios; HML is the difference in return between of high book-to-market and low book-to-market portfolios; and MOM is the difference in return between portfolios of stocks considered to have strong momentum and portfolios of stocks considered to have weak momentum. The portfolios are value-weighted and rebalanced annually. The observation period is from 2009 to 2018. ***, **, and * indicate significance at the 1%, 5% and 10% level, respectively. Standard errors are corrected for heteroskedasticity and autocorrelation using the Newey-West (1987) method. 37 Table 12 - Performance analysis of a non-green portfolio containing all the initial firms and a non-green portfolio containing only the ESGrated firms – Conditional model This table presents estimates of monthly abnormal returns, factor loadings, and the adjusted R^2 obtained from regressing equation (2). Mkt corresponds to the excess returns of the benchmark, proxied by market returns in Professor Kenneth French’s website (Kenneth French) or the returns of the MSCI EU Energy Index in excess of the risk-free rate (MSCI EU Energy); SMB is the difference in return between small-cap and large-cap portfolios; HML is the difference in return between of high book-to-market and low book-to-market portfolios; and MOM is the difference in return between portfolios of stocks considered to have strong momentum and portfolios of stocks considered to have weak momentum. TS corresponds to the term spread, computed as the difference between the yield of a long-term and short-term bond (proxied by the 10 year EMU benchmark and 3-month Euribor, respectively); DY corresponds to the dividend yield and is based on the STOXX Europe 600 Index. The portfolios are valueweighted and rebalanced annually. Panel A presents the results for the 2009 to 2018 period. Panel B presents the results for the subperiods using the KF and MSCI EU Energy Index, respectively. ***, **, and * indicate significance at the 1%, 5% and 10% level, respectively. Standard errors are corrected for heteroskedasticity and autocorrelation using the Newey-West (1987) method. 38 Table 12 - Performance analysis of a non-green portfolio containing all the initial firms and a non-green portfolio containing only the ESGrated firms – Conditional model (Continued) 39 5.3. Performance of environmentally ranked portfolios Table (13) presents the regression estimates for the unconditional 4-factor model of high and low-rated portfolios formed based on the Environmental Pillar and its individual dimensions (Emission Reduction, Resource Reduction and Product Innovation), considering a 30% cut-off. In all cases we also present the estimates of the difference between the high and low-rated portfolios to represent the performance of a strategy of a long position in high-rated portfolios and a short position in low-rated portfolios. Considering the full period (Panel A) and starting with the KF benchmark, we can see that whatever dimension considered, high and low-rated portfolios do not achieve statistically significant abnormal returns. Furthermore, no significant performance difference is observed between high and low-rated portfolios. Regarding the risk factors, the high-rated portfolios show negative coefficients associated to the size factor, although only significant at the 10% level in the portfolios formed on the Emission Reduction and Product Innovation dimensions. The coefficients of the HML factor are positive and statistically significant at the 5% level in the case of portfolios formed on the Environmental and Emission Reduction scores. The low-rated portfolio formed on the Emission Reduction dimension is exposed to the small size effect (at the 1% level). The results of the long-short portfolio show that high-rated portfolios formed on the individual dimensions of the Environmental pillar are less exposed to small firms than their low-rated peers. Also, high-rated portfolios formed on the Resource Reduction and Product Innovation dimensions are more exposed to the momentum effect than lowrated companies. As for the performance estimates obtained with the specialized benchmark for the full period, we observe statistically significant abnormal returns of all high-rated portfolios and on the low-rated portfolio formed on the Resource Reduction and Product Innovation dimensions (although in this case the level of significance is only 10%). However, there are no statistically significant differences in the performance of high-rated and low-rated portfolios. In terms of systematic risk, we observe that highrated portfolios have lower betas than low-rated portfolios, except for those based on 40 the Resource Reduction and Product Innovation dimensions. The exposure to the size factor shows a similar story to what we found previously with the general market benchmark: low-rated portfolios are more exposed to small size firms that high-rated firms. Furthermore, we find that low-rated portfolios are not exposed to the value factor, whereas high-rated portfolios are exposed to the value factor. Also, all low-rated portfolios have negative and statistically significant coefficients on the momentum factor, with low-rated portfolios formed on the Resource Reduction and Product Innovation dimensions being more exposed to firms with recent poor performance than their high-rated counterparts. Moving on to the subperiod analysis, when the market is proxied by the KF benchmark (Panel B), we observe that the alpha coefficients are neutral. In the second period, the results change slightly, as the high-rated portfolios formed on the Environmental pillar and Resource Reduction dimensions now exhibit positive and statistically significant alphas at the 10% level. Yet, none of the differences between the alphas of the portfolios are statistically significant, meaning that there are no abnormal gains from following a strategy of going long in high-rated portfolios and short in lowrated portfolios. As for the risk factors, from 2009-2013 the high-rated portfolios are not exposed to any risk source other than the market, whereas the low-rated portfolios are exposed to small firms, growth stocks and, for the Resource Reduction and Product Innovation dimensions, firms with recent poor performance. On the second period, the low-rated portfolios are no longer exposed to the small size effect, but they are now exposed to value stocks. Unlike the first period, and with exception of the Environmental dimension, high-rated portfolios show significant exposure to large firms and value stocks. In panel C, as with the MSCI Energy EU, the market risk exposure for the high-rated portfolios is lower and all the high-rated portfolios show exposure to value stocks. The low-rated portfolios show exposure to small cap firms, as all coefficients, except for the Environmental dimension, are statistically significant. Low-rated firms also show exposure to growth stocks, although only portfolios formed on the Emission Reduction and Resource Reduction dimensions have statistically significant coefficients on the HML factor. Also, with this benchmark the low-rated portfolios tend to show a negative exposure to the momentum factor, except for the Emission Reduction dimension. In the 41 2014 to 2018 period, we observe that neither high or low-rated portfolios have any significant exposure to the value factor and only the low-rated portfolios show exposure towards small cap stocks (whose coefficients are always statistically significant at least at the 5% level). Furthermore, we observe that the market risk exposure for the highrated portfolios is lower compared to low-rated portfolios, with the Environmental Pillar and Emission Reduction dimensions differing significantly, at the 1% level, from the lowrated portfolios. As for the abnormal returns estimates, in the 2009 to 2013 period only one portfolio (the high-rated portfolio on the Product Innovation dimension) shows a statistically significant alpha, although only at the 5% level. However, we can conclude that in this period high-rated portfolios did not outperform low-rated ones. In the 2014 to 2018 period, the only positive alphas are those of the highand low-rated portfolios formed on the Resource Reduction dimension, although only at the 10% level. In this period, we also observe that the high-rated portfolio formed on the Environmental Pillar dimension significantly outperformed the low-rated portfolio. 48 Table 14 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Conditional model This table presents estimates of monthly abnormal returns, factor loadings, and the adjusted R^2 obtained from regressing equation (2). Mkt corresponds to the excess returns of the benchmark, proxied by market returns in Professor Kenneth French’s website (Kenneth French) or the returns of the MSCI EU Energy Index in excess of the riskfree rate (MSCI EU Energy); SMB is the difference in return between small-cap and large-cap portfolios; HML is the difference in return between of high book-to-market and low book-to-market portfolios; and MOM is the difference in return between portfolios of stocks considered to have strong momentum and portfolios of stocks considered to have weak momentum. TS corresponds to the term spread DY corresponds to the dividend yield. The portfolios are value-weighted and rebalanced annually. Panels A and B present the results for the 2009 to 2018 period, using the KF and MSCI EU Energy Index, respectively. Panels C and D present the results for the subperiods using the KF and MSCI EU Energy Index, respectively. ***, **, and * indicate significance at the 1%, 5% and 10% level, respectively. Standard errors are corrected for heteroskedasticity and autocorrelation using the Newey & West, (1987) method. 49 Table 14 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Conditional model (Continued) 50 Table 14 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Conditional model (Continued) 51 Table 14 - Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 30% cut-off – Conditional model (Continued) 52 Table (15) presents the regression estimates for the unconditional four-factor model of high and low-rated portfolios formed based on the individual Environmental Pillar and its three categories, considering a 50% cut-off. In the full period (Panel A), with the KF benchmark, the 50% cut-off estimates show an increase in the exposure of the low-rated portfolios to small caps. The size coefficients are now statistically significant in all dimensions, except for the Environmental Pillar, while the value and momentum estimates remain mostly similar. Regarding the abnormal returns’ coefficients, the highrated portfolio formed on the Resource Reduction dimension has a positive and statistically significant alpha, at the 10% level, but the difference compared to the lowrated alpha is not statistically significant. The results further show that the low-rated portfolio formed on the Environmental dimension underperforms the high-rated one, although only at the 10% level. These results may be due to the low-rated portfolio increasing the number of firms that have higher Environmental scores. The results with the MSCI Energy EU benchmark show that high-rated portfolios exhibit more exposure to small size caps than low-rated portfolios. The value and momentum factors are still very similar to those of the 30% cut-off portfolios. The alpha estimates also show changes compared to the 30% cut-off portfolios. In this case, we observe that the highrated portfolio formed on the Resource Reduction dimension performs significantly better, at the 10% level, than its low-rated counterpart. In the 2009 to 2013 period (Panel B), the results using the KF benchmark show very similar estimates regarding the risk factors and alpha estimates, with most exposures and coefficients with the same significance as the estimates from the 30% cut-off. The high-rated and the low-rated portfolios show neutral performance. Yet, the high-rated portfolio formed on the Environmental pillar underperforms the low-rated portfolio at the 5% level. In the 2014 to 2018 period, using the KF benchmark, similarly to in the results of the 30% cut-off, we observe similar changes in risk exposures compared to the 2009 to 2013 period, with the high-rated portfolios exposed to large cap firms and value stocks and the low-rated portfolios tending towards small caps. Regarding performance, we observe improvements in the alpha estimates, with the high-rated portfolios presenting positive and significant estimates in the Emission Reduction and in Resource Reduction dimensions. And in the latter case, high-rated firms even outperform their low-rated peers at the 5% level. 53 The results of the first period with the MSCI Energy EU benchmark (Panel C) show mostly similar exposures as in the 30% cut-off portfolios, with low-rated portfolios’ exposure to growth stocks increasing for portfolios formed on the Product Innovation dimension and decreasing for the remaining ones. With this benchmark, the alpha estimates for the high-rated portfolios are positive but lower than the low-rated portfolios, resulting, in the Environmental Pillar dimension, in a statistically significant difference, at the 5% level. This means that the low-rated portfolio significantly outperforms the high-rated one in this dimension. As for the second period, the market risk exposures are slightly higher for the Environmental Pillar and Emission Reduction dimensions and lower for the other two dimensions, when compared to the results of the 30% cut-off portfolios. When compared to the first period, the trend in risk exposures is similar to the previous results, with low-rated portfolios tending towards small cap and value stocks and high-rated portfolios mostly exposed only to market risk. Regarding alphas, we observe slight improvements in the high-rated portfolios’ estimates when compared to the 2009 to 2013 period. In contrast, the low-rated portfolios exhibit lower alpha estimates, with the low-rated portfolio in the Resource Reduction dimension being negative and statistically significant, at the 10% level. As a result, we observe that in this dimension high-rated portfolios outperform, at 1% level, low-rated ones. 54 Table 15 – Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Unconditional model This table presents estimates of monthly abnormal returns, factor loadings, and the adjusted R^2 obtained from regressing equation (1). Mkt corresponds to the excess returns of the benchmark, proxied by market returns in Professor Kenneth French’s website (Kenneth French) or the returns of the MSCI EU Energy Index in excess of the risk-free rate (MSCI EU Energy); SMB is the difference in return between small-cap and large-cap portfolios; HML is the difference in return between of high book-to-market and low bookto-market portfolios; and MOM is the difference in return between portfolios of stocks considered to have strong momentum and portfolios of stocks considered to have weak momentum. SMB, HML and MOM are proxied using the data available in Professor Kenneth French’s website. The portfolios are valueweighted and rebalanced annually. Panel A presents the results for the2009 to 2018 period. Panels B and C present the results for the subperiods using the KF and MSCI EU Energy Index, respectively. ***, **, and * indicate significance at the 1%, 5% and 10% level, respectively. Standard errors are corrected for heteroskedasticity and autocorrelation using the Newey-West (1987) method. 55 Table 15 – Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Unconditional model (Continued) 56 Table (16) presents the regression estimates for the conditional four-factor model of high and low-rated portfolios formed based on the individual Environmental pillar dimensions, considering a 50% cut-off. For the full period (Panels A and B), and starting with the KF benchmark (Panel A), when compared to the 30% cut-off portfolios, the additional firms in the portfolios do not affect the results significantly, although we observe some slight changes in some of the risk exposures. The value factor associated Table 15 – Performance of high and low-rated portfolios performance analysis using the ASSET4 ESG Environmental Pillar dimensions with a 50% cut-off – Unconditional model (Continued) 57 with the term spread is now statistically significant for all portfolios and for all dimensions, and although the value factor associated with the dividend yield estimate is negative, the overall effects from the PIVs in this factor are positive. The alphas also change slightly, with the alphas associated to PIVs mostly cancelling each other out, resulting in no significant difference in these alphas. Comparing high and low-rated portfolios, the latter exhibits higher abnormal returns in the Environmental dimension, at 10% level, but in the Resource Reduction dimension the high-rated portfolio beats the low-rated one. The results obtained with the MSCI Energy EU benchmark (Panel B) show that the additional PIVs do not, for the most part, affect the performance or risk exposures, but the additional firms now increase the overall market risk in the Environmental Pillar and Emission Reduction dimensions and decrease it in the other two dimensions. Also, when compared to the 30% cut-off portfolios, the PIVs associated with the market risk lose their significance. In contrast to the results obtained with the KF benchmark, the alphas associated to the PIVs do not show any statistical significance, nor do they differ between the portfolios. However, the average abnormal performance estimates do show better estimates for the high-rated portfolios, even outperforming significantly, at the 5% level, the low-rated portfolio formed on the Resource Reduction dimension. Looking at the first period and using the KF benchmark (Panel C), the high-rated portfolios lose their previous statistical significance towards value stocks. Also, the lowrated portfolios seem to be, for the most part, the only ones that are significantly affected by the PIVs. Both PIVs associated with the value factor seem, overall, to affect this factor negatively, with the exception of the Environmental Pillar, showing a statistically significant difference for both coefficients. Regarding performance, the alphas associated to the term spread are only statistically significant at the 10%. Anyhow, the highand lowrated portfolios do not differ significantly from each other regarding the effect of this PIV on performance. The abnormal performance estimates further shows no statistically significant difference between high and low-rated portfolios. In the second period, regarding the PIVs, the size factor associated with the dividend yield differs significantly in the Environmental and Resource Reduction portfolios, at the 5% and 1% levels respectively. The PIV term spread affects the size factor significantly in all dimensions, except for the Environmental Pillar, although when 64 5.4. Environmental performance of green and non-green energy portfolios In this section, we evaluate the environmental performance of green and non-green portfolios and link environmental performance to financial performance. Starting with the green and non-green portfolios of companies that are rated by ASSET4 ESG. Figure (1) plots the evolution over time of the average Environmental ESG Pillar score, as well as the scores of each of its individual category. We can see that in all dimensions, the green portfolio’s annual average score is higher than that of the nongreen portfolio, with the difference being statistically significant at 1% level, as we can observe in Table (17). It is worth noting that in the Emission Reduction dimension, the non-green portfolio’s average score is not as low as in the other dimensions. Figure 1 - Evolution of Environmental ESG Pillar dimensions scores for the green and non-green portfolios This figure shows the mean Environmental, Resource Reduction, Emission Reduction and Product Innovation scores of portfolios of green and non-green firms from 2008 to 2018. 65 Table 17 – Mean Environmental, Emission Reduction, Resource Reduction and Product Innovation scores This table shows the Environmental, Emission Reduction, Resource Reduction and Product Innovation scores of the green, non-green, and difference portfolio. The significance of differences between means is based on a two-sample T-test assuming unequal variance. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels. The sample period is from 2009 to 2018. To further investigate how the scores progress along the years, we split the green and non-green portfolios of rated stocks in 50% and 30% cut-offs, according to their ASSET4 ESG ratings, and plot the evolution of high-ranked and low-ranked stocks of each portfolio in Figures (2) and (3). We can observe that the average scores of the highranked portfolio of non-green stocks almost match but are always lower than those of the high-ranked portfolio of green stocks, especially when considering the 30% cut-off. Also, while average scores of the low-rated portfolio of non-green stocks do not change much throughout the years, the scores of the low-rated portfolio of green stocks show an improvement over time, almost matching the scores of the high rated portfolio of non-green stocks in the Environmental and Product Innovation dimensions when considering the 30% cut-off, and even surpassing the high-ranked portfolio of non-green stocks in the Product Innovation dimension when considering the 50% cut-off. While there is a significant difference between the evolution of the green ESG ratings and the non-green ESG ratings, there isn’t any accompanying changes in performance of the green and non-green ESG rated portfolios. 66 Figure 2 – Evolution of Environmental ESG Pillar dimensions scores for the green and non-green portfolios considering a 50% cut-off This figure shows the mean Environmental, Resource Reduction, Emission Reduction and Product Innovation scores of green and non-green firms from 2009 to 2018. 67 Figure 3 – Evolution of Environmental ESG Pillar dimensions scores for the green and non-green portfolios considering a 30% cut-off This figure shows the mean Environmental, Resource Reduction, Emission Reduction and Product Innovation scores of green and non-green firms from 2009 to 2018. As for the ESG based portfolios, we can see in Figure (4) that the high-rated portfolios’ scores are usually in the mid to low 90's for all dimensions and cut-offs, while the low-rated portfolios scores are in the high 50's to low 40's (high 30's to low 30's), except for the Product Innovation dimension, where it stays around 30 (around 20) in the 50% cut-off (30% cut-off). The improvements of the 50% and 30% lower cut-off portfolios’ ESG ratings from 2014 to 2018 may be tied to the significantly higher abnormal returns of the high-rated portfolio when compared to the low-rated counterpart in the second subperiod in terms of the Environmental and Resource Reduction dimensions, as observed in tables 13 to 16. 68 Figure 4 – Evolution of Environmental ESG Pillar dimensions scores for the ESG rated portfolios considering 50% and 30% cut-offs. This figure shows the mean Environmental, Resource Reduction, Emission Reduction and Product Innovation scores of ESG based portfolios from 2009 to 2018. 6. Conclusions This paper addresses the performance of portfolios of European green and nongreen energy stocks from 2009 to 2018. We form green and non-green value-weighted portfolios based on all constituents from the "Energy - Fossil Fuel" and "Renewable Energy" tabs in the Eikon database as well as the altenergystocks.com website. We also form portfolios based on the categories of the Environmental ASSET4 ESG Pillar, namely the Environmental pillar and the Emission Reduction, Resource Reduction and Product Innovation categories, and compared the performance of high-rated portfolios against low-rated portfolios considering 30% and 50% cut-offs. Long-short portfolios (long in green or high-rated and short in non-green or low-rated portfolios were also formed in 69 order to compare the differences in performance from investing in these portfolios. Portfolio performance is evaluated using the four-factor model (Carhart, 1997) in an unconditional and conditional setting. Our results show that, for the most part, green energy portfolios are very similar to non-green portfolios in terms of performance, with our results not showing any statistical difference between green and non-green portfolios. As for the ESG based portfolios, their performance is also very similar between high and low-rated portfolios, but when considering the full and 2014 to 2018 periods, using the MSCI Energy EU benchmark, we observe some high-rated portfolios significantly outperforming the lowrated portfolios. Also, there are instances where we observe low-rated portfolios significantly outperforming high-rated portfolios formed on the Environmental Pillar dimension. It is also important to note that when considering the subperiods, in almost any case, we document a slight improvement in the abnormal performance estimates for green and high-rated portfolios and, in cases where the estimates deteriorate, the nongreen and low-rated portfolios estimates usually deteriorate more, even though most of the differences between green (high-rated) and non-green (low-rated) portfolios are not significant. The green energy portfolios show, overall, a similar performance when compared to the non-green energy portfolios. Also, both green and non-green portfolios’ alphas are similar to those obtained with the KF benchmark but are usually higher than those of the specialized benchmark. And when considering the ESG based portfolios, the high and low-rated portfolios show similar alphas when compared to the KF benchmark, while the high-rated portfolios show overall higher performance than the MSCI Energy EU benchmark. Overall, our results suggest that forming portfolios based on green energy screening does not hurt portfolio performance compared to investing in non-green energy firms. Our results seem to be consistent with the studies of Anderloni & Tanda (2017) and Ng & Zheng (2018), suggesting similar performance between green and non-green energy stocks. Our results are of interest for investors, businesses and regulators. To investors, our results suggest that they can invest in environmentally friendly energy companies without paying a green premium. Hence, investors can play an active role in transitioning 70 from carbon-intensive fossil fuels to renewable and clean energy sources without sacrificing financial performance. These results also encourage policymakers/regulators concerned with the reliance on fossil fuel energy to promote industries (such as renewable energy) that may help to achieve the goals of reducing greenhouse gas emissions and contribute to converge to a sustainable economy. It is also worth noting that despite the substantial decline in global energy demand following the global coronavirus pandemic that unfolded in December 2019, leading many countries to implement lockdown and confinement measures, renewable energy has so far been the source of energy most resilient to the Covid-19 crisis (IEA, 2020). The results in this research may be affected by the shocks in oil prices and recent changes in investment behavior favoring non-polluting green firms and divesting from non-green energy firms. Future research may be able to consider these factors. Furthermore, an opportunity for future research is to evaluate the performance of ASSET4 ESG rated firms and comparing them to non-rated ones. 71 References Ambec, S., & Lanoie, P. (2008). Does it pay to be green? A systematic overview. Academy of Management Perspectives, 22(4), 45–62. http://www.scopus.com/scopus/openurl/link.url?ctx_ver=Z39.882004&ctx_enc=info:ofi/enc:UTF8&svc_val_fmt=info:ofi/fmt:kev:mtx:sch_svc&svc.citedby=yes&rft_id=info:eid/2s2.058249109340&rfr_id=http://search.ebscohost.com&rfr_dat=partnerID:NnvIuKwx &rfr_dat= Anderloni, L., & Tanda, A. (2017). Green energy companies: Stock performance and IPO returns. Research in International Business and Finance, 39(Part A), 546–552. https://doi.org/10.1016/j.ribaf.2016.09.016 Bauer, R., Koedijk, K., & Otten, R. (2005). International evidence on ethical mutual fund performance and investment style. Journal of Banking and Finance, 29(7), 1751– 1767. https://doi.org/10.1016/j.jbankfin.2004.06.035 Bloomberg. (2019). Clean Energy Investment Exceeded $300 Billion Once Again in 2018. Bloomberg New Energy Finance. https://about.bnef.com/blog/clean-energyinvestment-exceeded-300-billion-2018/ Brzeszczyński, J., Ghimire, B., Jamasb, T., & McIntosh, G. (2019). Socially responsible investment and market performance: The case of energy and resource companies. Energy Journal, 40(5), 17–72. https://doi.org/10.5547/01956574.40.5.jbrz Capelle-Blancard, G., & Monjon, S. (2012). Trends in the literature on socially responsible investment: Looking for the keys under the lamppost. Business Ethics, 21(3), 239–250. https://doi.org/10.1111/j.1467-8608.2012.01658.x Carhart, M. M. (1997). On Persistence in Mutual Fund Performance. Journal of Finance (Wiley-Blackwell), 52(1), 57–82. https://doi.org/10.1111/j.15406261.1997.tb03808.x Christopherson, J. A., Ferson, W. E., & Glassman, D. A. (1998). Conditioning Manager Alphas on Economic Information: Another Look at the Persistence of Performance. Review of Financial Studies, 11(1), 111–142. http://10.0.4.69/rfs/11.1.111 Climent, F., & Soriano, P. (2011). Green and Good? The Investment Performance of US Environmental Mutual Funds. Journal of Business Ethics, 103(2), 275–287. https://doi.org/10.1007/s10551-011-0865-2 Cortez, M. C., Silva, F., & Areal, N. (2009). The Performance of European Socially Responsible Funds. Journal of Business Ethics, 87(4), 573–588. https://doi.org/10.1007/s10551-008-9959-xf 72 Freeman, R., & Evan, W. M. (1990). Corporate governance: A stakeholder interpretation. The Journal of Behavioral Economics, 19(4), 337–359. https://doi.org/10.1016/0090-5720(90)90022-Y Ferson, W. E., Sarkissian, S., & Simin, T. T. (2003). Spurious Regressions in Financial Economics? The Journal of Finance, 58(4), 1393–1413. https://doi.org/10.1111/1540-6261.00571 Ferson, W. E., & Schadt, R. W. (1996). Measuring Fund Strategy and Performance in Changing Economic Conditions. Journal of Finance, 51(2), 425–461. https://econpapers.repec.org/RePEc:bla:jfinan:v:51:y:1996:i:2:p:425-61 French, K. R. (2019). Kenneth R. French - Data Library. https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html Grossman, B. R., & Sharpe, W. F. (1986). Financial Implications of South African Divestment. Financial Analysts Journal, 42(4), 15–29. https://doi.org/10.2469/faj.v42.n4.15 Henriques, I., & Sadorsky, P. (2018). Investor implications of divesting from fossil fuels. Global Finance Journal, 38, 30–44. https://doi.org/10.1016/j.gfj.2017.10.004 Hunt, C., & Weber, O. (2019). Fossil Fuel Divestment Strategies: Financial and CarbonRelated Consequences. Organization and Environment, 32(1), 41–61. https://doi.org/10.1177/1086026618773985 Ibikunle, G., & Steffen, T. (2017). 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 International Energy Agency -IEA-. (2019). Global Energy and CO2 Status Report 2018. Retrieved September, 2019 from https://www.iea.org/geco/ International Energy Agency -IEA-. Global Energy Review 2020: The impacts of the Covid-19 crisis on global energy demand and CO2 emissions. Retrieved June, 2020 from https://doi.org/https://www.iea.org/reports/global-energy-review-2020 Javed, M., Rashid, M. A., & Hussain, G. (2016). When does it pay to be good – A contingency perspective on corporate social and financial performance: would it work? Journal of Cleaner Production, 133, 1062–1073. https://doi.org/10.1016/j.jclepro.2016.05.163 Kempf, A., & Osthoff, P. (2007). The Effect of Socially Responsible Investing on Portfolio Performance. European Financial Management, 13(5), 908–922. https://doi.org/10.1111/j.1468-036X.2007.00402.x Leite, C., Cortez, M. C., Silva, F., & Adcock, C. (2018). The performance of socially responsible equity mutual funds: Evidence from Sweden. Business Ethics, 27(2), 108–126. https://doi.org/10.1111/beer.12174 Leite, P., & Cortez, M. C. (2018). The performance of European SRI funds investing in bonds and their comparison to conventional funds. Investment Analysts Journal, 47(1), 65–79. https://doi.org/10.1080/10293523.2017.1414911 73 Lesser, K., Rößle, F., & Walkshäusl, C. (2016). Socially responsible, green, and faithbased investment strategies: Screening activity matters! Finance Research Letters, 16, 171–178. https://doi.org/10.1016/j.frl.2015.11.001 Mallett, J. E., & Michelson, S. (2010). Green Investing: Is it Different from Socially Responsible Investing? International Journal of Business, 15(4), 395–410. http://search.ebscohost.com/login.aspx?direct=true&db=edo&AN=55785947&sit e=eds-live Margolis, J. D., & Walsh, J. P. (2003). Misery Loves Companies: Rethinking Social Initiatives by Business. Administrative Science Quarterly, 48(2), 268-305. https://doi.org/10.2307/3556659 Martí-Ballester, C.-P. (2019). Do European renewable energy mutual funds foster the transition to a low-carbon economy? Renewable Energy, 143, 1299–1309. https://doi.org/10.1016/j.renene.2019.05.095 Marti‐Ballester, C. (2019). The role of mutual funds in the sustainable energy sector. Business Strategy & the Environment (John Wiley & Sons, Inc), 28(6), 1107–1120. https://doi.org/10.1002/bse.2305 Molina-Azorín, J. F., Claver-Cortés, E., López-Gamero, M. D., & Tarí, J. J. (2009). Green management and financial performance: A literature review. Management Decision, 47(7), 1080–1100. https://doi.org/10.1108/00251740910978313 Muñoz, F., Vargas, M., & Marco, I. (2014). Environmental Mutual Funds: Financial Performance and Managerial Abilities. Journal of Business Ethics, 124(4), 551–569. https://doi.org/10.1007/s10551-013-1893-x Newey, W. K., & West, K. D. (1987). A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix. Econometrica, 55(3), 703–708. https://doi.org/10.2307/1913610 Ng, A., & Zheng, D. (2018). Let’s agree to disagree! On payoffs and green tastes in green energy investments. Energy Economics, 69, 155–169. https://doi.org/10.1016/j.eneco.2017.10.023 Orlitzky, M., Schmidt, F. L., & Rynes, S. L. (2003). Corporate Social and Financial Performance: A Meta-analysis. Organization Studies, 24(3), 403–441. https://doi.org/10.1177/0170840603024003910 Porter, M. E., & van der Linde, C. (1995). Green and Competitive: Ending the Stalemate. Harvard Business Review, 73(5), 120-133. http://search.ebscohost.com/login.aspx?direct=true&db=edb&AN=9510041980& site=eds-live Reboredo, J. C., Quintela, M., & Otero, L. A. (2017). Do investors pay a premium for going green? Evidence from alternative energy mutual funds. Renewable and Sustainable Energy Reviews, 73, 512–520. https://doi.org/10.1016/j.rser.2017.01.158 Revelli, C., & Viviani, J.-L. (2015). Financial performance of socially responsible investing (SRI): What have we learned? A meta-analysis. Business Ethics, 24(2),