Sustainable and Governance Investment Funds in Brazil: A Performance Evaluation

Abstract

This study analyzes the financial performance of sustainable investments against conventional investment products in an emerging market context using a sample of sustainable and governance equity investment funds that focus solely on the Brazilian stock market. A quantitative analysis is used to compare monthly returns, volatility, and Jensen’s alpha for the period of January 2017 to December 2019 (bull market) and January 2020 and December 2021 (bear market). The study finds that sustainable investments do not diverge from conventional stock funds during a bear market period in terms of financial performance and present similar volatility. The main findings of this study corroborate the latest research from the ANBIMA reports related to sustainability practices in the Brazilian capital markets. The study also contributes to the academic literature by providing empirical evidence from an emerging economy such as Brazil of reasonable performance from sustainable investments in different periods of an economic cycle. The study has a few limitations such as the lack of a taxonomy and an ESG regulatory framework for the sustainable and governance equity investment funds and the small number of sustainable and governance equity investment funds in Brazil.

1. Introduction

The Environmental, Social, and Governance (ESG) concept first emerged in 2004 with the report “Who Cares Wins” developed by the U.N. Global Compact with the funding support of the Swiss government and the endorsement of many financial institutions around the world (e.g., World Bank Group, Goldman Sachs, and BNP Paribas). The overall goal of this framework is to develop a set of basic principles towards the business environment in order to achieve a more sustainable development and stronger and resilient financial markets [1].

The letters from this acronym stand for the non-financial dimensions that should also be considered in business and investment decisions. The first, E for environmental, deals with issues such as climate change, energy transition, and the development of a circular economy. The second, S for social, stands for how organizations manage and treat stakeholders. Some of the topics considered are diversity and inclusion among employees, fare treatment towards its supply chain, and support and funding for communities in which the organization operates [2]. The third dimension, G for governance, is mainly related to transparency, integrity, and management practices. Some of the topics considered under this theme are proper information disclosure and anti-corruption/bribery structures in place.

The sustainable finance term has also been gaining attention in the financial ecosystem. According to Boffo and Patalano [3], this can be defined as the process of considering ESG factors in investment decisions, positively contributing to the asset allocation into sustainable economic activities. In this context, many questions have been raised among academics, practitioners, and consultants. For example, considering corporations and their duties with shareholders and stakeholders towards value enhancement, a common doubt is the relationship between sustainable practices and corporate financial performance. Friede et al. [4] analyzed more than 2000 empirical studies over this topic and discovered that in 63% of them, a positive relation was identified between investing in sustainability and achieving higher equity returns. In only 8% of the studies was there a negative relationship. Some of the links between value enhancement and a strong ESG approach are cost reductions from less energy consumption and water intakes and productivity increase by boosting employee motivation and attracting top talent to the organization [5].

Regarding the investor’s perspective, it is important to explain the concept of sustainable investments (SI). It can be defined as an approach that considers ESG factors in the portfolio selection and asset allocation process. To do so, it might use different strategies such as negative screening of controversial industries, positive screening of best-in-class companies, and ESG integration into the risk management process and investment decision. It can also comprise similar terms such as ethical investing, socially responsible investing (SRI), and impact investing [6]. When analyzing the global scenario for SI, it is possible to observe a growth in this investment class. According to the Global Sustainable Investment Review [7], 35.9% of all assets under management (AUM) in the five major global markets (Europe, United States, Canada, Australasia, and Japan) are under the SI umbrella. This represents an amount of USD 35.3 trillion, an increase of 15% compared to 2018.

A common question in the investment fund industry regarding SI is related to the performance of these investment products when comparing them to conventional peers. Becchetti et al. [8] analyzed the performance of socially responsible funds (SRFs) and conventional funds between 1992–2012 with 22,000 funds being considered in the sample. The authors were not able to find any particular superiority of performance between the two investment styles throughout the period of analysis. However, during the 2008 global financial crisis, it was observed that the SRFs outperformed conventional funds. Furthermore, Yue et al. [9] performed a quantitative analysis comparing the performance and volatility of a sample of sustainable and conventional funds with 30 funds in each sample between the period of 2014 and 2018. Again, the sustainable funds sample proved to be less risky than its counterpart and there was no significant difference of financial performance between the samples.

Upon further analyzing of the literature, it is observed that most of the studies with similar objectives to the ones previously mentioned are focused on financial markets from developed economies such as US, Europe, and Japan. An exception is the paper developed by Silva and Iquiapaza [10] that analyzed the performance of traditional and sustainable funds in Brazil from 2009 to 2016 using the Jensen’s alpha as the main variable of analysis.

The objective of this paper was to perform such an analysis using also other financial performance indicators such as the return and variance. The sustainable and governance equity investment funds classification from the Brazilian Association of Capital Markets (ANBIMA) is used as a proxy for SI. A sample of these equity funds is compared to a sample of conventional funds in terms of risk (measured by the volatility of the fund portfolios) and return. Jensen’s alpha is also estimated for each sample using Carhart’s four-factor model [11]. The period considered for the analysis is between 2017 and 2021. The choice for this time window was made so that it contemplates a bull and bear market period in the Brazilian capital markets. The former stands for moments of optimism and higher demand by investors in the capital markets. In this study it is illustrated by the positive variation of Ibovespa Index, which is the main Brazilian Broad Market Index, growing from 64,671 points in January 2017 to 115,645 points in December 2019. The latter stands for moments of pessimism and uncertainty in capital markets. In this study it is illustrated by the exogenous economy shock during the years from 2020 and 2021 caused by the COVID-19 pandemic, generating a negative variation of Ibovespa Index from 113,761 points in January 2020 to 104,822 in December 2021.

From each sample of equity investment funds, two portfolios of sustainable and conventional funds are created. The monthly returns from each portfolio are compared through statistical tests for the mean (paired t-test, Wilcoxon signed rank test, unpaired t-test, and Mann–Whitney test) and the variance (Levene test) during the bull and bear market periods. Furthermore, the Jensen’s alpha is estimated for each portfolio considering the entire period so that the minimum of 60 observations is guaranteed for the linear multiple regression, which is in line with Carhart’s [11] previous work.

The findings from the analysis demonstrate that during the bear market, no statistically significant difference in financial performance is noticed between the sustainable and conventional investments. However, during the bull market period, the conventional portfolio outperformed the sustainable portfolio by a statistically significant margin. Moreover, all estimated alphas are not statistically significantly different from zero for both investment classes.

The main contributions from this study can be summarized in the following topics:

• The study provides empirical evidence on the financial performance of SI in an emerging market context, which is a valuable contribution to the literature.
• The study uses two different measures of financial performance (monthly returns and Jensen’s alpha), which allows for a more comprehensive analysis.
• The study considers both bull and bear market periods, which provides a more complete picture of the performance of SI.
• The study offers insights into the motivations of Brazilian investment managers in creating ESG investment vehicles.
• The study provides practical implications for investors interested in SI.

The remainder of this article is organized into four sections. Section 2 presents a literature review over SI, sustainable and governance investment funds in Brazil, and a summary of similar studies performed in other markets and regions. Section 3 presents the data and methodology used to perform the comparison. Section 4 presents the main findings of the study with their proper discussion. Finally, Section 5 presents the main conclusions from the research.

2. Literature Review

2.1. Sustainable Investments

Stock markets are of great importance to the economic development of a country as they reduce the friction and distance (cost transactions) among investors and companies from the real economy, allowing savings to be transformed into funds for long-term projects [12,13]. Moreover, stock markets may provide market liquidity (trading facilitation), portfolio diversification (improved risk management), and the possibility of enhancing corporate governance by aligning the interests of managers and owners [14,15,16].

Considering the rise of the sustainable development agenda in the past decades and the important role that stock markets play in the economic growth of nations, it is expected that the former should somehow reverberate into the latter. Energy transition, low-carbon economy, and social inequality are currently some of the topics of main concern for public and private organizations. Therefore, appropriate funding mechanisms are required to further promote sustainable practices in the global economy. In this context, sustainable finance products such as green bonds, microfinance, and SRFs are key to filling this gap [17,18]. However, it is important to demonstrate that investors can rely on sustainable financing products not only for ethical motivations, but also for their financial performance [19]. One particular aspect that can be exemplified is the decision-making process of investors towards their investment opportunities. In this sense, the SI trend has been gaining much attention worldwide.

The 2000s decade marked the growth of the SI concept and the ESG framework. In 2000, the UN Global Compact was launched with the goal to engage worldwide organizations into corporate social responsibility (CSR) practices. A set of universal principles was developed on how businesses should be conducted related to human rights, labor, environment, and anti-corruption practices [20]. In 2004, the UN Global Compact supervised the development of the report “Who Cares Wins” focused on the financial services industry, which coined the ESG term. The goal was to demonstrate the need of an integrated approach towards ESG issues by those organizations that affect or are affected (e.g., banks, insurance companies, and asset management firms) by the financial ecosystem [1].

In 2006, the UN Global Compact and the UNEP FI joined forces with some of the most influential worldwide institutional investors to develop a set of principles for responsible investment (PRI) that could somehow guide investment managers into how to better incorporate ESG material aspects into their investment practice. The principles created focused on the proper integration of non-financial criteria into the investment decision-making process and the proper ESG information disclosure by organizations (both financial and non-financial institutions) [21]. In 2017, PRI released an investment case demonstrating why the Sustainable Development Goals (SGDs) from the 2030 Sustainable Development Agenda should be considered in the investment management practice. Some of the relevant factors listed were as follows: (1) the macro risks from climate change and environmental destruction might represent a systemic risk to be considered by institutional investors; and (2) there are micro-opportunities to be considered from clean technology stocks, low-carbon infrastructure, green real estate, sustainable forestry, and agriculture [22].

Another important concept that became key to the continuous rise of the ESG framework has been the fiduciary duty associated mainly with institutional investors. According to Sandberg [23], fiduciary duty is the set of obligations that a trustee (e.g., a pension fund) must have towards its beneficiaries (e.g., future retirement pensioners). Since the beginning of the sustainable investments practices, there has been a debate on whether or not taking non-financial aspects such as the ones proposed by the ESG might not be in the best interest of the primary owners of the financial resources being invested through mutual funds or any other investment vehicle.

In this sense, the UNEP FI commissioned a report to the London-based law firm Freshfields Bruckhaus Deringer in order to shed light into the issue. The goal was to understand if considering ESG factors into any investment policy would be voluntarily permitted, legally required, or even hampered by law and regulation. The report which later became known as the Freshfields Report was launched in 2005. The conclusion was that considering ESG was not only in line with the fiduciary duty, but not considering these material aspects would actually be a breach of it [24]. The international acceptance of this conclusion took one more obstacle out of the way for the ESG growth [25].

In order to further strengthen the idea of corporate engagement from investors towards companies as a tool for exercising the fiduciary duty, the UK Financial Reporting Council (FRC) launched the UK Stewardship Code in 2010 for institutional investors. The goal was to develop a set of good practices for shareholders on monitoring their investee companies, acting collectively with other investors if and when necessary, reporting their stewardship practices and voting activities, and establishing clear guidelines on how and when to escalate activities to protect and enhance shareholder value [26]. According to the Brazilian Association for Capital Markets Investors (AMEC), there are already more than 10 similar codes in different countries worldwide. Moreover, the Brazilian Stewardship Code was launched in 2016 with similar principles to the initial set created in the UK [27].

The global scenario indicates that SI and ESG markets are booming. According to the Bank of International Settlements [28], the mutual funds and exchange-traded-funds (ETFs) industry alone has reached approximately US $2 trillion in AUM under the ESG umbrella. However, this amount only accounts for 3% of the current market size, indicating that there is still much room for growth. The continuation and strengthening of this trend will depend mainly on developing an international ESG taxonomy standardization and the availability of reliable ESG data [29].

2.2. Sustainable and Governance Investment Funds in Brazil

Sustainable and Governance Investment Funds are equity funds designed to invest in companies with high standards of corporate governance or that stand out among their peers when it comes to social responsibility and sustainability practices [30]. Though this classification was created in 2008 by ANBIMA, the first sustainable and governance equity fund actually appeared in Brazil in 2001 and is still active today [31]. It is the “Ethical Ações Sustentabilidade Special” managed by Santander Asset Management, which has as an investment policy of investing in companies with better ESG performance using the organization’s proprietary methodology.

Since then both the number of sustainable and governance equity funds and their AUM grew by more than 20 times from 2001 to 2021. In Figure 1 it is possible to observe the historical evolution of this data.

A phenomenon to be observed from the above graph is the recent growth of 100% of both the number of funds (from 22 in 2019 to 45 in 2021) and the AUM (from R $1.8 billion in 2019 to R $2.9 billion in 2021). There were 16 new sustainable and governance equity investment funds created in the year 2020 alone with 15 out these 16 new funds having the acronyms “ESG” or “ASG” in their names [31]. From that, it is possible to infer the rise of the impact of the ESG agenda in the Brazilian capital markets. Finally, it is important to consider that many classifications of SI are also being taken into account by the umbrella of this definition for sustainable and governance investments created by ANBIMA. These are the SRFs, Ethical Investment Funds, and ESG Investment Funds [29].

Specifically analyzing the 45 active sustainable and governance equity investment funds at the end of 2021, it is possible to observe that 69% of them are managed by asset management firms from traditional commercial banks (e.g., Itaú Unibanco S.A, Banco do Brasil S.A, and Banco Bradesco S.A). Another 29% are managed by independent asset management firms (e.g., JGP Gestão de Recursos ltda.) and 2% are managed by a non-commercial bank (Banco Plural S.A) [31]. Furthermore, the majority consider IBOVESPA (31%) or the Corporate Sustainability Index (ISE, 29%) as either a benchmark for the performance fee or as a profitability reference [31].

Another aspect to be considered is how investors and professionals involved in the Brazilian investment fund industry view this ESG agenda impact and how they are preparing themselves to cope with the challenges imposed by it. According to the report “A Sustainability Portrait of the Brazilian Capital Markets”, 84% of the 209 asset management firm’s respondents are already considering sustainability issues in their investment process [32]. Moreover, the penetration of this agenda is mainly related to risk management requirements (45%), followed by institutional decisions (30%), opportunities of higher returns (9%), fiduciary duties (6%), and, lastly, client demands (4%) [32].

Also at the beginning of 2022, ANBIMA launched its second ESG guide. Some of the conclusions presented by the report about the Brazilian context are as follows: (1) 71% of the asset management firm respondents declared to have some sort of structure (specific committee or employee) dedicated to deal with the ESG topic. In the first edition of the guide (2018) the percentage was only of 34%. (2) 80% of the asset management firms already have a responsible investment policy formalized or under development. In the first edition of the guide (2018) the percentage was about of 69%. (3) The COVID-19 pandemic increased the perception of the importance regarding ESG issues in the risk management framework of banks, asset management firms, and other financial institutions. (4) 52% of all respondents declared that sustainability issues will gain even more importance in the next 12 months [33].

Moreover, even though the ESG agenda has gained awareness from the market, the three pillars might not be valued the same way by investors in Brazil. According to Miralles-Quirós et.al [34], investors tend to value more governance and social practices from sensitive industries (e.g., energy, chemical, and steel manufacturing) and environmental practices from non-sensitive industries.

Lastly, in order to further strengthen the fight against “greenwashing”, a new regulatory framework for SI has been recently launched in Brazil by ANBIMA named “Rules and Procedures for Identification of Sustainable Investment Funds (IS)”. One of the novelties provided by it is the understanding that a sustainable investment product must be related not only to the stocks from the portfolio of the equity funds, but also to the management practices within the organizations responsible for them [35]. Firms will have to provide transparency on how they train their teams towards sustainability issues and how they incorporate ESG topics into their standard investment analysis process for all their investment products. Additionally, a distinction was created between equity funds that should be characterized only as “considering ESG issues” and those that are to be considered as a sustainable investment fund or “IS Fund”. The former must belong to an organization with a transparent and formalized policy of ESG integration into its risk management analysis. The latter, aside from the basic requirements previously mentioned, must also have an investment thesis specifically focused on a sustainability issue from one or more of the three ESG pillars [35].

2.3. Sustainable Investments vis-à-vis Conventional Investments

Since the inception of the SRI trend in the 1970s and more recently of SI and the ESG framework, there has been a great concern regarding the financial performance of such an investment strategy. An illustration of this debate is the article written by Milton Friedman [36] in the New York Times entitled “A Friedman doctrine: The Social Responsibility of Business Is to Increase Its Profits”. The famous economist explains that a corporation is an artificial institution and, therefore, should not have any responsibility aside from maximizing shareholder value by increasing its profits over time. If that would be the case, CSR practices would not have influence on the stock performance of companies, and by consequence, neither on socially responsible portfolios.

On the other hand, there is another line of thought which states that by enhancing their social and environmental performance, companies can obtain new market opportunities and increase their efficiency, hence increasing their value in the long-term [37,38]. Moreover, investors can further strengthen their risk management process and portfolio allocation by analyzing a more complete set of information from private organizations, enhancing their chances to predict financial distress from companies and protecting themselves from exogenous economic shocks [39,40].

According to Becchetti et al. [8], there are there main differences attributed to SI against conventional investments. The first is related to additional CSR information required from investable companies. Even though nowadays many organizations have a sustainability report consolidating non-financial material aspects, fund managers still need specialized resources to analyze this data and add it to screening policies or to a risk management process.

The second difference is related to the restricted available universe for portfolio diversification. Markowitz [41] in his portfolio selection theory proposed that investors should strive to maximize the expected return and minimize the variance of their portfolio by applying a diversification process from a universal set of companies, creating an efficient frontier of possible combinations between return and risk as measured by the variance of returns. This would be generated by a combination of different weights of the stocks in the universal set contained in the portfolio. In this sense, by creating a constraint into this universe of selectable stocks due to positive or negative stock screening, it is possible that the resulting portfolio will have a smaller return with the same expected variance when comparing it with a conventional portfolio [42].

Finally, the third difference is related to timing costs that emerge from the necessity of SRI fund managers to comply with rules related to non-financial matters. A company might change its behavior and lose its CSR performance, forcing the portfolio manager to take the stock from his or her portfolio even though its financial performance might be better than counterparts at that moment [43]. These three constraints mentioned above have generated a lot of skepticism regarding SI and its capacity to generate value for investors in the long-term.

Results from past empirical studies on this matter are controversial and mixed, but based on the summary from

Table 1. Summary of previous studies comparing sustainable and conventional investment funds performance.

Authors	Year Published	Period of Analysis	Data Frequency	Method	Period of Crisis?	Performance Variable Compared between SI (SRI or ESG) and Conventional Investments
						Return	Risk	Alpha
Pisani & Russo [44]	2021	2020	Daily	Regression	Yes	SI dominance ***	SI less risky ***	-
Arefeen & Shimada 1 [45]	2020	2016	Daily	Regression	Yes	No dominance	SI less risky ***	-
Broadstock et al. [46]	2020	2020	Daily	Regression	Yes	SI dominance **	SI less risky **	-
Yue et al. [9]	2020	2014 to 2018	Daily	Regression	No	No dominance	No dominance	SI dominance *
Durán-Santomil et al. 2 [47]	2019	2016 to 2018	Annual	Regression	No	SI dominance *	SI less risky **	SI dominance **
Silva & Iquiapaza 3 [10]	2017	2009 to 2016	Monthly	Regression	Yes	-	-	No dominance
Becchetti et al. 4 [8]	2014	1992 to 2012	Monthly	Regression	Yes	No dominance	-	No dominance (non-crisis) SI dominance (crisis) ***
Nofsinger & Varma [48]	2014	2000 to 2011	Annual	Regression	Yes	No dominance	-	Conventional dominance (non-crisis) * SI dominance (crisis) *
Barnett & Salomon 5 [49]	2006	1972 to 2000	Monthly	Regression	No	No dominance	-	-
Bello [50]	2005	1994 to 2001	Monthly	Regression	No	-	-	No dominance
Bauer et al. [51]	2004	1990 to 2001	Annual	Regression	No	No dominance	-	SI dominance *
Guerard [52]	1997	1987 to 1994	Monthly	Hypothesis Test	No	No dominance	-	-

Notes: statistical significance levels: (***) denotes 0.01, (**) denotes 0.05, and (*) denotes 0.1. ¹—The study actually analyzes the abnormal returns from the period of economic shock. ²—The study uses the value at risk (VaR) variable to measure the risk. ³—The authors also performed the analysis considering the sharpe index as a dependent variable. SRI funds presented a better relation between return and risk. ⁴—The Jensen’s alpha was also calculated for the specific period of the 2007 global financial crisis. ⁵—The authors considered the risk-adjusted performance (RAP) as the financial performance measure.

it is possible to observe a few trends. Firstly, it seems that SI funds tend to present less volatility over time. Moreover, this characteristic might be even more exacerbated in periods of crisis from exogenous economic shocks. Secondly, even though SI funds might suffer from constraints in the diversification process and additional costs related to the analysis of extra non-financial data, it is not clear that conventional funds present a better financial performance both in terms of return and alpha.

What is actually observed in some cases is that in periods of economic crisis, SI might have a financial dominance (return and alpha) over conventional investments. Moreover, only in Nofsinger and Varma’s [48] study, is a dominance observed of conventional investments in non-crisis periods. In all others, no dominance of one investment style over the other is observed.

Considering this scenario, it is important to empirically verify if these same phenomena can be observed in an emerging market context, as illustrated by the sustainable and governance equity investment funds in Brazil. The following hypotheses emerge from the literature review and will guide the analysis:

Hypothesis 1.

SI stock funds do not diverge from conventional stock funds in terms of financial performance.

Hypothesis 1.1.

Sustainable investment stock funds do not diverge from conventional stock funds in terms of returns.

Hypothesis 1.2.

Sustainable investment stock funds do not diverge from conventional stock funds in terms of Jensen’s alpha.

Hypothesis 2.

SI stock funds present less volatility compared to conventional stock funds, especially in moments of exogenous economic shock.

3. Data and Methods

To test the hypotheses that arose in the last subsection, two portfolios (sustainable and conventional) of equity investment funds are developed and compared regarding their historical returns and volatility. Furthermore, the alphas of each portfolio are estimated and compared using the four-factor model of Carhart [11]. In order to address different scenarios of market return, the alpha estimations are made using both IBOVESPA, which is the main Brazilian Broad Market Index, and ISE, which is the most well-known sustainable stock market index in Brazil.

Finally, the analysis is developed considering the timeframe from January 2017 to December 2021. The choice for this period was made so that it contemplates both a bull market period and a bear market period due to an economic shock. The former is illustrated by the positive variation of Ibovespa Index from 64,671 points in January 2017 to 115,645 points in December 2019. The second part of the study considers the years from 2020 and 2021 marked by the economic shock caused by the COVID-19 pandemic, which began to gain global public awareness by the World Health Organization (WHO) in January 2020 (WHO COVID-19 timeline: https://www.who.int/news/item/27-04-2020-who-timeline---COVID-19 (accessed on 14 May 2023)). The event generated breakdowns in many industries due to shortages in their supply chains, decrease of economic activity worldwide, and volatility in the capital markets due to uncertainties related to the perspective of how the disease would develop in the long-term. During this period, a negative variation of the Ibovespa Index was observed from 113,761 points in January 2020 to 104,822 in December 2021.

3.1. Data

Data was obtained from the following sources to perform the study: the monthly returns of the portfolios sorted by size, book-to-market and momentum, and the risk-free rate, which is required to build the risk factors for the Carhart four-factormodel [11]. They were downloaded from the Center for Research in Financial Economics of the University of São Paulo (NEFIN). The data with monthly returns from ISE and Ibovespa was downloaded from the B3 website.

The data from the Brazilian equity investment funds were obtained from the Economatica platform that consolidates data from the financial services industry in Brazil and Latin America. A database with all active and non-active equity investment funds was downloaded. At first, 6153 funds were mapped. After building the SI and the conventional fund portfolios, specific data was obtained for each fund. The following data from January 2017 to December 2021 were obtained from the funds selected to compose each sample (SI and conventional):

• Total Assets (TA_i,t): historical monthly data series of total assets for each selected fund (last trading session of the month), adjusted by dividend distribution.
• Beta (βi): the beta (market or systemic risk) of each fund considering IBOVESPA as the broad market index. A 5 year window was considered for the estimation with the last trading session of December 2021 being the reference point.
• Monthly Return (R_i,t): historical data series with the monthly return for each selected fund calculated by the Economatica platform as:

$$Ri,t=\frac{T{A}_{i,t} - T{A}_{i,t-1}}{T{A}_{i,t-1}}$$

(1)

3.2. Methodology

3.2.1. Building Sustainable and Conventional Equity Investment Fund Portfolios

The development process for the SI and conventional equity fund portfolios can be summarized in two steps: building two samples of stock investment funds and calculating the average of the fund returns for each sample. Following Becchetti et al. [8] and Yue [9], equally weighted portfolios should be developed. Moreover, two more portfolios should also be created with a weighted average of the fund returns. As the weighting criteria, the amount of total assets of each fund at the time i is used to estimate the portfolio return at that same period. For each set of portfolios, the hypothesis for the difference of returns between portfolios and the volatility of each one will be tested. Finally, the Jensen’s alpha will also be estimated and compared.

The SI equity funds sample was obtained from the initial database downloaded from Economatica. A filter was applied to identify all the active and non-active sustainable and governance investment funds and a total of 71 funds were identified. The second step was to remove all the sustainable and governance equity funds that did not have data for one of the two economic cycles contemplated in the study. All the funds created after January 2020 or that were canceled by December 2019 were excluded from the sample. The final SI sample contemplated 22 sustainable and governance equity funds.

The conventional equity funds sample was built based on a risk factor exposure matching style procedure, similar to the one executed by Becchetti et al. [8]. Considering that there are many more conventional equity investment funds available to construct the conventional portfolio, it is important that the selection process for this sample match the same risk profile as the SI funds. Therefore, for each fund from the SI portfolio, two conventional equity funds were selected in an attempt to minimize the beta-distance between sustainable (fund i) and conventional (fund j) funds, as shown below:

$$d_{\beta (i,j)}={\sum }_{k=1}^k\left|\beta ik-\beta jk\right|$$

(2)

where βi stands for the beta (market or systemic risk) of each SI fund and βj stands for the beta of each conventional fund. The criteria of constrains regarding the data availability for the period, similar to the one used in the SI portfolio, was also used for the conventional portfolio. Finally, investment funds with a majority of their assets invested in foreign companies and exchange–traded funds were also not considered for the sample development process. The final conventional sample contemplated 44 equity funds.

3.2.2. Statistical Tests for the Mean and the Variance

The next step after having the historical series of monthly returns for each set of portfolios (SI and conventional) was to test the hypotheses established. The first test was to check if there were any differences in the financial performance illustrated by Hypothesis 1.1. To do so, the difference between the returns of each portfolio was calculate as shown below:

Diff_(SI-CONV.),t = R_SI,t − R_conv,t

(3)

where R_SI stands for the returns from the SI portfolio and R_conv stands for returns from the conventional portfolio. A paired test for dependent samples was used to test the following null hypothesis considering both the timeframe of January 2017 to December 2019 (bull market) and January 2020 to December 2021 (bear market):

$${\mathrm{H}}_{0\left(1.1\right)}:\frac{{\sum }_{\mathrm{t}=1}^{\mathrm{t}}\mathrm{Diff}\left(\mathrm{SI}-\mathrm{CONV}.\right),\mathrm{t}}{\mathrm{t}}=0$$

(4)

$${\mathrm{H}}_{1\left(1.1\right)}:\frac{{\sum }_{\mathrm{t}=1}^{\mathrm{t}}\mathrm{Diff}\left(\mathrm{SI}-\mathrm{CONV}.\right),\mathrm{t}}{\mathrm{t}}\ne 0$$

(5)

Moreover, the average returns of each window of analysis (bear and bull markets) were compared using an unpaired test for independent samples. Due to the small number of funds contained at each portfolio, the tests were performed using both a parametric and a non-parametric method, increasing the robustness of the analysis against previous work in the literature.

The paired test for dependent samples performed were the t-test for dependent samples for the parametric assumption and Wilcoxon signed-rank test for the non-parametric assumption. The unpaired tests for independent samples performed were the t-test for independent samples for the parametric assumption and the Mann–Whitney test for the non-parametric assumption.

To test hypothesis 2, the variances from the average of monthly returns of each portfolio (SI and conventional) were compared. The comparison was made considering the period of bull market (January 17 to December 19) and bear market (January 2020/December 2020). Moreover, to verify possible differences of volatility between market periods, a comparison between the variances from the periods of bull and bear market was also developed for each portfolio. The sample variances were estimated as presented in the equation below:

$${\mathsf{\sigma }}^2=\frac{{\sum }_{\mathrm{t}=1}^{\mathrm{t}}{\left(\mathrm{Rt}-\overline{\mathrm{R}}\right)}^2}{\mathrm{t}-1}$$

(6)

where Rt stands for each monthly return observation for the SI and conventional portfolios and $\overline{\mathrm{R}}$ stands for the average of monthly returns also for each portfolio. Finally, in order to compare the variance from the portfolios and between the two windows of analysis, the Levene test was performed.

3.2.3. Carhart Four Factor Model

To test the hypothesis established at H (1b), the alphas of each portfolio needed to be estimated for the period. Considering the robustness of the Carhart model, this was the model chosen to perform the alpha estimation. A multiple linear regression was performed for each portfolio using the ordinary least square (OLS) method. To ensure the robustness and validity of the coefficients estimated, the residuals normality and independency were tested, respectively, using the Shapiro test and Durbin–Watson test. Moreover, the Breusch–Pagan test was performed to check for error variance homogeneity (homoscedasticity). The variance inflation factor (VIF) test was used to check for multicollinearity among the factors of the models. The models were all estimated using the R programming language within the R Studio software version 4.0.4 (All R scripts and databases used for the analysis are available for disclosure upon request). To build each risk factor considered in the model, the NEFIN database and methodology were used as shown below:

R_i − R_f = α_i + β₁MKTPR + β₂SMB + β₃HML + β₄WML + ε₁

(7)

where:

R_i—Equally weighted portfolio monthly returns based on data obtained from the Economatica platform for each fund composing the SI and conventional portfolios.

R_f—Risk-free rate represented by the monthly returns of the 30-day DI Swap.

α_i—Jensen’s alpha represented the portion not explained by the model, which was estimated.

MKTPR—Market risk factor as calculated by the difference of monthly returns between a market portfolio (IBOVESPA or ISE) and the risk-free rate mentioned above.

SMB—Size premium, which was estimated from the portfolios sorted by size (small and big) obtained from the NEFIN website. Stocks are sorted by market capitalization and divided into terciles. The small portfolio contains stocks from the 1st tercile and the big portfolio contains stocks from the 3rd tercile. Finally, the SMB factor was calculated by the difference of monthly returns from the small portfolio minus the big portfolio.

HML—The value premium was estimated from the portfolios sorted by book-to-market returns obtained from the NEFIN website. Stocks were sorted by book-to-market ratio and divided in terciles. The growth portfolio contains stocks from the 1st tercile and the value portfolio contains stocks from the 3rd tercile. Finally, the HML factor was calculated based on the difference of monthly returns from the value portfolio minus the growth portfolio.

WML—The momentum premium was estimated from the portfolios sorted by momentum returns obtained from the NEFIN website. Stocks were sorted by their cumulative returns between t-12 and t-2 and were divided into terciles. The loser portfolio contains stocks from the 1st tercile and the winner portfolio contains stocks from the 3rd tercile. Finally, the WML factor was calculated based on the difference of monthly returns from the winner portfolio minus the loser portfolio.

ε₁—Error term from model estimation.

The alphas for both sets of portfolios (equally weighted and weighted by equity fund’s total assets) considering both IBOVESPA and ISE as the market portfolio were compared in order to analyze if there are any statistically significant differences among them.

4. Results and Discussion

4.1. Statistical Tests for the Mean and the Variance

The historical series of returns from the sustainable and conventional portfolios were compared in terms of returns and variance.

Table 2. Shapiro test: normality check for paired t-test diff (SI-CONV).

Market ¹	Portfolio Weighting Criteria	Frequency 2	Shapiro Test (Normality Check)
			W-Statistics	p-Value
Bull Market	Equally Weighted	36	0.959	0.112
Bull Market	Total Assets Weighted	36	0.959	0.207
Bear Market	Equally Weighted	24	0.980	0.861
Bear Market	Total Assets Weighted	24	0.946	0.218

¹—Bull market: 17 January = 19 December and bear market: 20 January–21 December. ²—Monthly observation.

and

Table 3. Paired t-test diff (SI-CONV).

Market ¹	Portfolio Weighting Criteria	Frequency 2	Paired t-Test Diff (SI-CONV)
			T-Statistics	p-Value	Difference in Means (95% Confidence Interval)		Mean of the Differences
Bull Market	Equally Weighted	36	−2.465	0.019	−0.64%	−0.06%	−0.35%
Bull Market	Total Assets Weighted	36	−2.097	0.043	−0.53%	−0.01%	−0.27%
Bear Market	Equally Weighted	24	−1.239	0.228	−0.75%	0.19%	−0.28%
Bear Market	Total Assets Weighted	24	−1.377	0.188	−0.61%	0.12%	−0.25%

¹—Bull market: 17 January–19 December and bear market: 20 January–21 December. ²—Monthly observations.

present, respectively, the Shapiro test and the paired t-test. The former aims at verifying the normality assumption required to use a parametric method, while the latter estimates possible statistically significant differences between the returns of each investment style for different market periods.

The results from Table 2 indicate that we failed to reject the null hypothesis established by the Shapiro test (p > 0.05) so the normality assumption required for performing the paired t-test is confirmed for all groupings of market period and portfolio weighting criteria.

The results from Table 3 indicate that we should reject the null hypothesis (p < 0.05) for the bull market period for both equally weighted and total assets weighted portfolios. However, we fail to reject the null hypothesis (p > 0.05) for the bear market period for both portfolios with different weighting criteria. In other words, it was only possible to identify statistically significant differences in the returns between the sustainable and conventional portfolios in moments of greater attractiveness of the Brazilian capital markets. During the COVID-19 crisis, the sustainable portfolio had no statistically significant difference in financial performance against the conventional portfolio.

Table 4. Wilcoxon signed rank test diff (SI-CONV).

Market ¹	Portfolio Weighting Criteria	Frequency 2	Wilcoxon Signed Rank Test Diff (SI-CONV)		Median of Diff (SI-CONV)
			V-Statistics	p-Value	Median	IQR
Bull Market	Equally Weighted	36	199	0.035	−0.10%	1.00%
Bull Market	Total Assets Weighted	36	197	0.032	−0.40%	1.20%
Bear Market	Equally Weighted	24	111	0.277	−0.20%	1.30%
Bear Market	Total Assets Weighted	24	106	0.218	0.00%	0.90%

¹—Bull market: 17 January–19 December and bear market: 20 January–21 December. ²—Monthly observations.

presents the results for a similar comparison of statistically significant differences between the returns of each investment style, but with a non-parametric assumption. The Wilcoxon signed rank test indicates a similar conclusion to the paired t-test. The null hypothesis should be rejected (p < 0.05) for the bull market period, but not for the bear market period.

The initial results are similar to the findings from Arefeen and Shimada [45], which found no statistically significant superiority of abnormal returns in periods of economic shock from one investment style over the other. Furthermore, the results are partially similar to Becchetti et al. [8] and Nofsinger and Varma [48]. These last two studies found no dominance between the two investment styles when considering different periods of an economic cycle (bull and bear market).

These findings might partially corroborate the line of thought which states that additional costs such as constraints at portfolio diversification may arise in a sustainable investment portfolio. However, it is important to also consider the bigger downside that might come from a conventional investment portfolio. Sustainable equity investment funds might be less exposed to an economic crisis compared to their conventional peers.

Table 5. Levene test for the variance—conventional and sustainable portfolios.

Market ¹	Portfolio Weighting Criteria	Frequency 2	Levene Test
			F-Value	p-Value
Bull Market	Equally Weighted	36	0.000	0.999
Bull Market	Total Assets Weighted	36	0.013	0.909
Bear Market	Equally Weighted	24	0.032	0.858
Bear Market	Total Assets Weighted	24	0.001	0.978

¹—Bull market: 17 January–19 December and bear market: 20 January–21 December. ²—Monthly observations.

and

Table 6. Levene test for the variance—bear and bull markets.

Portfolio	Portfolio Weighting Criteria	Frequency ¹	Levene Test
			F-Value	p-Value
Conventional	Equally Weighted	36 + 24	5.923	0.018
Conventional	Total Assets Weighted	36 + 24	5.881	0.018
Sustainable	Equally Weighted	36 + 24	5.646	0.023
Sustainable	Total Assets Weighted	36 + 24	6.934	0.011

¹—Contemplates bull market (17 January–19 December) and bear market period (20 January–21 December).

present the results from the Levene test comparing the variance between each investment style (sustainable and conventional) and between different market periods (bull and bear Market). From the former it is possible to conclude that we fail to reject the null hypothesis (p > 0.05) that the variances between each investment style are similar. However, from the latter it is possible to conclude that we should reject the null hypothesis (p < 0.05) that the variances between different market periods are similar.

The results from Table 5 differ from previous findings made by Pisani and Russo [44], Arefeen and Shimada [45], and Broadstock et al. [46], which found that SI were less risky than their traditional peers. However, this could be partially explained by the risk factor exposure matching procedure used to build the conventional portfolio even with the beta distance considered in the methodology being based on the entire period (17 January–21 December). On the other hand, the variance analysis compared the variance between the portfolios considering each market period separately. Finally, the results from Table 6 indicate that both investment styles suffered from a higher volatility during the COVID-19 crisis.

Table 7. Unpaired t-test—diff (bear and bull markets).

Portfolio	Portfolio Weighting Criteria	Frequency ¹	Unpaired t-Test for Distinct Variances
			T-Statistics	p-Value	Mean Bear Market	Mean Bull Market	Diff Mean (Bear—Bull)
Conventional	Equally Weighted	36 + 24	−1.114	0.274	0.05%	2.09%	−2.04%
Conventional	Total Assets Weighted	36 + 24	−1.002	0.324	0.22%	2.09%	−1.86%
Sustainable	Equally Weighted	36 + 24	−1.064	0.296	−0.23%	1.74%	−1.98%
Sustainable	Total Assets Weighted	36 + 24	−0.967	0.341	−0.02%	1.80%	−1.83%

¹—Contemplates bull market (17 January–19 December) and bear market period (20 January–21 December).

and

Table 8. Mann–Whitney test—diff (bear and bull markets).

Portfolio	Portfolio Weighting Criteria	Frequency ¹	Mann-Whitney Test		Median of Returns of Bear and Bull Market
			W-Statistics	p-Value	Median—Bear Market	Median—Bull Market	Diff Median (Bear—Bull)
Conventional	Equally Weighted	36 + 24	366	0.326	−0.10%	1.60%	−1.70%
Conventional	Total Assets Weighted	36 + 24	365	0.318	−0.20%	1.80%	−2.00%
Sustainable	Equally Weighted	36 + 24	365	0.318	−0.30%	1.50%	−1.80%
Sustainable	Total Assets Weighted	36 + 24	366	0.326	−0.30%	1.60%	−1.90%

¹—Contemplates bull market (17 January–19 December) and bear market period (20 January–21 December).

present, respectively, the results from the unpaired t-test (parametric assumption) and Mann–Whitney test (non-parametric assumption). The analysis performed compares the historical series of returns between bull and bear market for both the conventional and sustainable portfolios.

Both tests presented similar conclusions. We fail to reject the null hypothesis (p > 0.05) that the mean or the median of returns between the bull and bear markets are similar. However, it is possible to observe that the difference between the means (Table 7) is bigger for the conventional investment portfolio and the difference between the medians (Table 8) is bigger for the sustainable investment portfolio.

This phenomenon can be explained by the fact that the historical series of returns (17 January–21 December) for each portfolio have a skewed distribution. All four portfolios have differences between the distribution mean and median mainly from outlier negative results due the COVID-19 crisis (e.g., returns from March 2020 due to the COVID-19 pandemic strike in Brazil). This is illustrated by Figure 2, which shows the unbalanced bell curve from the histograms for each portfolio created for the analysis.

4.2. Carhart Model

Table 9. Validity tests for Carhart model.

Validity Tests		Conventional Portfolio—IBOV	Conventional Portfolio—ISE	Sustainable Portfolio—IBOV	Sustainable Portfolio—ISE
Residuals Normality (Shapiro-test)	W-statistics	0.968	0.988	0.993	0.987
	p-value	0.118	0.805	0.974	0.753
Residuals Independency (Durbin-Watson test)	D-W Statistic	1.577	1.747	1.826	1.913
	p-value	0.086	0.366	0.464	0.692
Homoscedasticity check (Breusch-Pagan test)	BP	5.267	3.256	2.480	2.727
	DF	4	4	4	4
	p-value	0.261	0.516	0.648	0.605
Multicollinearity check (VIF test)	MKTPR	1.663	1.532	1.633	1.532
	SMB	1.633	1.703	1.633	1.703
	HML	1.334	1.263	1.334	1.263
	WML	1.019	1.019	1.019	1.019

and

Table 10. Coefficient results for Carhart model.

Regression Results	Conventional Portfolio—IBOV	Conventional Portfolio—ISE	Sustainable Portfolio—IBOV	Sustainable Portfolio—ISE
Dependent Variable: Ri—Rf
Independent variables
Intercept (Jensen’s alpha)	0.001	0.001	−0.001	−0.001
	(0.001)	(0.002)	(0.002)	(0.001)
MKTPR	0.850 ***	0.848 ***	0.874 ***	0.930 ***
	(0.020)	(0.040)	(0.034)	(0.024)
SMB	0.152 ***	0.129 **	0.121 *	0.055 *
	(0.028)	(0.056)	(0.048)	(0.033)
HML	−0.059 *	0.157 ***	−0.133 *	0.083 **
	(0.031)	(0.057)	(0.051)	(0.033)
WML	0.066 **	0.062	0.027	0.022
	(0.027)	(0.052)	(0.044)	(0.030)
N. Obs	60	60	60	60
Adjusted R-squared	0.9822	0.9344	0.9506	0.9779
F-statistic	815.7	210.9	284.7	654.8
Prob F-statistics	0.0000	0.0000	0.0000	0.0000

Notes: Significance levels: (***) denotes 0.01, (**) denotes 0.05, and (*) denotes 0.1. Standard errors in parentheses.

present respectively the results from the model validity tests and the coefficient results for the Carhart model of each portfolio considered in the study. The former is required to verify the robustness of the regression performed considering the risk factors and portfolio returns used for the analysis. The latter presents the risk factor coefficients and the intercept coefficient results, which represents the Jensen’s alpha for each portfolio.

From the validity tests it is possible to state that we fail to reject the null hypothesis (p > 0.05) for the Shapiro test, Durbin–Watson test, and Breusch–Pagan test. For all models estimated, the residuals present a normal distribution and are independent and homoscedastic. Moreover, all VIF tests present a result lower than ten for all coefficients estimated. In line with the literature, it is possible to state that there is no multicollinearity among the regressors of all models.

The intercept coefficient results from Table 10 indicate that for all four portfolios the estimated alphas were not statistically significant. In other words, it was not possible to identify any dominance of one investment style over the other in terms of Jensen’s alpha. This finding is in line with the results from Bello et al. [50], and more recently Silva and Iquiapaza [10]. The latter developed an analysis also considering sustainable and governance investment funds as a proxy for SI in the Brazilian market. However, the results differ from the findings of Yue et al. [9], Durán-Santomil et al. [47], and Bauer et al. [51], which identified a dominance for SI.

Regarding the market risk factor coefficients, it is possible to observe that the sustainable portfolios tend to be less influenced by the SMB factor compared to the conventional portfolio. Possibly the sustainable and governance equity funds have a lower degree of exposure to small caps compared to their conventional counterparts. Such a conclusion would be in line with the findings from Boffo and Patalano [3], which stated that large, capitalized companies tend to have more ESG ratings and coverage than small caps. Finally, the WML factor presented no statistically significant results for both sustainable portfolios and for the conventional portfolio with ISE as the market index. It seems that the sustainable and governance equity investment funds in Brazil present less exposure to momentum strategies based on past stock performance.

5. Conclusions

This study analyzes the financial performance of SI against conventional investment products in an emerging market context. A sample of sustainable and governance equity investment funds (ANBIMA classification) focused solely on the Brazilian stock market is used as a proxy for SI. Two main hypotheses arose from the literature review and guided the analysis. Firstly, SI stock funds do not diverge from conventional stock funds in terms of financial performance (monthly returns and Jensen’s alpha). Secondly, SI stock funds present less volatility compared to conventional stock funds, especially in moments of exogenous economic shock.

The proposed methodology provides a novelty to the academic literature by using two different approaches to how the portfolios were developed (equally weighted and total assets weighted). Moreover, the statistical tests performed considered both a parametric (paired t-test and unpaired t-test) and a non-parametric assumption (Wilcoxon signed rank test and Mann–Whitney test).

The paired t-test and Wilcoxon signed rank test with the monthly returns of each portfolio indicated that hypothesis H(1) can be only partially confirmed. During the bull market period, the conventional investments portfolio outperformed the SI portfolio, but during the bear market period no statistically significant difference was observed between the average returns of both investment styles. Moreover, the unpaired t-test and the Mann–Whitney test demonstrate that all four portfolios have similar average return performances between the bear and bull markets.

The Jensen’s alpha estimation also demonstrated that there is no statistically significant difference among portfolios. All regression models presented an alpha not statistically different from zero. In other words, neither investment styles outperformed the market (Ibovespa or ISE).

After performing the Levene test for the variance, it is possible to state that hypothesis H(2) cannot be confirmed from the study. All SI portfolios had no statistically significant differences of volatility against the conventional investment portfolios. Furthermore, this result stands for all market periods considered in the analysis. It is also possible to observe that all portfolios had statistically significant differences of volatility between bull and bear markets, indicating that the COVID-19 pandemic did impact both sustainable and conventional investments.

The main findings of this study corroborate the latest research from the ANBIMA reports related to sustainability practices in the Brazilian capital markets (a sustainability portrait of the Brazilian capital markets). Only 9% of asset management firm respondents declared that the penetration of the sustainability into the Brazilian capital markets is related to opportunities of higher returns. Whereas 45% of asset management firm respondents declared that risk management requirements are the main driver for the ESG integration into their investment process. Additionally, different results for statistical tests for the mean of monthly returns are observed in periods of bull and bear markets. This might indicate that Brazilian investment managers, when creating ESG investment vehicles, are more concerned about protecting investors from a downside in moments of economic shocks than to provide higher returns for them.

In terms of managerial and practical contributions, the small performance differences obtained in the paired t-test (−0.35% for the bull market and −0.25% for the bear market) and in the Wilcoxon signed rank test (−0.4% for the bull market and 0.00 for the bear market) might represent an incentive for investors with moral and ethical motivations. It shows that even in a stock market from an emerging economy it is possible to obtain reasonable returns with little downside in periods of crisis with SI vehicles.

This study also contributes to the academic literature by providing empirical evidence from an emerging economy such as Brazil of reasonable performance from SI. It innovates by using two different measures for financial performance: fund returns month by month while contemplating a period of bull and bear market and Jensen’s alpha estimated for the entire period. Previous similar studies such as Silva and Iquiapaza [10] used a fund-by-fund approach, whereas this study provides an aggregated overview by developing four fund portfolios.

6. Limitations and Future Studies

A few limitations of the study should also be addressed. Firstly, the lack of a taxonomy and ESG regulatory framework for the sustainable and governance equity investment funds may be facilitating the practice of greenwashing, which is not addressed in this study. Secondly, the sample size of sustainable and governance equity investment funds in Brazil is relatively small, which limits the generalizability of the findings. Moreover, the study only considers a 5 year period, which may not be sufficient to draw definitive conclusions about the long-term financial performance of SI. Thirdly, the study does not consider the impact of transaction costs or management fees on the financial performance of the funds, which may affect the conclusions. Finally, the analysis only considers funds focused solely on the Brazilian stock market, which may not be representative of SI in other markets or asset classes.

A suggestion for future studies would be that similar analyses could be performed in other emerging market economies while also considering other periods and bigger windows of analysis. Furthermore, different approaches for selecting the conventional equity investment funds sample could be used, aside from the risk factor exposure matching style procedure used in this study.