A Year in Review: Financial and Climate Performance of European Union Sustainable Finance Disclosure Regulation (SFDR) Classifications (2021 - 2022)

Kisara Dang

McCombs School of Business

FIN 372: ESG Investing

May 9, 2021

Abstract

This study seeks to examine the financial and climate performance associated with different European Sustainable Finance Disclosure Regulation (SFDR) classifications in the first year of its use, from March 10, 2021, to March 10, 2022. The paper first, discusses the implications of the SFDR classification system, followed by a study utilizing equity funds listed on Bloomberg to analyze the financial and ESG performance of funds over the past year. Bloomberg funds, split into three categories (Articles 6, 8, and 9), were used for statistical analysis in Rstudio including one-way ANOVA tests to determine significant findings, if any. While Article 6 funds had higher percentage returns, Article 9 funds had higher ESG performances, representing a trade-off in the short term.

01 | Introduction

European Union's (EU) Sustainable Finance Disclosure Regulation (SFDR) was created and introduced by the European Commission with the primary objectives of improving transparency and reducing greenwashing concerning the ESG features of investment portfolios. The SFDR imposes mandatory ESG disclosure for financial market participants (FMPs) and became effective beginning March 10, 2021 (KPMG, 2021). The regulation requires asset managers to provide standardized disclosures on ESG factors and risk at the product and entity level. The SFDR manifests into required disclosure on websites, prospectuses, and periodic reports for FMPs.

Funds available for sale in the EU are classified as either Articles 6, 8, or 9 with different sustainability objectives and disclosure standards. Despite a lack of clear policy guidance for product classification, tracking the performance of funds could potentially provide insight into the benefits of the SFDR classification system. Considering that assets in Articles 8 and 9 have reached 4.05 Trillion EUR (42.5% of funds in the EU) (Morningstar 2022), it is important to analyze whether is an effective strategy in the short and long term. This research aims to look at the evolving landscape of Article 8 and 9 funds, particularly analyzing their financial performance, ESG performance, and climate intensity as a full year has passed since its adoption.

Considering the timeline of 2021 to 2022, little similar research exists for comparison. Yet, the paper also fits more broadly into the growing field of ESG research. First, a general description of the SFDR, article categorizations, and its connection to the EU taxonomy will be provided. Then, the literature review connects the study to peer articles and the ESG landscape more broadly. Next, this paper seeks to examine the ESG and financial performance of the three article classifications. The data sample includes equity funds (specifically ETFs) publicly listed on Bloomberg. A one-way ANOVA test and other exploratory statistics were used to test the statistically significant variance of the returns of each asset classification (measured in total return and hypothetical growth); the weighted average carbon intensity (T CO2E/$M Sales); and the ESG quality score. Findings from the statistical analysis suggest a higher financial performance for Article 6 funds. Yet, in examining ESG performance, Article 9 funds performed significantly better in terms of implied temperature rise and ESG scores. However, Article 8 funds tended to be in the middle range with the most variance.

02 | Background Context

EU Sustainable Finance Action Plan (SFAP)

The EU SFAP is a policy objective to promote sustainable investment across the 27 nations in the Union. Understanding the importance of the private sector in the green transition, the European Commission announced its initial sustainable finance strategy in 2018 and the SFAP on March 8, 2018  (Busch, et al. 2021). The SFAP is part of a broader effort to connect finance to the economy for the betterment of society and the planet. The plan proposes major regulatory changes to enable a low-carbon transition and a stable financial system. The action plan outlines ten reforms in three areas: increasing capital flow, making sustainability mainstream in risk management, and increasing transparency (Appendix A). The SFAP offers a series of interlinked regulations to redirect capital into a more sustainable economy.

EU Sustainable Finance Disclosure Regulation (SFDR)

        The SFDR is part of the action plan and imposes ESG disclosures for all FMPs and financial advisers in the EU effective March 10, 2021with other regulations such as the EU taxonomy to follow suit. The SFDR was introduced with the goal of providing greater transparency and standardization on environmental and social characteristics and creating common standards for disclosure. In addition, two other objectives of the regulation are to curb the greenwashing of products and provide investors with the ability to compare investment options (J.P. Morgan Asset Management, 2021). The SFDR disclosures manifest into additional disclosure on websites, prospectuses, and periodic reports. Under the SFDR, firms must disclose how they address two key considerations: principle adverse impacts and sustainability risks (J.P. Morgan Asset Management, 2021). First, “any negative effects that investment decisions or advice could have on sustainability factors”  (J.P. Morgan Asset Management, 2021). For example, investing in a company with poor water practices would constitute a principal adverse impact. Second, sustainability risks refer to the “environmental, social or governance events, or conditions, such as climate change, which could cause a material negative impact on the value of an investment” (J.P. Morgan Asset Management, 2021).

Additionally, investors must classify funds into three distinct categories based on the degree to which sustainability is a consideration. Table 1 summarizes the distinction between the three categories. Both Article 8 and 9 financial products must disclose a variety of ESG topics such as sustainability indications, good governance practices, sustainable investments, benchmarks, carbon emissions, and principal adverse impacts.

Table 1: Article Descriptions

Source: Morningstar Direct, Morningstar Research. Data as of December 2020.

Connection to EU Taxonomy Regulation

The EU taxonomy regulation is a classification system and “tool to help investors, companies, issuers, and project promoters navigate the transition to a low-carbon, resilient and resource-efficient economy” (PRI, 2021). The taxonomy sets performance thresholds referred to as technical screening criteria (TSC). The taxonomy became effective on January 1, 2022, and is integrated with the SFDR, which sets the overarching framework for reporting (J.P. Morgan Asset Management, 2021). The SFDR and taxonomy are designed to be complementary, with the taxonomy to be used for addressing claims made under the SFDR. Once the taxonomy became enacted, Articles 8 and 9 products must state their sustainable investments and whether investments are in activities that align with the taxonomy (J.P. Morgan Asset Management, 2021). Appendix B includes diagrams of how the two are interrelated and connected to the EU’s SFAP.

03 | Literature Review

Considering the timeline of the SFDR, few studies have uncovered its implications. Yet, more broadly, it can be connected to the landscape of ESG disclosure regulation for the EU and on a global scale. First, studies show a lack of a globally accepted taxonomy for sustainable activities that possess a high level of clarity and ability for comparison (Martini 2021). As an effect, ESG integration is discouraged at the global level.  Martini (2021) argues that a “lack of a globally accepted taxonomy on what constitutes sustainable activities and of regulatory clarity, practical complexity, and behavioral issues” is critical in addressing increased standards and access to high-quality data across industries and regions on a global scale.

The EU SFAP is the first step in addressing the issues that Martini (2021) articulates, as they are “the broadest and most comprehensive regulatory initiatives developed in sustainable finance” to date (Geiger 2020). The EU SFDR and taxonomy represent a direct intervention to “actively promote and accelerate sustainable finance” (Driessen 2021). In fact, because of the EU SFAP in March 2018m sustainable finance rose to the top of the EU legislative, regulatory, and supervisory agendas for the EU and national supervisors (Driessen 2021).

Before the EU SFAP,  ESG in financial markets places was “mainly on the principles, guidelines, and initiatives of voluntary application” FMPs. The SFDR is part of a “broader package of legislative tools designed to reorient capital towards more sustainable businesses” (Humpreys, 2021). The SFDR is not intended to regulate market access, but to govern the activities and how they are disclosed. Transparency and disclosure requirements will apply at both the product and entity level for FMPs in terms of their (1) sustainability risks, (2) principle adverse impacts (PAIs), and (3) ESG positioning. According to Doyle (2021), PAIs are “essential to understanding the regulation as they represent a basic SFDR unit,” and can be defined as the “negative effects on sustainability factors that an investment decision or advice might have.” For example, some actions under the SFDR include pre-contractual disclosures on how sustainability is incorporated and a description of their sustainable investments target (Geiger 2020). While the SFDR increases the administrative burden, it also has a “deeper more far-reaching normative impact” on FMPs (Doyle 2021). Entity and product-level disclosures are shown in Table 2 below:

Table 2: SFDR disclosure at product and entity level

Source: Doyle 2021

In addition, FMPs will need to keep dates in mind for different requirements. Appendix J illustrates a timeline for entity-level requirements. Several important dates include:

Source: Doyle 2021

Investor Concerns with SFDR Article Classifications

According to Siri and Zhu (2019), the success of the SFAP on investor protection regulation is endangered by “ the existence of many challenges, weaknesses, and contradictions raised by economists and stakeholders concerning the definition of sustainability, ESG data availability and reliability, the development of an EU taxonomy, conflicts of interest, product governance, and suitability assessment.” Tejas (2021) notes that the “costs associated with compliance with the regulation will ultimately be passed on to investors” with many investors noting their confusion and increasing costs with SFAP regulations and disclosure requirements.  

Indeed, investors are still trying to “get a handle on the multitude of ESG-based regulations coming out of Europe” (Humpreys, 2021). The SFDR will require significant additional disclosure for investors including developing and communicating new processes and procedures enacted by additional disclosure requirements (Doyle 2021). Further, the SFDR will increase the amount of data that needs to be gathered and interpreted (Teja, 2021). According to Florence Fontan, head of company engagement at BNP Paribas Securities Services, “the main issue for asset managers is that they will be required to access ESG data that is not made publicly available by investee companies. The data… is currently difficult to access, sometimes non-existent and generally very costly” (Teja, 2021). The accessibility of data could mean that asset managers will struggle to meet reporting obligations in the short term. Structural gaps in reporting and data collection were found through investor perspectives that must be addressed along the timeline for SFDR disclosure requirements. On the other side, Fontan also noted that the increased regulation will “make it easier for end investors to understand the true ESG and sustainability risks of their investments and make better-informed decisions” (Teja, 2021). Overall, the impact of the SFDR revolution is still in its infancy, with new perspectives and challenges unveiled to come.

A Year in Review

        Two year-long reviews of Article 8 and 9 funds exist. First, Morningstar’s SFDR Article 8 and Article 9 Funds: 2021 in Review provides a review of the evolving landscape of Article 8 and 9 funds with a focus on the last three months of the year. The Morningstar review (2022) also analyzes the ESG credentials of funds. Second, Goldman Sachs' ‘SFDR, one year on The Trends and Anatomy of Article 8 & 9 funds’ describes the growth and flows into Article 8 and 9 funds. The Goldman Sachs report also examines client feedback which shows the confusion and difficulty in addressing SFDR reporting requirements (Tylenda, et al., 2022). In all, both provide an overview of the effects of SFDR classifications on the market over the past year.

        First, both describe the capital flow towards Article 8 and 9 investments. According to Morningstar (2022), assets in Articles 8 and 9 funds have reached 4.05 trillion and represent 42.4% of assets sold in the EU. The Goldman Sachs report notes that flows into Articles 8 and 9 have significantly outpaced non-ESG funds which present more than double the number of funds (Tylenda, et al., 2022). Based on SFDR data collected from prospectuses, 25.2% of funds were classified as Article 8, and 3.4% of funds were classified as Article 9 (Morningstar 2022). Both categories leaned more towards equity but varied in terms of asset class exposure. Additionally, SFDR spurred product innovation and development with 600 new Article 8 or 9 funds entering the market since March 2021 (Morningstar 2022). Another way capital has moved is through the reclassification of strategies as asset managers reclassify Article 6 funds into Articles 8 or 9 through increasing ESG integration. Around 1,800 funds were upgraded from Article 6 to Articles 8 or 9 since March 2021; some funds have even used SFDR to overhaul their strategies and align with sustainable objectives (Morningstar 2022). Amundi, Nordea, and Swedbank dominate and are the largest providers of Article 8 and 9 funds. The acceleration of flow into SFDR funds has implications for the capital flows from the non-ESG sector to ESG-friendly sectors (Tylenda, et al., 2022). The Goldman Sachs report predicts (1) greater differentiation in ESG strategies and (2) Taxonomy adoption as a catalyst for increasing capital in sustainability (Tylenda, et al., 2022).

        Second, looking at the ESG performance through ESG risk, Morningstar uses metrics to assess the ESG characteristics of Article 8 and 9 funds. In aggregate, Articles 8 and 9 perform better on ESG metrics than Article 6 funds (Morningstar 2022). Further, Article 9 offers the highest sustainability credentials (Morningstar 2022). In particular, the rating distribution for sustainability ratings is skewed towards Article 8 and 9 funds; 70.4% of Article 9 funds received the highest ratings for Morningstar (2022). Morningstar also examined the controversy levels of funds as the “do no significant harm” principle is a key element of the SFDR. The Morningstar report (2022) examines exposure to controversial weapons, tobacco, severe controversies, thermal coal involvement, fossil fuel involvement, and carbon solutions involvement. One particularly interesting finding is the higher fossil fuel involvement of Article 9 funds in fossil fuels; this is because these funds are invested in transitioning companies. In all, Article 8 and 9 funds performed better on ESG risk and controversy in comparison to Article 6 funds.

Third, both articulate the challenges and confusions related to a lack of clear policy guidance. The Goldman Sachs report describes the “significant confusion and difficulty in interpreting and addressing SFDR reporting requirements” as indicated through client feedback (Tylenda, et al., 2022). The report outlines how funds are addressing “ 1) classifying Article 8 and 9 funds; 2) incorporating Principle Adverse Impacts (PAIs); 3) conducting Do No Significant Harm analysis; 4) ensuring Good Governance; 5) selecting and disclosing reference benchmarks, and 6) addressing EU Taxonomy reporting” (Tylenda, et al., 2022). In addition, the Morningstar report (2022) articulates the different interpretations of definitions that asset managers have taken (Morningstar 2022). In all, classifying funds remains a challenge as the regulatory language is vague, and the EU SFAP implies ongoing changes for what firms will be required to disclose.

04 | Methodology and Data

The study employs quantitative research methods. The data are equity funds, predominately ETFs, from Bloomberg split into the categorical variables of Articles 6, 8, and 9. Equity funds are chosen as they represent the majority of Article 8 and 9 funds by asset class; equity funds account for about half of Article 8 funds and around two-thirds of Article 9 funds (Morningstar 2021). The data capture the performance from March 10, 2021, to March 10, 2022, and onwards. In total, there were fifty samples collected. The dataset (Appendix C) includes the variables (Appendix D): SFDR Classification; expense ratio (%); one-year hypothetical growth of $10,000 (from March 10, 2021, to March 10, 2022) (%); one-year annualized total return (%); one-year benchmark return (%); P/E ratio; equity beta; MSCI ESG fund rating (CCC to AAA); MSCI weighted average carbon intensity (Tons CO2e/sales); and MSCI implied temperature rise (ITR) (0-3.0+ °C).  

On a broad level, this research seeks to analyze the financial and ESG performance of Equity Funds classified as Articles 8 and 9 and how they compare to the performance of Article 6 funds, which represent the larger market. Considering the information available, the data will first be collected, cleaned, and analyzed in Excel (Appendix C). After, data will be analyzed predominately with R-studio (Appendix F) to conduct various tests using SFDR classification as the independent variable.

Data Exploration

Data was first summarized in Excel using a pivot table showing the means of each group for quantitative variables (Appendix E). Table 3 shows a summary of the means collected from the pivot table. The variables shown were also used for one-way ANOVA testing against the SFDR classification.

Table 3: Pivot Table Means by SFDR Classification

Data was then explored and described in Rstudio using summary statistics and using the “psych” package (Appendix I). For instance, the mean annualized total return for the sample set is 1.00%.  

Two-way Interactions Between Categorical Variables

How do ITR and ESG scores vary by SFDR classification? To answer this question, two two-way interactions were done between categorical variables: ESG fund ratings against SFDR classifications (Appendix H2) and ITR against SFDR classifications (Appendix H3). After creating tables, box plots were made to visualize how instances of each varied with SFDR classifications (Appendix H).

One-way ANOVA tests         

        Analysis of variance, (ANOVA) testing is useful in determining if the difference between groups is larger than the difference within groups and identifying if there is any statistical significance. The independent variable is SFDR classifications, a categorical variable. Five one-way ANOVA tests were run to analyze the interaction between SFDR classifications and other quantitative variables on interest: SFDR Classification; one-year hypothetical growth of $10,000 (from March 10, 2021, to March 10, 2022) (%); one-year annualized total return (%); one-year benchmark return (%); MSCI weighted average carbon intensity (Tons CO2e/sales); and MSCI ESG quality score (1 - 10). The first three variables are financial performance metrics. The last two are ESG performance metrics. The weighted average carbon intensity is an ESG performance metric measured by estimated emissions per $1 million in sales (MSCI). Meanwhile, the ESG quality score is a weighted average of the ESG scores of fund holds with consideration to funding exposures to laggard categories and ESG rating trends.

        The ANOVA tests consisted of four parts. First, the assumptions of an ANOVA test must be met; normality and equal variance were tested by plotting and analyzing the standard deviations between groups (Appendix H1). Second, box plots were created for each variable to visualize the data (Appendix H4, H5). Third, ANOVA tests were run on the four variables. The hypotheses are below:

: mean dependent variable will be equal amongst SFDR classifications

: at least one mean dependent variable will not be equal amongst SFDR classifications

The hypothesis implies there must be statistically significant variance between each asset classification. A critical F-value of 3.19958 and a p-value of .05 will be used for a 95% confidence level. Lastly, for the results with statistical significance, Tukey’s Honest Significant Differences (Tukey HSD) were performed. The Tukey HSD function reports a p-value for each pairwise comparison with a 95% confidence interval. The elements of the Tukey HSD are:

• diff: the difference between means of the two groups
• lwr: the lower endpoint of the confidence interval at 95%
• upr: the upper endpoint of the confidence interval at 95%
• p-adj: p-value after adjustment for the multiple comparisons

Using an alpha value of .05 for a confidence interval of 95%, any p-value below this can be rejected which suggests that the two groups are different. Notably, a Tukey HSD test was not conducted for the 1Y Hypothetical Growth of $10,000 as the results are not statistically significant.

Limitations of Study

The limitations lie in the scope and data collection of the study. It is a rather rudimentary analysis with ETFs only found on Bloomberg which creates significant bias. Further, data samples were collected by hand, which can lead to human error. Further analysis would also be more consistent and consider confounding variables. Lastly, a larger sample size including funds from other asset managers would provide more statistical validity and better model the real world.

Other limitations of the study include time and geographic implications. Again, only funds listed in the EU with an SFDR classification were used and pulled from Bloomberg’s equity funds. While Bloomberg controls for certain factors such as how different numbers were calculated and provided consistency, the sample set is biased as a result and does not represent the real world. Second, the study was run only for the year following the implementation of the SFDR. Finally, the study is rudimentary in its analysis, rather than a robust consideration of the implications of the SFDR.

Lastly, considering the time element of the study, an event study may have been conducted as a substitute to look at the mean abnormal cumulative returns. However, the study was interested in the variance between groups in particular. In addition, CUSIP identifications were not available for all samples. Further analysis may consider the viability of an event study for EU SFAP announcements or enactment dates.

05 | Data Analysis and Results

The overall study is concerned with SFDR classifications’ interactions with financial and ESG performance. Financial performance was analyzed using annualized returns as a percentage using one-way ANOVA tests. The ESG performance was analyzed using tables and one-way ANOVA tests for quantitative variables. Article 9 funds had higher ESG performances, while Article 8 funds tended to have more variation and fall in the middle. In addition, Article 6 funds had a higher financial performance with no statistical significance found in the returns between Article 8 and 9 funds. Overall, the results suggest a trade-off in the short term between the financial performance of Article 6 funds with the higher ESG disclosure and the performance of Article 8 and 9 funds.

SFDR Classifications and Financial Performance

As explained below, the results of the one-way ANOVA tests suggest that the null hypothesis that financial return does not differ between SFDR classifications can be rejected. Appendix H4 shows the box plots for the three indicators; Article 6 has a much higher mean return than Articles 8 and 9. Meanwhile, there seems to be no statistically significant difference in the returns for Articles 8 and 9.

Hypothetical Growth of $10,000: The hypothetical growth reflects the growth of an investment over a period of time, in this case, March 10, 2021, to March 10, 2022. Fund expenses are also deducted from the calculation and the value is expressed in terms of a percentage. The box plot (Appendix H4) shows that Article 6 has a higher mean growth percentage, and Articles 8 and 9’s growth rate was quite similar. However, the ANOVA test shows that the results were not statistically significant as the p-value (0.137) is greater than the alpha level of 0.05. The null hypothesis, that the means for growth differ between classifications, cannot be rejected and a Tukey HSD could not be performed.

Annualized Total Returns: The annual returns represent changes to the net asset value over a period of time, in this case, one year, expressed in a percentage. The box plot (Appendix H4) shows, again, that Article 6 funds have a higher return percentage. In addition, using the ANOVA test, the F value was 8.75 > the F critical of 3.20 and the p-value of 0.006 < the alpha level of 0.05. Therefore, the results were statistically significant and a Tukey HSD function could be used. Article 6 funds have the largest difference from Articles 8 and 9. However, the difference between Articles 8 and 9 had a p-value of 0.99 meaning there was no statistical significance between the two. While there is no significant change in total returns between Articles and 9, Article 6 funds had a significantly better financial performance using annualized total returns as an indicator.

Annualized Benchmark Returns: The benchmark returns are the returns in comparison to a benchmark portfolio. Once again, Article 6 funds have a much higher return percentage. In addition, the ANOVA tests also suggest that the null that means did not differ can be rejected as the p-value of 0.001 < the alpha level of 0.05 and the F-value of 7.304 > the F critical of 3.200. The Tukey HSD shows similar results to the one ran for annualized total returns with no statistical significance between Article 8 and 9 funds (p-value = .869).

SFDR Classifications and ESG Performance

The following analysis refers to an examination of how SFDR classifications interact with ESG scores, ITR, and weighted average carbon intensity. For ESG scores and ITR, Article 9 funds performed better than Articles 6 and 8 with Article 6 often performing the worst. Article 8 funds tended to fall in the middle of Articles 6 and 9 with much more variance. However, Article 8 funds had the lowest carbon intensity, but the means between Articles 8 and 9 were not statistically significant. Overall, the null hypothesis that ESG performance did not differ between the SFDR classifications can be rejected as Article 6 funds tend to have both lower ESG scores, ITRs, and carbon intensities. Box plots of each analysis can be found in Appendix H.

ESG Scores: For ESG fund ratings, all the funds in the sample had an ESG score of BBB, A, AA, or AAA. Funds with BBB or A scores represent average funds with firms with mixed or unexceptional records of managing ESG risk and opportunities in comparison to peers. AA and AAA scorers are funds that have leader firms that manage ESG risks or opportunities most significantly. As shown in the graph (Appendix H2), the majority of Article 6 funds have an AA rating, but four funds have a score of BBB. In addition, the majority of Article 9 funds (90%) have an ESG score of AAA which is the highest possible. Article 8 funds have the most in variation of ESG scores with the majority having either an ESG score of AA or AAA (95%). In all, it seems that Article 9 funds, which have the highest level of sustainability disclosure, have higher ESG scores than Article 6 and 8 funds.

The ESG quality score correlates with the ESG fund ratings (CCC to AAA) directly. Looking at the box plot of the ESG quality scores (Appendix H5), Article 6 has the lowest mean score (7.31) while Article 9 (9.29) has the highest mean score with only one outlier. As expected, Article 8 scores fall in the middle with a mean score of 8.14. Looking at the ANOVA test, the F value is 8.322 > the F-critical value of 3.200 and the p-value of 0.000805 < the alpha level of .05 demonstrating statistical significance. The Tukey HSD function confirms the statistically significant difference in means between Articles 6 and 9 with a p-value of 0.0005. Therefore, we can reject the null that there is no difference between ESG scores for different SFDR classifications.

Implied Temperature Rise (ITR): ITR is split into four categories: 1.5 - 2.0°C, 2.0 - 2.5°C, 2.5 - 3.0°C, 3.0°C+. Creating a table of ITR split by SFDR classifications, the results show that 80% of Article 6 funds have an ITR of 2.5°C or higher. Again, Article 8 funds were the most varied with the majority (70%) falling in the range of 2.0 - 3.0°C. Lastly, Article 9 has a majority (90%) of its funds with a lower ITR of 2.5°C or below. ITR, which measures alignment with the temperature goals of the Paris Agreement, is used to show the potential of a fund to reduce emissions over time. Article 9 funds have lower ITRs which suggest better future climate performance than Article 6 and 8 funds. Appendix H3 shows a plot of the ITR split by SFDR classifications.  

Weighted Average Carbon Intensity: The variable is expressed in Tons/CO2e/$M in sales or tons of carbon dioxide equivalents per one million in sales. First, the box plot (Appendix H5) shows that Article 8 funds have the lowest mean carbon intensity of 76.5405. Yet, the Article 9 funds have the least amount of variance and a mean carbon intensity of 107.16. On the other hand, Article 6 funds have the highest mean carbon intensity of 207.43 with the widest range as well. The F-value of 9.152 > 3.200 and p-value of 0.00044 < the alpha level of 0.05 demonstrates statistical significance. The Tukey HSD was then used which shows Article 6 has the largest mean difference from Articles 8 and 9, which were both statistically significant. However, the difference between means (30.62) for Articles 8 and 9 was not statistically significant (p-value = 0.71). The results of the ANOVA test and Tukey HSD function show that Article 6 funds have a much higher mean carbon intensity than both Articles 8 and 9, with Article 8 funds having the lowest mean carbon intensity, yet more variance.

Future Area of Exploration

There is many directions of future exploration that can be taken. For instance, the EU Sustainable Finance Action Plan is still in its infancy, so future analysis may have long time scales and more robust statistical analysis of other factors as well. It may also be of interest to track the movement of Article 8 and 9 inflows and outflows as the EU taxonomy is finalized.

06 | Conclusion

A year has passed since the rolling out of SFDR disclosure requirements. Again, the main objective of the SFDR is to ensure FMPs can achieve sustainable financial growth in the long term while combatting greenwashing and increasing transparency. As part of the SFDR, asset managers must keep adverse impacts into account and must report on the entity and product level. First, the study seeks to connect the SFDR requirements to the broader ESG regulatory landscape. Then, the literature review unveils current challenges and perspectives related to SFDR classification with two year-long reviews that provide the foundations for the study. The study conducted then uses quantitative analysis to unveil and analyze the financial and ESG performance of different SFDR classifications. The study conducted provides validity for the assumption that funds with an Article 9 classification have a higher ESG performance than its counterparts. As of now, it seems to be at a trade-off as Article 6 funds have better financial performance in the short term. However, it is still unclear how this will play out in the long term as EU SFAP requirements continue to shift.

07 | Bibliography

Broome, J. (2022, March 3). EU taxonomy in limbo - reporting alignment of Article 8 and 9 funds in 2022. Retrieved May 14, 2022, from https://www.sustainalytics.com/esg-research/resource/investors-esg-blog/eu-taxonomy-alignment-of-article-8-and-9-funds-reporting 

Busch, D., Ferrarini, G., van den Hurk, A. (2021). The European Commission’s Sustainable Finance Action Plan and Other International Initiatives. In: Busch, D., Ferrarini, G., Grünewald, S. (eds) Sustainable Finance in Europe. EBI Studies in Banking and Capital Markets Law. Palgrave Macmillan, Cham. https://doi-org.ezproxy.lib.utexas.edu/10.1007/978-3-030-71834-3_2 

Doyle, D. (2021, April 01). What is the impact of the EU Sustainable Finance Disclosure Regulation (SFDR)? Retrieved May 14, 2022, from https://www.spglobal.com/marketintelligence/en/news-insights/blog/what-is-the-impact-of-the-eu-sustainable-finance-disclosure-regulation-sfdr 

Driessen, M. (2021). Sustainable Finance: An Overview of ESG in the Financial Markets. In: Busch, D., Ferrarini, G., Grünewald, S. (eds) Sustainable Finance in Europe. EBI Studies in Banking and Capital Markets Law. Palgrave Macmillan, Cham. https://doi-org.ezproxy.lib.utexas.edu/10.1007/978-3-030-71834-3_10 

European Commission. (2021, July 6). Communication From the Commission. Retrieved May 14, 2022, from https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=CELEX%3A52021DC0390&from=BG 

Geiger, S. (2020, December 09). EU action plan – deep dive on disclosure regulation (SFDR). Retrieved May 14, 2022, from https://www.ey.com/en_ch/sustainability-financial-services/eu-action-plan-deep-dive-on-disclosure-regulation-sfdr 

Humphreys, N. (2021, February 5). Demystifying the Sustainable Finance Disclosure Regulation. Retrieved May 14, 2022, from https://www.bloomberg.com/professional/blog/demystifying-the-sustainable-finance-disclosure-regulation/ 

J.P. Morgan Asset Management. (2021, June 30). Explaining the European Union Sustainable Finance Disclosure Regulation (EU SFDR). Retrieved May 14, 2022, from https://am.jpmorgan.com/kr/en/asset-management/institutional/investment-strategies/sustainable-investing/understanding-SFDR/ 

KPMG. (2021). SFDR - a Snapshot : Changes Effective from 10 March 2021. Retrieved May 14, 2022, from https://assets.kpmg/content/dam/kpmg/ie/pdf/2021/03/ie-sustainable-finance-disclosure-reg-sfdr.pdf 

Martini, A. (2021). Socially responsible investing: From the ethical origins to the sustainable development framework of the european union. Environment, Development and Sustainability, 23(11), 16874-16890. doi:https://doi.org/10.1007/s10668-021-01375-3

Morningstar Manager Research. (2022, February 4). SFDR Article 8 and Article 9 funds: 2021 in review a rapidly evolving ... Retrieved May 14, 2022, from https://assets.contentstack.io/v3/assets/blt4eb669caa7dc65b2/blt989542c4fba32038/6203ce883d56844907a2b5e3/SFDR-Article-8-and-Article-9-Funds-2021-in-Review.pdf 

Siri, & Zhu. (2019). Will the EU Commission Successfully Integrate Sustainability Risks and Factors in the Investor Protection Regime? A Research Agenda. Sustainability (Basel, Switzerland), 11(22), 6292–. https://doi.org/10.3390/su11226292 

Teja, N. (2021). Primer: The EU’s sustainable finance disclosure regulation. International Financial Law Review, Retrieved from http://ezproxy.lib.utexas.edu/login?url=https://www-proquest-com.ezproxy.lib.utexas.edu/trade-journals/primer-eu-s-sustainable-finance-disclosure/docview/2511011295/se-2?accountid=7118 

Tylenda, E., Chen, G., Meyer, M., Aggarwal, R., & Corbett, B. (2022, March 10). SFDR, one year on: The Trends and Anatomy of Article 8 & 9 funds. Retrieved May 14, 2022, from https://www.goldmansachs.com/insights/ 

Appendices

Appendix A: EU Sustainable Finance Action Plan (SFAP) Reforms

Table: Action Plan Reforms

Source: Principles for Responsible Investing (PRI)

According to the PRI, the EU Action Plan involves the following 10 actions:

1. Sustainability taxonomy
2. Standards and labels
3. Fostering investments in sustainable projects
4. Incorporating sustainability when providing financial advice
5. Developing sustainability benchmarks
6. Better integrating sustainability into ratings and research
7. Clarifying institutional investors and asset managers’ duties
8. Incorporating sustainability into prudential requirements
9. Strengthening sustainability disclosure and accounting rule-making
10. Fostering sustainable corporate governance and attenuating short-termism in capital markets

Appendix B: Interrelations between EU Taxonomy and SFDR

Source: European Commission 2021

Table: EU sustainability disclosure regime for financial and non-financial companies

Source: European Commission 2021

Appendix C: Excel Dataset

Appendix D: Variable Descriptions

Developed from Blackrock Variable Descriptions

Appendix E: Pivot Table of Means

Appendix F: RStudio Code

eudata <- FIN_372_Data

#factoring categorical

as.factor(eudata$`SFDR Classification`)

factor(eudata$`SFDR Classification`, levels = c("Article 6", "Article 8", "Article 9"))

as.factor(eudata$`MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022`)

factor(eudata$`MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022`, levels = c("BBB", "A", "AA", "AAA"), ordered = TRUE)

as.factor(eudata$`MSCI Implied Temperature Rise (0-3.0+ °C)`)

factor(eudata$`MSCI Implied Temperature Rise (0-3.0+ °C)`, levels = c("> 1.5 - 2.0", "> 2.0 - 2.5", "> 2.5 - 3.0", "> 3.0"), ordered = TRUE)

#PART I: DATA EXPLORATION

summary(eudata)

install.packages("psych")

library(psych)

describe(eudata)

#correlation between return and carbon intensity

cor(eudata$`1Y Growth of 10K (3/10/21 - 3/10/22)`, eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)`)

plot(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized`,eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)`)

cor(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized`,eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)`)

aggregate(eudata$`P/E Ratio`,list(eudata$`SFDR Classification`), FUN=mean)

#table of two-way interaction of ESG Score and SFDR Classification

esgscore <- table(eudata$`MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022`, eudata$`SFDR Classification`)

esgscore

barplot(esgscore, legend.text = TRUE, beside = TRUE,

        xlab = 'SFDR Classification',

        main = "ESG Scores of Different SFDR Classifications",

        col = c("#bf5700", "#005f86", "#a6cd57", "#00a9b7"))

addmargins(esgscore)

prop.table(esgscore, 2) # by column

prop.table(esgscore, 1) # by row

barplot(prop.table(esgscore, margin = 2),

        legend.text = TRUE,

        ylab = "Proportion",

        col = c("#bf5700", "#005f86", "#a6cd57", "#00a9b7"))

#implied temperature and SFDR Classification two-way interaction

implied <- table(eudata$`MSCI Implied Temperature Rise (0-3.0+ °C)`, eudata$`SFDR Classification`)

implied

barplot(implied, xlab = 'SFDR Classification',

        main = "Implied Temperature",

        legend.text = TRUE,

        beside = TRUE,

        col = c("#bf5700", "#005f86", "#a6cd57", "#00a9b7"))

#PART II: ANOVA TESTING

#ANOVA tests whether any of the group means are

#different from the overall mean of the data by checking

#the variance of each individual group against the overall

#variance of the data. If one or more groups falls outside

#the range of variation predicted by the null hypothesis

#(all group means are equal), then the test is statistically significant.

#10K Hypothetical Growth

tenaov <- aov(eudata$`1Y Growth of 10K (3/10/21 - 3/10/22)` ~ eudata$`SFDR Classification`, data = eudata)

summary(tenaov) #results were not statistically significant

boxplot(eudata$`1Y Growth of 10K (3/10/21 - 3/10/22)` ~ eudata$`SFDR Classification`, data = eudata,

        xlab = "SFDR Classification", ylab = "Return - Hyp. Growth of 10K (%)",

        main = "Mean 1Y Growth of 10K of each SFDR Classification",

        frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

aggregate(eudata$`1Y Growth of 10K (3/10/21 - 3/10/22)`,list(eudata$`SFDR Classification`), FUN=mean) #means by group

#Weighted Average Carbon Intensity

tempaov <- aov(eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)` ~ eudata$`SFDR Classification`)

summary(tempaov)

boxplot(eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)` ~ eudata$`SFDR Classification`, data = eudata,

        xlab = "SFDR Classification", ylab = "Weighted Average Carbon Intensity (Tons CO2E/$M SALES)",

        main = "Mean Carbon Intensity of each SFDR Classification",

        frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

#As the ANOVA test is significant, we can compute Tukey HSD

#(Tukey Honest Significant Differences, R function: TukeyHSD()) for

#performing multiple pairwise-comparison between the means of groups.

#diff: difference between means of the two groups

#lwr, upr: the lower and the upper end point of the confidence interval at 95% (default)

#p adj: p-value after adjustment for the multiple comparisons.

TukeyHSD(tempaov)

aggregate(eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)`,list(eudata$`SFDR Classification`), FUN=mean)

#Annualized Total Return

annualaov <- aov(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized` ~ eudata$`SFDR Classification`)

summary(annualaov) #stat significance

TukeyHSD(annualaov)

boxplot(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized` ~ eudata$`SFDR Classification`, data = eudata,

        xlab = "SFDR Classification", ylab = "1Y  Total Return (%) as of 4/30/2022 - Annualized",

        main = "Mean Total Return of each SFDR Classification",

        frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

aggregate(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized`,list(eudata$`SFDR Classification`), FUN=mean)

by(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized`, eudata$`SFDR Classification`, summary) #another way

#Annualized Benchmark Return

benchaov <- aov(eudata$`1Y Benchmark Return (%) as of 3/31/2023 - Annualized` ~ eudata$`SFDR Classification`)

summary(benchaov) #stat significance

TukeyHSD(benchaov)

boxplot(eudata$`1Y Benchmark Return (%) as of 3/31/2023 - Annualized` ~ eudata$`SFDR Classification`, data = eudata,

        xlab = "SFDR Classification", ylab = "1Y Benchmark Return (%) as of 3/31/2023 - Annualized`",

        main = "Mean Benchmark Return of each SFDR Classification",

        frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

aggregate(eudata$`1Y Benchmark Return (%) as of 3/31/2023 - Annualized`,list(eudata$`SFDR Classification`), FUN=mean)

#ESG Quality Score

esgaov <- aov(eudata$`MSCI ESG Quality Score (0-10) - 4/7/2022` ~ eudata$`SFDR Classification`)

summary(esgaov)

TukeyHSD(esgaov)

boxplot(eudata$`MSCI ESG Quality Score (0-10) - 4/7/2022` ~ eudata$`SFDR Classification`, data = eudata,

        xlab = "SFDR Classification", ylab = "MSCI ESG Quality Score (0-10)",

        main = "Mean ESG Quality Score of each SFDR Classification",

        frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

aggregate(eudata$`MSCI ESG Quality Score (0-10) - 4/7/2022`,list(eudata$`SFDR Classification`), FUN=mean)

#testing normality

plot(benchaov, 2)

plot(annualaov, 2)

plot(tenaov, 2)

plot(tempaov, 2)

Appendix G: RStudio Output

> eudata <- FIN_372_Data

>

> #factoring categorical

> as.factor(eudata$`SFDR Classification`)

 [1] Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6

[10] Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6

[19] Article 6 Article 6 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8

[28] Article 8 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8

[37] Article 8 Article 8 Article 8 Article 8 Article 9 Article 9 Article 9 Article 9 Article 9

[46] Article 9 Article 9 Article 9 Article 9 Article 9

Levels: Article 6 Article 8 Article 9

> factor(eudata$`SFDR Classification`, levels = c("Article 6", "Article 8", "Article 9"))

 [1] Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6

[10] Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6 Article 6

[19] Article 6 Article 6 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8

[28] Article 8 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8 Article 8

[37] Article 8 Article 8 Article 8 Article 8 Article 9 Article 9 Article 9 Article 9 Article 9

[46] Article 9 Article 9 Article 9 Article 9 Article 9

Levels: Article 6 Article 8 Article 9

> as.factor(eudata$`MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022`)

 [1] BBB AA  AAA AA  BBB AA  AA  AA  AA  AA  AAA BBB AA  A   AA  BBB AA  AA  AA  AA  AA  AAA BBB

[24] AAA AAA BBB AA  AAA AA  AAA AA  AAA AAA AA  AA  AA  AA  AAA A   AAA AAA AAA AAA A   AAA AAA

[47] AAA AAA AAA AAA

Levels: A AA AAA BBB

> factor(eudata$`MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022`, levels = c("BBB", "A", "AA", "AAA"), ordered = TRUE)

 [1] BBB AA  AAA AA  BBB AA  AA  AA  AA  AA  AAA BBB AA  A   AA  BBB AA  AA  AA  AA  AA  AAA BBB

[24] AAA AAA BBB AA  AAA AA  AAA AA  AAA AAA AA  AA  AA  AA  AAA A   AAA AAA AAA AAA A   AAA AAA

[47] AAA AAA AAA AAA

Levels: BBB < A < AA < AAA

> as.factor(eudata$`MSCI Implied Temperature Rise (0-3.0+ °C)`)

 [1] > 3.0       > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0 > 3.0       > 2.0 - 2.5 > 2.5 - 3.0

 [8] > 2.0 - 2.5 > 3.0       > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0

[15] > 2.0 - 2.5 > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0 > 2.0 - 2.5 > 2.5 - 3.0 > 2.0 - 2.5

[22] > 2.5 - 3.0 > 3.0       > 2.5 - 3.0 > 2.0 - 2.5 > 3.0       > 1.5 - 2.0 > 2.5 - 3.0

[29] > 1.5 - 2.0 > 2.0 - 2.5 > 2.0 - 2.5 > 2.0 - 2.5 > 1.5 - 2.0 > 2.5 - 3.0 > 2.5 - 3.0

[36] > 1.5 - 2.0 > 2.5 - 3.0 > 2.0 - 2.5 > 2.0 - 2.5 > 2.5 - 3.0 > 2.0 - 2.5 > 2.0 - 2.5

[43] > 2.0 - 2.5 > 1.5 - 2.0 > 2.0 - 2.5 > 2.0 - 2.5 > 2.0 - 2.5 > 2.0 - 2.5 > 2.0 - 2.5

[50] > 3.0      

Levels: > 1.5 - 2.0 > 2.0 - 2.5 > 2.5 - 3.0 > 3.0

> factor(eudata$`MSCI Implied Temperature Rise (0-3.0+ °C)`, levels = c("> 1.5 - 2.0", "> 2.0 - 2.5", "> 2.5 - 3.0", "> 3.0"), ordered = TRUE)

 [1] > 3.0       > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0 > 3.0       > 2.0 - 2.5 > 2.5 - 3.0

 [8] > 2.0 - 2.5 > 3.0       > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0

[15] > 2.0 - 2.5 > 2.5 - 3.0 > 2.5 - 3.0 > 2.5 - 3.0 > 2.0 - 2.5 > 2.5 - 3.0 > 2.0 - 2.5

[22] > 2.5 - 3.0 > 3.0       > 2.5 - 3.0 > 2.0 - 2.5 > 3.0       > 1.5 - 2.0 > 2.5 - 3.0

[29] > 1.5 - 2.0 > 2.0 - 2.5 > 2.0 - 2.5 > 2.0 - 2.5 > 1.5 - 2.0 > 2.5 - 3.0 > 2.5 - 3.0

[36] > 1.5 - 2.0 > 2.5 - 3.0 > 2.0 - 2.5 > 2.0 - 2.5 > 2.5 - 3.0 > 2.0 - 2.5 > 2.0 - 2.5

[43] > 2.0 - 2.5 > 1.5 - 2.0 > 2.0 - 2.5 > 2.0 - 2.5 > 2.0 - 2.5 > 2.0 - 2.5 > 2.0 - 2.5

[50] > 3.0      

Levels: > 1.5 - 2.0 < > 2.0 - 2.5 < > 2.5 - 3.0 < > 3.0

>

> #PART I: DATA EXPLORATION

> summary(eudata)

     Name           SFDR Classification Expense Ratio    1Y Growth of 10K (3/10/21 - 3/10/22)

 Length:50          Length:50           Min.   :0.0300   Min.   :-0.20080                    

 Class :character   Class :character    1st Qu.:0.1200   1st Qu.:-0.09115                    

 Mode  :character   Mode  :character    Median :0.1900   Median :-0.01860                    

                                        Mean   :0.2222   Mean   :-0.01048                    

                                        3rd Qu.:0.3025   3rd Qu.: 0.06210                    

                                        Max.   :0.6500   Max.   : 0.15420                    

 1Y  Total Return (%) as of 4/30/2022 - Annualized

 Min.   :-0.29460                                

 1st Qu.:-0.10065                                

 Median :-0.02560                                

 Mean   :-0.01543                                

 3rd Qu.: 0.06388                                

 Max.   : 0.23930                                

 1Y Benchmark Return (%) as of 3/31/2023 - Annualized   P/E Ratio     Equity Beta (3Y)

 Min.   :-0.293200                                    Min.   :11.15   Min.   :0.7900  

 1st Qu.:-0.066525                                    1st Qu.:13.44   1st Qu.:0.9980  

 Median :-0.004300                                    Median :15.97   Median :0.9990  

 Mean   : 0.004552                                    Mean   :16.73   Mean   :0.9963  

 3rd Qu.: 0.105725                                    3rd Qu.:19.19   3rd Qu.:1.0000  

 Max.   : 0.287200                                    Max.   :28.94   Max.   :1.2000  

                                                                      NA's   :1      

   P/B Ratio      MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022

 Min.   : 1.280   Length:50                                

 1st Qu.: 1.683   Class :character                        

 Median : 2.365   Mode  :character                        

 Mean   : 2.734                                            

 3rd Qu.: 3.283                                            

 Max.   :11.000                                            

 MSCI ESG Quality Score (0-10) - 4/7/2022

 Min.   : 4.760                          

 1st Qu.: 7.360                          

 Median : 8.175                          

 Mean   : 8.039                          

 3rd Qu.: 9.092                          

 Max.   :10.000                          

 MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)

 Min.   : 11.14                                            

 1st Qu.: 59.50                                            

 Median :104.61                                            

 Mean   :135.02                                            

 3rd Qu.:161.19                                            

 Max.   :661.19                                            

 MSCI Implied Temperature Rise (0-3.0+ °C)

 Length:50                                

 Class :character                        

 Mode  :character                        

>

> install.packages("psych")

  There is a binary version available but the source version is later:

      binary source needs_compilation

psych  2.2.3  2.2.5             FALSE

installing the source package ‘psych’

trying URL 'https://cran.rstudio.com/src/contrib/psych_2.2.5.tar.gz'

Content type 'application/x-gzip' length 1566977 bytes (1.5 MB)

==================================================

downloaded 1.5 MB

* installing *source* package ‘psych’ ...

** package ‘psych’ successfully unpacked and MD5 sums checked

** using staged installation

** R

** data

*** moving datasets to lazyload DB

** inst

** byte-compile and prepare package for lazy loading

** help

*** installing help indices

** building package indices

** installing vignettes

** testing if installed package can be loaded from temporary location

** testing if installed package can be loaded from final location

** testing if installed package keeps a record of temporary installation path

* DONE (psych)

The downloaded source packages are in

        ‘/private/var/folders/46/qjs5n_sj2cbc3z3gh1jcc_cr0000gn/T/RtmpuBNFkk/downloaded_packages’

> library(psych)

> describe(eudata)

                                                            vars  n   mean     sd median

Name*                                                          1 50  24.30  13.38  24.50

SFDR Classification*                                           2 50   1.80   0.76   2.00

Expense Ratio                                                  3 50   0.22   0.14   0.19

1Y Growth of 10K (3/10/21 - 3/10/22)                           4 50  -0.01   0.09  -0.02

1Y  Total Return (%) as of 4/30/2022 - Annualized              5 50  -0.02   0.12  -0.03

1Y Benchmark Return (%) as of 3/31/2023 - Annualized           6 50   0.00   0.12   0.00

P/E Ratio                                                      7 50  16.73   4.50  15.97

Equity Beta (3Y)                                               8 49   1.00   0.07   1.00

P/B Ratio                                                      9 50   2.73   1.66   2.37

MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022*                    10 50   2.58   0.78   3.00

MSCI ESG Quality Score (0-10) - 4/7/2022                      11 50   8.04   1.44   8.18

MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)   12 50 135.02 114.80 104.62

MSCI Implied Temperature Rise (0-3.0+ °C)*                    13 50   2.54   0.84   3.00

                                                            trimmed   mad   min    max  range

Name*                                                         24.38 16.31  1.00  47.00  46.00

SFDR Classification*                                           1.75  1.48  1.00   3.00   2.00

Expense Ratio                                                  0.21  0.10  0.03   0.65   0.62

1Y Growth of 10K (3/10/21 - 3/10/22)                          -0.01  0.11 -0.20   0.15   0.36

1Y  Total Return (%) as of 4/30/2022 - Annualized             -0.01  0.12 -0.29   0.24   0.53

1Y Benchmark Return (%) as of 3/31/2023 - Annualized           0.00  0.12 -0.29   0.29   0.58

P/E Ratio                                                     16.23  4.43 11.15  28.94  17.79

Equity Beta (3Y)                                               0.99  0.00  0.79   1.20   0.41

P/B Ratio                                                      2.45  1.02  1.28  11.00   9.72

MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022*                     2.55  1.48  1.00   4.00   3.00

MSCI ESG Quality Score (0-10) - 4/7/2022                       8.17  1.35  4.76  10.00   5.24

MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)  117.27 67.20 11.14 661.19 650.05

MSCI Implied Temperature Rise (0-3.0+ °C)*                     2.55  1.48  1.00   4.00   3.00

                                                             skew kurtosis    se

Name*                                                       -0.06    -1.20  1.89

SFDR Classification*                                         0.33    -1.23  0.11

Expense Ratio                                                1.06     0.75  0.02

1Y Growth of 10K (3/10/21 - 3/10/22)                        -0.12    -1.06  0.01

1Y  Total Return (%) as of 4/30/2022 - Annualized           -0.13    -0.51  0.02

1Y Benchmark Return (%) as of 3/31/2023 - Annualized         0.03    -0.43  0.02

P/E Ratio                                                    0.90     0.21  0.64

Equity Beta (3Y)                                             0.27     2.70  0.01

P/B Ratio                                                    2.89    10.72  0.24

MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022*                   0.11    -0.57  0.11

MSCI ESG Quality Score (0-10) - 4/7/2022                    -0.65    -0.41  0.20

MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)  2.17     6.70 16.24

MSCI Implied Temperature Rise (0-3.0+ °C)*                  -0.02    -0.65  0.12

>

> #correlation between return and carbon intensity

> cor(eudata$`1Y Growth of 10K (3/10/21 - 3/10/22)`, eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)`)

[1] 0.1146669

> plot(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized`,eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)`)

> cor(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized`,eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)`)

[1] 0.2947159

> aggregate(eudata$`P/E Ratio`,list(eudata$`SFDR Classification`), FUN=mean)

    Group.1       x

1 Article 6 16.7355

2 Article 8 16.3240

3 Article 9 17.5170

>

> #table of two-way interaction of ESG Score and SFDR Classification

> esgscore <- table(eudata$`MSCI ESG Fund Rating (AAA-CCC) - 4/7/2022`, eudata$`SFDR Classification`)

> esgscore

      Article 6 Article 8 Article 9

  A           1         1         1

  AA         13         8         0

  AAA         2         9         9

  BBB         4         2         0

> barplot(esgscore, legend.text = TRUE, beside = TRUE,

+         xlab = 'SFDR Classification',

+         main = "ESG Scores of Different SFDR Classifications",

+         col = c("#bf5700", "#005f86", "#a6cd57", "#00a9b7"))

>

> addmargins(esgscore)

      Article 6 Article 8 Article 9 Sum

  A           1         1         1   3

  AA         13         8         0  21

  AAA         2         9         9  20

  BBB         4         2         0   6

  Sum        20        20        10  50

> prop.table(esgscore, 2) # by column

      Article 6 Article 8 Article 9

  A        0.05      0.05      0.10

  AA       0.65      0.40      0.00

  AAA      0.10      0.45      0.90

  BBB      0.20      0.10      0.00

> prop.table(esgscore, 1) # by row

      Article 6 Article 8 Article 9

  A   0.3333333 0.3333333 0.3333333

  AA  0.6190476 0.3809524 0.0000000

  AAA 0.1000000 0.4500000 0.4500000

  BBB 0.6666667 0.3333333 0.0000000

> barplot(prop.table(esgscore, margin = 2),

+         legend.text = TRUE,

+         ylab = "Proportion",

+         col = c("#bf5700", "#005f86", "#a6cd57", "#00a9b7"))

>

> #implied temperature and SFDR Classification two-way interaction

> implied <- table(eudata$`MSCI Implied Temperature Rise (0-3.0+ °C)`, eudata$`SFDR Classification`)

> implied

              Article 6 Article 8 Article 9

  > 1.5 - 2.0         0         4         1

  > 2.0 - 2.5         4         7         8

  > 2.5 - 3.0        13         7         0

  > 3.0               3         2         1

> barplot(implied, xlab = 'SFDR Classification',

+         main = "Implied Temperature",

+         legend.text = TRUE,

+         beside = TRUE,

+         col = c("#bf5700", "#005f86", "#a6cd57", "#00a9b7"))

>

> #PART II: ANOVA TESTING 

> #ANOVA tests whether any of the group means are

> #different from the overall mean of the data by checking

> #the variance of each individual group against the overall

> #variance of the data. If one or more groups falls outside

> #the range of variation predicted by the null hypothesis

> #(all group means are equal), then the test is statistically significant.

>

> #10K Hypothetical Growth

> tenaov <- aov(eudata$`1Y Growth of 10K (3/10/21 - 3/10/22)` ~ eudata$`SFDR Classification`, data = eudata)

> summary(tenaov) #results were not statistically significant

                             Df Sum Sq  Mean Sq F value Pr(>F)

eudata$`SFDR Classification`  2 0.0336 0.016784   2.075  0.137

Residuals                    47 0.3802 0.008089              

>

> boxplot(eudata$`1Y Growth of 10K (3/10/21 - 3/10/22)` ~ eudata$`SFDR Classification`, data = eudata,

+         xlab = "SFDR Classification", ylab = "Return - Hyp. Growth of 10K (%)",

+         main = "Mean 1Y Growth of 10K of each SFDR Classification",

+         frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

>

> aggregate(eudata$`1Y Growth of 10K (3/10/21 - 3/10/22)`,list(eudata$`SFDR Classification`), FUN=mean) #means by group

    Group.1         x

1 Article 6  0.021025

2 Article 8 -0.028630

3 Article 9 -0.037180

>

> #Weighted Average Carbon Intensity

> tempaov <- aov(eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)` ~ eudata$`SFDR Classification`)

> summary(tempaov)

                             Df Sum Sq Mean Sq F value  Pr(>F)    

eudata$`SFDR Classification`  2 181011   90505   9.152 0.00044 ***

Residuals                    47 464779    9889                    

---

Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

>

> boxplot(eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)` ~ eudata$`SFDR Classification`, data = eudata,

+         xlab = "SFDR Classification", ylab = "Weighted Average Carbon Intensity (Tons CO2E/$M SALES)",

+         main = "Mean Carbon Intensity of each SFDR Classification",

+         frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

>

> #As the ANOVA test is significant, we can compute Tukey HSD

> #(Tukey Honest Significant Differences, R function: TukeyHSD()) for

> #performing multiple pairwise-comparison between the means of groups.

>

> #diff: difference between means of the two groups

> #lwr, upr: the lower and the upper end point of the confidence interval at 95% (default)

> #p adj: p-value after adjustment for the multiple comparisons.

>

> TukeyHSD(tempaov)

  Tukey multiple comparisons of means

    95% family-wise confidence level

Fit: aov(formula = eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)` ~ eudata$`SFDR Classification`)

$`eudata$`SFDR Classification``

                         diff        lwr        upr     p adj

Article 8-Article 6 -130.8845 -206.98918 -54.779818 0.0003858

Article 9-Article 6 -100.2690 -193.47782  -7.060181 0.0323990

Article 9-Article 8   30.6155  -62.59332 123.824319 0.7079680

>

> aggregate(eudata$`MSCI Weighted Average Carbon Intensity (Tons CO2E/$M SALES)`,list(eudata$`SFDR Classification`), FUN=mean)

    Group.1        x

1 Article 6 207.4250

2 Article 8  76.5405

3 Article 9 107.1560

>

> #Annualized Total Return

> annualaov <- aov(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized` ~ eudata$`SFDR Classification`)

> summary(annualaov) #stat significance

                             Df Sum Sq Mean Sq F value   Pr(>F)    

eudata$`SFDR Classification`  2 0.1960 0.09801    8.75 0.000588 ***

Residuals                    47 0.5264 0.01120                    

---

Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

> TukeyHSD(annualaov)

  Tukey multiple comparisons of means

    95% family-wise confidence level

Fit: aov(formula = eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized` ~ eudata$`SFDR Classification`)

$`eudata$`SFDR Classification``

                         diff         lwr         upr     p adj

Article 8-Article 6 -0.128575 -0.20956968 -0.04758032 0.0010461

Article 9-Article 6 -0.126240 -0.22543782 -0.02704218 0.0095011

Article 9-Article 8  0.002335 -0.09686282  0.10153282 0.9982125

>

> boxplot(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized` ~ eudata$`SFDR Classification`, data = eudata,

+         xlab = "SFDR Classification", ylab = "1Y  Total Return (%) as of 4/30/2022 - Annualized",

+         main = "Mean Total Return of each SFDR Classification",

+         frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

>

> aggregate(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized`,list(eudata$`SFDR Classification`), FUN=mean)

    Group.1         x

1 Article 6  0.061250

2 Article 8 -0.067325

3 Article 9 -0.064990

> by(eudata$`1Y  Total Return (%) as of 4/30/2022 - Annualized`, eudata$`SFDR Classification`, summary) #another way

eudata$`SFDR Classification`: Article 6

    Min.  1st Qu.   Median     Mean  3rd Qu.     Max.

-0.11450 -0.00800  0.06760  0.06125  0.13605  0.23930

------------------------------------------------------------------------

eudata$`SFDR Classification`: Article 8

    Min.  1st Qu.   Median     Mean  3rd Qu.     Max.

-0.29460 -0.15050 -0.04565 -0.06732  0.02710  0.16680

------------------------------------------------------------------------

eudata$`SFDR Classification`: Article 9

    Min.  1st Qu.   Median     Mean  3rd Qu.     Max.

-0.17390 -0.12335 -0.04690 -0.06499 -0.02725  0.05390

>

> #Annualized Benchmark Return

> benchaov <- aov(eudata$`1Y Benchmark Return (%) as of 3/31/2023 - Annualized` ~ eudata$`SFDR Classification`)

> summary(benchaov) #stat significance

                             Df Sum Sq Mean Sq F value  Pr(>F)  

eudata$`SFDR Classification`  2 0.1710 0.08552   7.304 0.00173 **

Residuals                    47 0.5503 0.01171                  

---

Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

> TukeyHSD(benchaov)

  Tukey multiple comparisons of means

    95% family-wise confidence level

Fit: aov(formula = eudata$`1Y Benchmark Return (%) as of 3/31/2023 - Annualized` ~ eudata$`SFDR Classification`)

$`eudata$`SFDR Classification``

                         diff        lwr         upr     p adj

Article 8-Article 6 -0.111270 -0.1940842 -0.02845585 0.0059156

Article 9-Article 6 -0.132475 -0.2339012 -0.03104879 0.0076117

Article 9-Article 8 -0.021205 -0.1226312  0.08022121 0.8687994

>

> boxplot(eudata$`1Y Benchmark Return (%) as of 3/31/2023 - Annualized` ~ eudata$`SFDR Classification`, data = eudata,

+         xlab = "SFDR Classification", ylab = "1Y Benchmark Return (%) as of 3/31/2023 - Annualized`",

+         main = "Mean Benchmark Return of each SFDR Classification",

+         frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

>

> aggregate(eudata$`1Y Benchmark Return (%) as of 3/31/2023 - Annualized`,list(eudata$`SFDR Classification`), FUN=mean)

    Group.1         x

1 Article 6  0.075555

2 Article 8 -0.035715

3 Article 9 -0.056920

>

> #ESG Quality Score

> esgaov <- aov(eudata$`MSCI ESG Quality Score (0-10) - 4/7/2022` ~ eudata$`SFDR Classification`)

> summary(esgaov)

                             Df Sum Sq Mean Sq F value   Pr(>F)    

eudata$`SFDR Classification`  2  26.48  13.240   8.322 0.000805 ***

Residuals                    47  74.77   1.591                    

---

Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

> TukeyHSD(esgaov)

  Tukey multiple comparisons of means

    95% family-wise confidence level

Fit: aov(formula = eudata$`MSCI ESG Quality Score (0-10) - 4/7/2022` ~ eudata$`SFDR Classification`)

$`eudata$`SFDR Classification``

                      diff         lwr      upr     p adj

Article 8-Article 6 0.8295 -0.13580646 1.794806 0.1050470

Article 9-Article 6 1.9800  0.79774587 3.162254 0.0005438

Article 9-Article 8 1.1505 -0.03175413 2.332754 0.0579925

> boxplot(eudata$`MSCI ESG Quality Score (0-10) - 4/7/2022` ~ eudata$`SFDR Classification`, data = eudata,

+         xlab = "SFDR Classification", ylab = "MSCI ESG Quality Score (0-10)",

+         main = "Mean ESG Quality Score of each SFDR Classification",

+         frame = FALSE, col = c("#bf5700", "#005f86", "#00a9b7"))

> aggregate(eudata$`MSCI ESG Quality Score (0-10) - 4/7/2022`,list(eudata$`SFDR Classification`), FUN=mean)

    Group.1      x

1 Article 6 7.3110

2 Article 8 8.1405

3 Article 9 9.2910

>

> #testing normality

> plot(benchaov, 2)

> plot(annualaov, 2)

> plot(tenaov, 2)

> plot(tempaov, 2)

Appendix H: Plots and Graphs

H1. Total Return and Carbon Intensity

H2. ESG Scores and SFDR Classification

H3. Implied Temperatures of SFDR Classifications

H4. Box Plots of Returns by SFDR Classification

Table: Return Means by Group

H5. ESG and Climate Performance by SFDR Classification

Table: Performance Means by Group

H6. Normality tests

Financial performance

ESG Performance

Appendix I: Summary Statistics

Table: Summary statistics

Appendix J: SFDR Dates for Entity Level Requirements

Source: Humpreys 2021