Stockholm Stock Exchange and Environmental 
Rating – A Multifactor Analysis 
 
 
Carl Helldén & Julia Lamers 
 
 
 
 
 
 
 
 
 
Supervisor: Taylan Mavruk 
 
Master’s thesis in Finance, 30 hec 
Spring 2022 
Graduate School  
 
School of Business, Economics and Law at the University of Gothenburg 
Acknowledgements
To begin with, we would like to express our gratitude to Taylan Mavruk, our supervisor
throughout this master thesis. The support and time he has put into our thesis have been
invaluable and great guidance to the success. We would also like to thank the School
of Business, Economics and Law at University of Gothenburg for two insightful years
of studies.
Abstract
The thesis investigates if investors can generate positive abnormal performance by in-
vesting in Environmental high-rated stocks on the Stockholm stock exchange based
on three screening strategies; positive, negative and best-in-class for value-weighted,
long-only and long-short portfolios. The sample is between 2010-2020, using CAPM,
Fama-French three factor model and Carhart four factor model. The results show that
the long-only portfolios with the positive and negative screening strategies generate
positive and significant results, where the negative generates the strongest result with
a monthly return of 0.0156% with Carhart. The best-in-class screening strategy gen-
erates mixed and inconclusive results for all portfolios. The thesis concludes that the
result is mixed and that investors do wisely by investing in long-only portfolios using
the positive and-or negative screening strategies.
Keywords: ESG, Environmental, asset pricing models, screening strategies
Contents
1 Introduction 1
2 Literature Review and Theory 4
2.1 Theoretical Background . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.1 Efficient Market Hypothesis . . . . . . . . . . . . . . . . . . . 4
2.1.2 Shareholder Theory: A Friedman Doctrine . . . . . . . . . . . 5
2.1.3 Stakeholder Theory . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Previous Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2.1 ESG Research . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2.2 Environmental Research . . . . . . . . . . . . . . . . . . . . . 9
3 Data and Methodology 11
3.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1.1 Sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1.2 Environmental Rating . . . . . . . . . . . . . . . . . . . . . . 12
3.1.3 Data Management . . . . . . . . . . . . . . . . . . . . . . . . 13
3.1.4 Portfolio Construction . . . . . . . . . . . . . . . . . . . . . . 13
3.1.5 Re-balancing . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1.6 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2.1 Capital Asset Pricing Model . . . . . . . . . . . . . . . . . . . 16
3.2.2 Fama-French Three Factor Model . . . . . . . . . . . . . . . . 16
3.2.3 Carhart Four Factor Model . . . . . . . . . . . . . . . . . . . . 17
3.2.4 Portfolio Formation and Screening . . . . . . . . . . . . . . . . 17
3.2.5 GICS Industry and Classification . . . . . . . . . . . . . . . . 19
4 Results 21
4.1 Screening Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2 Alternative Portfolio Construction . . . . . . . . . . . . . . . . . . . . 24
4.2.1 Sub-Periods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.2.2 Equally-Weighted Portfolios . . . . . . . . . . . . . . . . . . . 27
5 Analysis 29
5.1 Screening Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.2 Alternative Portfolio Construction . . . . . . . . . . . . . . . . . . . . 31
5.2.1 Sub-Periods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
i
5.2.2 Equally-Weighted Portfolios . . . . . . . . . . . . . . . . . . . 32
6 Conclusion 34
A Appendix 39
A.1 Screening Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
A.2 Total Return of GICS Sectors . . . . . . . . . . . . . . . . . . . . . . . 47
A.3 Score Grade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
ii
1 Introduction
The objective of the thesis is to offer a guide for investors (professionals and amateurs)
on how to invest in Environmental rated stocks based on screening strategies, and if the
screening strategies generate positive abnormal alphas. The field of sustainable finance
is relevant today for researchers, asset managers, and investors because of the increased
awareness during the last decade regarding the environmental, social and governmental
questions. It has led to the creation of the acronym ESG, which stands for Environmen-
tal, Social, and Government with the purpose of simplifying the effects a company’s
operations have on these issues (Who Cares Wins n.d.). The increased interest has led
to numerous ratings of ESG scores created by institutions such as Bloomberg, Sus-
tainalytics, and Thomson Reuters Refinitiv Eikon DataStream (Refinitiv Eikon). The
financial market often highlights the ESG values as a tool to contribute to sustainable
investments and to promote green transition (Blackrock n.d.b). The increased interest
in ESG has simultaneously spurred academic research, how to invest in ESG stocks,
and if higher ESG rating can generate positive significant alphas (Derwall et al. 2005,
Kempf & Osthoff 2007, Halbritter & Dorfleitner 2015).
The increased demand about ESG data intensified as of 2013-2014 on the global mar-
kets (Kell, 2019), and on the Swedish markets further in 2017-2018 with the imple-
mentation of the new sustainability law (Swedish Government, 2017). The law re-
quires public traded companies to include a sustainability report in the annual report,
highlighting the increased awareness and importance. On a larger scale, the European
Union decided to implement a new Taxonomy regarding sustainability (European Par-
liament, 2020) with the purpose to guide organisations and investors on how to conduct
a sustainable operation. With the regulatory introduction analysts have begun to em-
phasizes the sustainability criterion to a larger degree, as it can be argued that the score
a firm receives can be viewed as a precaution against potential lawsuits, and that a high
score can increase positive opinion on the company by the public (Bansal et al. 2018).
However, it is found that companies with higher ESG rating tend to have larger market
capitalization and are therefore more mature on the market, giving them a possible ad-
vantage on the market (Sargis & Wang n.d.).
1
In line with the increased demand about ESG data, the research field of ESG has in-
creased. The majority of academic papers on the topic are on the United States (US)
markets, and the studies (Derwall et al. 2005, Kempf & Osthoff 2007, Nofsinger &
Varma 2014, Halbritter & Dorfleitner 2015, Hartzmark & Sussman 2014) find different
results of the relation between stock returns and a company’s ESG score. For exam-
ple, Kempf and Osthoff (2007) and Derwall et al. (2005) find a positive relationship.
On the contrary, both Halbritter and Dorfleitner (2015), and Hartzmark and Sussman
(2019) argue that investors cannot expect abnormal returns by trading on a long-short
portfolio regarding ESG ratings, whereas Nofsinger and Varma (2014) find mixed re-
sults depending on the time period. Kaiser (2020) expands the analysis by studying the
relation in the US and in Europe and finds that investors in US and Europe can increase
a portfolios ESG level, and risk performance at the same time.
There is a consensus among researches that the scoring of ESG by providers is im-
portant, but considering the ambiguity by previous research and the lack of research
on the Swedish stock market, this topic should be further investigated. Therefore, the
purpose of this thesis is to narrow down the research and to only study the effect of
Environmental rating (in ESG) on stock returns on the Swedish stock exchange. This
is an interesting topic as little research on the Environmental factor alone has been
conducted, whereas emission and stock returns (Oestreich & Tsiakas 2015) in EU and
Germany have been investigated previously. The Environmental awareness among in-
vestors has increased as they express their concerns about Environmental sustainability
(OECD 2021).
To be able to achieve the purpose, the thesis follows the report of Kempf and Osthoff
(2007) and particularly their screening strategies. Kempf and Osthoff (2007) use a posi-
tive, negative and best-in-class screening strategy on the US market. Positive screening
includes all companies, negative screening excludes companies that are involved con-
troversial business (Hong & Kacperczyk 2007), and the best-in-class screening strat-
egy is balanced across industries. This thesis deviates from Kempf and Osthoff mainly
in two different ways. First, this thesis focuses only on the Environmental rating of
stock returns for Stockholm Stock Exchange (SSE), mid- and large cap stocks by using
data from Refinitiv Eikon when constructing portfolios. Secondly, for the best-in-class
screening strategy this thesis uses GICS industries rather than balance across industries
as Kempf and Osthoff (2007). This is the same best-in-class method as Derwall et al
(2005) and Halbritter and Dorfleitner (2015) use in their reports. Therefore, this thesis
2
will contribute to the literature as it is lacking an investment focus on the Environmen-
tal pillar, and will come with new insights from an investment perspective on the SSE.
Moreover, for the negative screening strategy, this thesis broadens the concept by in-
cluding ESG controversies as well.
With this background the main question of the report is: Does a trading strategy in
stocks, based on higher Environmental rating on the Stockholm stock exchange, lead to
positive abnormal performance?
To answer the question, this thesis uses the classic asset pricing models which are
CAPM, Fama-French three factor, and Carhart and creates long-only, and long-short
portfolios for each screening strategy. The long-only is an easy concept for the amateur
investor, as the investors only buys and holds stocks. The long-short strategy includes
a more active engagement, such as for professional investors, since the investors also
short-sells the low-rated portfolios. The sample period is between 2010-2020 with
monthly data, and 62-128 number of stocks will be investigated. The data sample grew
with time as more firms started to disclose public information about ESG related activi-
ties, in combination with increased regulatory pressure in 2017. The new law increased
the sample size significantly as the portfolios are re-balanced annually. The rating of
stocks is collected from Refinitiv Eikon, where the Environmental scores is between
A+ to D-. The A+ score is the best rating and the D- score is the worst rating.
The choice of using the letter rating from Refinitiv Eikon, rather than the more com-
mon approach in previous literature (Halbritter & Dorfleitner 2015) which is to use
the numerical values of rating, is because the letter rating is shown for investors when
searching for ESG rating on the Refinitiv Eikon database. Additionally, it is more ap-
propriate to use letters and absolute value, rather than relative, when sorting the stocks
out for the high-rated and low-rated portfolios. Since a portfolio construction based on
a relative measurement could sort stocks that perform well, with stocks that performs
bad, because they both would be evaluated as the bottom 50% of the portfolio.
The results of the thesis are strongest with Carhart model for long-only portfolios,
where the highest return is generated with the negative screening strategy with a monthly
return of 0.0156% between 2010-2020. The best-in-class screening strategy generates
mixed and inconclusive results for the long-only and long-short portfolios for all mod-
els between 2010-2020. When dividing the full-period to two periods with respect to
3
the sustainability law implementation in 2017, the long-only portfolio shows significant
results before 2017 for the positive and negative screening strategies. After the sustain-
ability law implementation, Fama-French shows positive and significant results with the
negative screening strategy, and Carhart shows positive and significant results for both
the positive and negative screening strategies. The long-short portfolios only show sig-
nificant results before the sustainability law implementation with CAPM and Carhart,
and those returns are negative. The best-in-class screening strategy generates mixed
and inconclusive results for all periods, models and strategies. The equally-weighted,
long-only portfolios generates positive and significant returns for all regressions for the
positive and negative screening strategies, whereas the long-short portfolios generates
negative and significant results with the positive screening strategy. The best-in-class
screening strategy with equally-weighted portfolios continues to be mixed and incon-
clusive for both long-only and long-short portfolios.
The rest of the thesis has the following outline: section 2 summarizes the thesis theo-
retical predictions and previous research on ESG, and the Environmental part of ESG.
Section 3 presents the data and methodological approach of the thesis, and a summary
of the stock returns of GICS industries on the Stockholm stock exchange. The results of
the thesis are presented in section 4 and in section 5 the results are interpreted. Section
6 presents the conclusion of the thesis.
2 Literature Review and Theory
2.1 Theoretical Background
2.1.1 Efficient Market Hypothesis
The efficient market hypothesis (EMH) argues that all prices of the market reflect all
available information. According to EMH stocks always trade at the correct price at
any given time, which makes it impossible for investors to outperform or underperform
the market by stock selecting or market timing. For investors to obtain higher returns
than the market, they must invest in stocks with higher risk. Fama (1970) argues that
even if the hypothesis is strong, the degree of efficiency depends on the market situ-
ation and changes after that. Fama therefore divide these degrees into three different
versions: a weak; semi-strong; and strong form.
The weak-form suggests that today’s stock prices include all information and data of
4
historical prices. If the weak form holds it is not possible to achieve higher returns
by using a strategy based on historical prices. The semi-strong form suggest that all
public information is incorporated to the current stock price, like announcements of
stock splits and annual earnings (ibid). The theoretical scenario of the EMH is of great
relevance to this thesis as EMH indicates that it is not possible for investors to achieve
higher returns than the market with different screening strategies because the stock
price would already include the firms Environmental-score.
If the semi-strong hypothesis holds, a high-or low Environmental-score, would not be
to a firms advantage with respect to the stock price. The strong-form suggests that all
information, including both public and not public information, is incorporated into the
stock price at all time. The implication is that no other information can provide any
investor the possibility to gain advantage of the market. Subsequently, it would also
not be possible to gain higher returns through insider trading (Rozeff & Zaman 1988).
2.1.2 Shareholder Theory: A Friedman Doctrine
With his New York Times article, Milton Friedman (1970), laid the foundation for the
Shareholder Theory to enter the public. The theory rose to prominence in the 1980s
with his TV-series ”Free to Choose” in 1980, which got it name and inspiration from
the book written by himself and his wife, Rose Friedman (Friedman & Friedman 1980).
The shareholder theory states that the sole purpose of a company is to provide returns
for the shareholders, not to take responsibility outside of that realm on the expense of
the shareholders asset and return. If an executive, or shareholder, of the firm wishes
to do so, they shall do so on their own expanse (A Friedman doctrine - The Social Re-
sponsibility Of Business Is to Increase Its Profits n.d.).
Milton Friedman ideas were not taken from nowhere as similar thoughts had been
formulated by him earlier (Friedman 1962), with inspiration from a long history of
economists which argue the similar thoughts, dating back to the Enlightenment (Smith
1776-1790). Friedman argues that the business executive responsible for the share-
holders firm can engage in activities which does not maximize the value of the firm,
and does so in the executives own interest, rather than the shareholders.
5
With the increased regulations regarding sustainability disclosure and market demand,
it is now questionable if this theory holds as it can be considered to be value creation
for the shareholders if emphasis is put on Environmental related questions. With the
increased demand a new theory of Stakeholder Theory have come to prominence the
last decade which partly argue against the shareholder theory.
2.1.3 Stakeholder Theory
As Friedman’s shareholder theory received criticism, the work of R.Edward Freeman
(1984) gained attention as it broadened the scope to include more than the sharehold-
ers. Freeman’s thoughts were not new, as Klaus Schwab (Schwab 1971) had shared
the same thoughts in 1971 with his work on how the modern mechanical firm should
be run by executives. The stakeholder theory includes more than the shareholders such
as employees, customers, suppliers, and other parties which the firms comes in contact
with, directly or in-directly.
Freeman (1984) argues that if the focus lies on the stakeholders, the firm will create
long-term benefits for all, not only the shareholders. Criticism for this theory states
that the executives have a fiduciary responsibility to serve the shareholders first (Miles
2012), and not the stakeholders. Despite the criticism, several authors argue that a
stakeholder approach creates value as there is a connection between stakeholder activity
and economic performance (Berman et al. 2017) as the business creates a competitive
advantage when including stakeholders in their business plan (Clarkson n.d., Freeman
1984) whilst creating value for the shareholders as well.
2.2 Previous Research
The interest in exploring the relationship between sustainability and stock returns has
increased during the last years, where the results from academic reports are not con-
sistent with each other. Kempf and Osthoff (2007) argue that it pays to be sustainable,
and others find mixed results such as Nofsinger & Varma (2014), Halbritter & Dorfleit-
ner (2015). The report of Kempf and Osthoff (2007) have been influential as multiple
papers (Nofsinger & Varma 2014, Halbritter & Dorfleitner 2015) have applied their
screening strategies. This part of the thesis will guide through earlier research on ESG
and the Environmental part of ESG.
6
2.2.1 ESG Research
The main focus of earlier research is to study the effect of ESG and stock returns on
the US stock market. A prominent paper in the literature comes from Kempf and Os-
thoff (2007) who study the relationship between sustainability and stock prices. The
report find that buying stocks with high social responsible ratings (SRI), and selling
stocks with low SRI leads to higher abnormal returns. In their 2007 report, Kempf and
Osthoff use the KLD database ratings from 1992-2004 to measure the SRI of compa-
nies listed on the US stock market, ranging from 500-3000 firms. The authors form
value-weighted portfolios, one with high SRI rating, and another with low SRI rat-
ing. For the main analysis the high-rated portfolio includes the top rated 10% of all
stocks, and the low-rated portfolio includes the bottom 10% of all stocks, in addition
Kempf and Osthoff (2007) divided the portfolios in top and bottom 5%, 25%, and 50%.
Furthermore, Kempf and Osthoff (2007) study the impact of various SRI criterion’s
on the performance of a screened stock portfolio where they apply a negative, posi-
tive and best-in-class screening strategy. The negative screening strategy excludes all
companies involved in controversial business areas such as Tobacco, Gambling, and
Alcohol (Hong & Kacperczyk 2007). The positive screening do not exclude companies
involved in controversial business, but rates all companies depending on a set of crite-
rion’s such as Human Rights, Gambling and Environment. The best-in-class screening
strategy is the same approach as the positive screening but the portfolio is balanced
across industries (Kempf & Osthoff 2007).
By using the Carhart (1997) four factor model to measure the performance, Kempf
and Osthoff (2007) conclude that the balanced long-short portfolio perform better than
both the high-rated and low-rated portfolio. The authors find that the positive and best-
in-class screening strategy both yields a positive alpha, and the negative screening of
the long-short portfolio does not. The best-in-class strategy yields the highest positive
alpha of 0.7% monthly with Carhart four factor model. To validate their results, Kempf
and Osthoff (2007) also investigate the sensitivity of the value-weighted portfolio by
forming equal-weighted portfolios, where they find that the results are similar to the
value-weighted portfolios. The equal-weighted, long-short portfolio of the negative
screening strategy does not yield a positive alpha, but so do the positive and best-in-
class screening strategy of the equal weighted, long-short portfolios (Kempf & Osthoff
2007). As a last test Kempf and Osthoff (2007) divide their sample in two time periods
for the value-weighted portfolios which generate the same results as for the full period.
7
In line with Kempf and Osthoff (2007) report, Halbritter and Dorfleitner (2015) in-
vestigate the relationship between a company’s financial performance which is based
on ESG ratings on the US stock market. With the application of the screening strat-
egy as mentioned previously, with the difference of the best-in-class screening strategy
where they use GICS classification, they obtain different results compared to Kempf
and Osthoff (2007). Halbritter and Dorfleitner (2015) collect ESG data from ASSET4,
Bloomberg, and KLD for a sample period between 1991-2012 and apply Carhart’s
(1997) four factor model and Fama-MacBeth (1973) regression. The 20% best firms
are assigned to the top portfolio, and the 20% worst are assigned to the bottom portfo-
lio, and from this they generate long-short portfolios whilst simultaneously analysing
top and bottom portfolio separately. The results are positive, but with insignificant ab-
normal returns, invalidating any claim. Halbritter and Dorfleitner argue that investors
should not expect abnormal returns by trading on a long-short portfolio based on ESG
ratings, contradicting Kempf and Osthoffs (2007).
Diaz et al. (2021) investigate the importance of ESG ratings for different industry
returns during the Covid-19 pandemic on the US market. As Halbritter and Dorfleitner
(2015), they use GICS industry classification. With Fama-French three factor model
they find that the Environmental and Social part of ESG are the drivers of the patterns
that was observed in the result, and that the impact of ESG and its pillars varies across
industries (Diaz et al. 2021). A report written by Van duuren et al. (2016) study ESG
factors from another perspective in the US, compare to Kempf and Osthoff (2007),
Halbritter and Dorfleitner (2015) and Diaz et al. (2021). Van duuren et al. (2016) study
how conventional asset managers include ESG factors in their investment processes
using a survey among fund managers. As Kempf and Osthoff (2007), and Halbritter
and Dorfleitner (2015), they apply the screening strategies to evaluate the performance
of stocks on the US market. The reports main conclusion is that conventional fund
managers have adopted aspects of responsible investing in their investment process.
Furthermore, Van duuren et al. (2016) find that for professional asset managers, Gov-
ernance is more important than the Environmental and Social factors.
Other reports on the topic in the US have find mixed results. The reports from Hartz-
mark and Sussman (2014), and Nofsinger and Varma (2014) find mixed and insignif-
icant results on the US market, as Halbrittner and Dorfleitner (2015). Hartzmark and
Sussmann (2014) do not find any evidence of high-rated fund portfolios outperform-
8
ing low-rated, when applying the Carhart four factor model. Nofsinger and Varma
(2014) apply CAPM, Fama-French three factor model and Carhart with inspiration
from Kempf and Osthoff (2007) screening strategies. The results from the Nofsinger
and Varma report show mixed results depending on the time-period during crisis, ESG
funds outperform, whereas during non-crisis the conventional funds outperform (ibid).
The research about ESG and stock returns outside the US market is limited, though
Kaiser (2020) expands the research from studying the US market to also include the
European market. Kaiser concludes that US and European investors can increase their
portfolio’s ESG level, and increase the risk-adjusted performance simultaneously. The
sample in the research consists of 1079 European firms and 1756 US firms between
2002-2015, excluding those that do not have ESG ratings for the full period from the
report. Kaiser uses the impact of ESG integration on value, momentum and growth and
Carhart four factor model to find the results. Most noticeably he finds that systematic
risk is lower after including ESG, apart from the European value portfolio, secondly he
finds that ESG integration lowers the size exposure (Kaiser 2020).
The researchers Limkriangkrai, Koh and Durand (2015) investigate ESG ratings effect
on stock returns and corporate financing decisions on the Australian equity market.
ESG data is collected by the Australian company Regnan from 200 companies that
are listed on the Australian exchange over the sample period 2009-2014. The authors
find inspiration from Kempf and Osthoff (2007) and apply the same screenings s, with
GICS Sectors for the best-in-class screening. Limkriangkrai, Koh and Durand (2015)
create a high-score portfolio of the stocks with the highest rating, and a low-score port-
folio of the stocks with the lowest rating using Carhart four factor model. The results
from Limkriangkrai, Koh and Durand show that the alphas were insignificant. (Limkri-
angkrai et al. 2017).
2.2.2 Environmental Research
Other studies have narrowed down their research to focus on the Environmental factor
in ESG, but this research is even more limited and has also mainly been studied on
the US stock market. In 2005, Derwall et al. (2005) find that equity portfolios with
high scores on eco-efficiency, score higher risk-adjusted returns than portfolios with
low scores on eco-efficiency between 1995-2003 in the US. They use the best-in-class
screening strategy to investigate if a long-run premium or penalty exists for holding
environmentally responsible companies. Derwall et al. (2005) create two mutually ex-
9
clusive stock portfolios with distinctive eco-efficiency characteristics, one high (low)
ranked portfolio which consists of companies making up to the 30% of total capital-
ization rated the highest (lowest). By subtracting the low-ranked portfolio with the
high-ranked portfolio returns, Derwall et al (2005) are able to generate the returns for
the difference portfolio and create an additional portfolio.
With the CAPM model, Derwall et al (2005) find that alpha for all three portfolios
are positive, but the result is not statistically significant for either. With Carhart four
factor model they find that the high-ranked portfolio earns a significant average monthly
return of 0.3%, and the low-ranked portfolio do poorly. Moreover, they find that the
high-rated and long-short portfolio is significant on a 10% level. Derwall et al. (2005)
conclude that companies that perform well along with the Environmental dimensions
have higher returns compared to those that perform poorly.
With respect to Environmental rating and the effects of it on the European markets,
little research have been conducted. One interesting report is written by Oesterich
and Tsiakas (2015), who study the relationship between greenhouse emissions trading
scheme and stock returns for the German market from 2005 to 2012, making it one of
few who solely focus on a European market. The report examine the data by applying
the Fama-French three factor and Carhart four factor model, and find that investors can
expect a carbon premium for investing in firms which invest more in containing their
emission.
The previous research literature focus on ESG, emissions, or on the social responsi-
bilities on the US markets. Few researches focus solely on the Environmental pillar in
ESG, creating a void in the earlier research, and literature. This thesis aim is to broaden
the research and only focus on the Environmental pillar in ESG, and investigate stock
returns for value-weighted, long-only and long-short portfolios, based on the ratings in
the Environmental pillar, on SSE. This can be achieved by implementing the previous
methods and screening strategies Derwall et al (2005), Kempf and Osthoff (2007), and
Halbrittner and Dorfleinter (2015) use in their reports. This thesis follows the outline
of Kempf and Osthoff (2007) and will, therefore, apart from using the screening strate-
gies for value-weighted, long-only and long-short portfolios, generate sub-periods (one
period before the sustainability law implementation of 2017, and one period after) and
create equally-weighted portfolios to investigate the sensitiveness of the results.
10
3 Data and Methodology
3.1 Data
3.1.1 Sample
The sample is selected from the stocks listed on the Nasdaq Stockholm Stock Exchange
(SSE) main list, which include large-, mid-, and small cap companies. Small caps are
excluded from this thesis sample as there is a low amount of small caps with an Environ-
mental rating between 2010-2016. The increase of firms, both small cap and in general,
with ESG rating from 2017 is due to the new sustainability law that was implemented
in 2017 which states that firms shall implement a sustainability report (Sustainability
Report, n.d.) in addition to the annual reports (Government n.d.) and this increased the
firms public information on ESG. There are numerous databases providing informa-
tion about ESG rating, such as Morningstar, Bloomberg, Thomson Reuters, and KLD
Research, which all have been used in various studies (Derwall et al. 2005, Kempf &
Osthoff 2007, Hartzmark & Sussman 2014). This thesis examines the rating for the En-
vironmental rating provided by Thomson Reuters Eikon (Refinitiv 2020) as it provides
a thorough historical ESG rating for the Swedish market.
The ratings are downloaded from the Refinitiv Eikon database and is based on pub-
lic information available for all investors. It is one of the most extensive financial
databases, with ESG rating dating back to 2002, where the data is analyzed by more
than 150 analysts. The reason for choosing Thomson Reuters Eikon is because the
database offers the most extensive data on the Environmental factor making it a rel-
evant database for the research question. A screening process tool provided on the
Refinitiv Eikon database is used to ensure that the desired data for the research ques-
tion is collected. The criterion’s for the paper are that stocks are listed on the SSE and
have an Environmental rating from 2010 to 2020.
The sample size for the paper grew in scope as the there is an increased awareness
and demand for ESG by investors, institutions, and professionals (Blackrock n.d.a) on
a global level. As this paper follows the report of Kempf and Osthoff, the optimal
sample period would have been between 2008-2020, to match their 12 year period.
However, since the demand of ESG rating by investors has increased during the last
years the ratings before 2010 are of inadequate size, and therefore this thesis decided to
collect data from 2010 and forward. One of the reasons why the pressure has increased
11
recently is because of the new sustainability law that was implemented in 2017. An-
other reason is that starting the data in 2008 instead of 2010, in the midst of the Great
Financial Crises, would skew the results for a long period of time as there was a strong
recovery period on the stock market in late 2008 and through 2009. It could be argued
that the last year (2020) should be excluded from the thesis due to Covid-19, but since
volatility is a part of the stock market it has been included as it only occurred for a few
months.
3.1.2 Environmental Rating
Refinitiv Eikon (Refinitiv 2020) offers an ESG score for companies between A+ to D-,
where A+ is the highest ranking, and D- is the lowest ranking. The letter rating comes
from the numerical values Refinitiv Eikon offers, where the ESG score rating for the
three pillars – Environmental, Social, and Corporate Governance are calculated by a
matrix of criterion’s. The pillar score is the relative sum of the category weights, which
varies depending on the industry for the Environmental and Social categories whereas
for the Governance pillar, the weighting is the same for all firms. The Environmental
scoring is divided into three categories which are resource use, emissions and innova-
tion. The pillars have in total 10 subcategories that forms the basis for the overall score.
On the Refinitiv Eikon database, the data for the pillar weights are normalised, and
given a value between 0 to 1 after evaluation, as it simplifies the rating. The pillar
score is transformed into quarterlies grading, defined by the letters A+ to D-, and this
is the grading which is displayed on the Refinitiv Eikon database when searching for
the ESG ratings of a firm. The logic behind the transformation and rating is illustrated
by Appendix A.3 provided by Refinitiv Eikon. The ESG rating of a firm is a relative
measurement, and it is evaluated in the context of it peers on the stock market. Fur-
thermore, the thesis examines the grading system A+ to D- as it provides the basis of
investment decision by the investors who browse Refinitiv Eikon to differentiate and
decide which companies to invest in depending on the rating. This will provide a better
reflection of the rating as the worst rated stocks are grouped together, otherwise would
be possible to group stocks with B rating in the low-rated portfolio, together with D
rated stocks, only because it could belong to the bottom 50% in total, and not reflecting
the true rating.
12
3.1.3 Data Management
The price data is collected from the Swedish House of Finance Data Base (SHoFDB)
which offers historical data on all stocks listed on SSE, including equity market value,
total market equity value, and stock prices adjusted for corporate actions (SHoF n.d.).
Companies with foreign ISIN or dual stock listing, are included in the data sample as
the ESG rating of the firm is independent of its country origin. When there is missing
data with respect to price, the price from the previous period is used. If a stock has
more than three periods of data missing during a year, the stock is omitted from the
portfolio to ensure accurate data (Bodie et al. 2021).
When constructing the portfolios, the market value of equity is used, and when there
is lacking data entries, the total market equity is used instead. When a stock has dif-
ferent classes such as A, or B shares, and one of the stocks classes market equity is
missing, the value of equity is calculated by subtracting the share equity of either share
with the total market equity for the combined classes of shares. The Fama-French fac-
tors are collected from the AQR’s database and are collected by monthly observations
from February 2010 to December 2020 (AQR n.d.). The variables for the Fama-French
regression are Rf, MKT, SMB, HML, and UMD (see chapter 3.2 for factors) and are
calculated from the SSE, making them highly relevant for this study. With respect to
the standard errors, the robust standard errors is used in order to achieve standard errors
which are unbiased and address heteroskedasticity.
3.1.4 Portfolio Construction
The prices are collected monthly, ranging from January 2010 to December 2020. The
stock prices are adjusted for corporate actions, and in addition to the stock prices,
SHoFDB is also used to collect the total market equity value, and the market value
for the stock. With the data from SHoFDB the simple monthly returns for the individ-
ual stock is calculated. The procedure enables the creation of a time-series regression
for the monthly returns, based on the chosen interval between 2010-2020. The choice
of working with absolute values (reflected by the alphabetical letter rating) instead of a
relative measurement when constructing the portfolios are that it reflects the true rating
better as it is what investors sees on the Refinitv Eikon when searching for ESG rating.
The letter rating and subsequently portfolio construction creates two portfolios for each
screening process, and a long-short portfolio in addition. The portfolios are constructed
by dividing the stocks in one high-rated and one low-rated portfolio. To sort the stocks
13
according to their rating, the rating from Refinitiv Eikon will act as a threshold. All
stocks with a rating of A (+/-) or B (+/-) is included in the high-rated portfolio, and
all stocks with a rating of C (+/-) or D (+/-) is included in the low-rated portfolio (See
chapter 3.2.4).
The simple returns, used for both portfolios is calculated by subtracting the previous
periods stock price, from the price of the stock of the current period and subsequently
dividing it with the price of the stock from the previous period. Value weighted portfo-
lios are constructed by taking the sum of market capitalization for the given time-period
used, with the individual stock market cap in the numerator and multiplying it with the
return of the given period. The formula for the value-weighted portfolios is specified
as follows:
MVi,t
V alue weighted = ∑ (1)
MVi,t
MVi,t = Stock i’s market capitalization on day t
The choice of focusing on the value weighted portfolio construction, rather than an
equally weighted, stems from the fact that this thesis follows the report of Kempf and
Osthoff where value-weighted portfolios are constructed for the main analysis. Also,
equally-weighted portfolios exposure to size and value factors, is greater than for value
weighted portfolios (Plyakha et al. 2021). Therefore, value-weighted portfolios are
more suitable for the thesis as value-weighed addresses the issue of smaller stocks,
with high risk and high returns, and gives a better balance for portfolio construction. To
further investigate if the value-weighted portfolios are sensitive to the tilting of stocks,
this thesis conducts equally-weighted portfolios for the full sample period based on the
stocks returns for the given portfolio. The equally-weighted formula is specified as fol-
lows:
1
Equally weighted = ∑ (2)
Ni,t
Ni,t = Number of stocks i on day t
14
3.1.5 Re-balancing
Following the work of Kempf and Osthoff (2007) and Derwall et al (2005), the first
reason for re-balancing the portfolios on the first trading day of the year is because they
remain the same throughout the year. The other reason is that the rating from Thom-
son Reuters Eikon is provided on an annual basis and that the portfolios subsequently
will reflect the expected return more accurately. Firms that are listed during the year,
provided that the firm has a rating, are omitted from the portfolio as it shall remain
unchanged throughout the year.
Newly listed firms are not included in the portfolio until the coming year (t+1) pro-
vided that the stock has an Environmental score rating. For delisted stocks there is a
difference depending on the reason. If a stock is delisted due to bankruptcy the firm
will be given a price equal to zero for the reminder of the year. If a stock is delisted due
to a tender offer the last prices of the firm will be used for the remainder of the year,
provided that the firm has more than seven months of price data for the year.
3.1.6 Limitations
In the thesis, data have been omitted with respect to abnormal returns. As the purpose
of the thesis is to analyze if investing in high score Environmental rated stocks gener-
ate abnormal performance, the thesis aims to reflect the underlying value with respect
to the models used, and the assumed normal distribution. A winsorization method is
conducted, and omissions for outliers for the returns is conducted where it is appropri-
ate, which have resulted in the exclusion of the Energy sector and Technology sector
from the best-in-class screening strategy. The two portfolios are excluded due to lack
of stocks and because of the the high volatility of the technology portfolio between the
years 2014-2017. The Energy sector only constitutes of one stock for nine out of the
11 years and cannot be considered to be a portfolio.
The asset pricing models might generate abnormal returns, whether these returns are
abnormal due to omitted variables bias or not, is not possible to examine in this thesis
(Fama & French 1993). It is reasonable to expect that some unobserved variables could
explain the result more accurately. The choice of time-period (2010-2020) could be a
concern as more data could generate more accurate results. Additionally, it is possible
that the exclusion of small caps can affect our results as they tend to generate larger
return than large cap stocks.
15
3.2 Methodology
3.2.1 Capital Asset Pricing Model
The Capital Asset Pricing model (CAPM) is a well known model in asset pricing and
the model explains the relation between expected return and risk. The CAPM model
provided the first framework for this relationship between risk and return when it was
developed in the early 1960s by William Sharpe (1964), Jack Treynor (1962), John
Lintner (1965a, b) and Jan Mossin (1966). This thesis purpose is to measure portfolio
return and therefore the equation of Jensens Alpha is used. Jensens Alpha is specified
in the following way:
Rit − rft = αit + βi(rmt − rft) + ϵit (3)
Rit = Return of stock i at time t
rft = Risk-Free-Rate at time t
α = Intercept (abnormal return) for stock i at time t
(rmt − rft) = Market Risk Premium (MKT) at time t
ϵit = Error term for stock i at time t
3.2.2 Fama-French Three Factor Model
The CAPM model was further developed in 1993 by Fama and French and they propose
two additional factors to CAPM, SMB (”Small Minus Big”) and HML (”High Minus
Low”). The SMB factor is constructed to measure if investing in relatively small mar-
ket capitalization stocks gives additional return historically. The SMB factor is also
called the ”size premium”. The HML factor is designed to measure if higher book-to-
market value generates higher returns and is referred to as the ”value premium” (Fama
& French 1993). When the asset pricing model performs well, it explains all expected
return. The model is specified in the following way:
Rit − rft = αit + β1MKTt + β2SMBt + β3HMLt + ϵit (4)
β2 = Coefficient of SMB and gives an indication whether the portfolio tilts to growth
companies or value companies at time t
β3 = Coefficient of HML and gives an indication whether the portfolio tilts to growth
companies or value companies at time t
16
SMBt = Average return of Small-Minus-Big firms at time t
HMLt = Average return of High-Minus-Low firms at time t
Small firms with small market capitalization tend to outperform large stocks with large
market capitalization. When β2 is less than zero the portfolio is tilted towards value
stocks, and if β2 is larger than zero the portfolio is tilted towards growth stocks. The
opposite is true for β3, where, theoretically, high book-to-market stocks (value stocks)
tend to outperform low book-to-market stocks (growth stocks). If β3 is larger than zero
it indicates that the portfolio is more invested in value stocks, if β3 is smaller than zero
it indicates that the portfolio is more invested in growth stocks (Fama & French 1993).
3.2.3 Carhart Four Factor Model
Carhart (1997) adds one more factor to the Fama-French three factor model. The forth
factor added to the regression is the momentum factor (UMD). The momentum strategy
goes long in the winner portfolio, and short in the loser portfolio based on historical
returns (Carhart 1997). The regression is defined in the following way:
Rit − rft = αit + β1MKTt + β2SMBt + β3HMLt + β4UMDt + ϵit (5)
β4 = Coefficient of UMD gives an indication in which way the portfolio tilt to momen-
tum for stock i
UMDt = Average return of winning stock minus loser stock at time t
The momentum factor is included in the regression to take the momentum effect into
consideration for the excess return. If β4 is larger than zero it indicates that the portfolio
is mainly invested in winner stocks, and if β4 is smaller than zero it indicates that the
portfolio is mainly invested in loser stocks (Carhart 1997).
3.2.4 Portfolio Formation and Screening
Three regression models based on monthly observations will be conducted to evaluate
the data; The Jensen’s model (referred to as CAPM in the result section) (Perold 2004),
the Fama-French three factor model (Fama & French 1993), and the Carhart model
(Carhart 1997). Following the report of Kempf and Osthoff (2007) this thesis applies
the screening strategies; positive, negative and best-in-class. The screening strategies
will enable an analysis of long-only portfolios, meaning going long in the high-rated
stocks, long-short portfolios, meaning going long in the high-rated stocks, and short in
17
the low-rated stocks. Of the two portfolios, the long-only is the easier choice for the
investor as it is a buy-and-hold strategy, whereas the long-short demands a more active
part from the investor in the short-selling. Short-only portfolios (low-rated) are not part
of the analysis as the thesis wishes to investigate the relationship between Environmen-
tal high-rated stocks and alpha. The choice of including the long-short portfolio comes
from the desire to investigate if it could increase the significance and the alphas of the
results.
As Kempf and Osthoff (2007), the main analysis is based on the value-weighted portfo-
lios, but to test if the value-weighted portfolios results are dependent on the time-period,
and, or weighting scheme as Kempf and Osthoff (2007); the full sample is divided in
sub-periods where the first period is between 2010-2016 before the sustainability law
implementation and the second is between 2017-2020 after the sustainability law im-
plementation, also equally-weighted for the full sample is constructed for long-only
and long-short portfolios. As this report has narrowed down the focus to Environmen-
tal rating, only the Environment criteria is used throughout all screening strategies and
models.
Kempf and Osthoff (2007) divided the portfolios in top and bottom 5%, 25% and 50%
respectively. For this thesis only the top 50% stocks, and bottom 50% stocks is applied
because it is not possible to divide the portfolios into the same cut-offs as Kempf and
Osthoff, due to limited number of stocks listed on SSE. The portfolios of the positive
screening strategy includes firms with the rating A+ to B- in the high-rated portfolio
(top portfolio) and companies with C+ to D- rating in the low-rated (bottom portfolio),
creating a threshold line. The negative screening strategy excludes stocks which are in-
volved in ESG controversies and controversial business from the high-rated portfolio,
and are included in the low-rated portfolio, inspired by Hong and Kacperczyk (2007)
and there sin stock definition. The definition of ESG controversies is provided by Re-
finitiv Eikon as they have an additional ranking where they provided annual data and
ranking for companies involved in controversies. The ranking is the same as for the
ESG scores. Controversial business include companies that are involved with military,
gambling and tobacco for example. The exclusion of the controversial stocks is done
manually, based on the Hong and Kacperczyk (2007) method.
For the best-in-class screening strategy all companies are divided into eight industry
classes based on their GICS sector, and the stocks are ranked according to their rating
18
within each industry class. The choice of constructing portfolios based on industry sec-
tor, comes from Halbrittner and Dorfleitner (2015) who follows the same approach for
the best-in-class screening strategy. Thereafter, the companies with the rating of A+ to
B- are included in the top portfolio and the companies with C+ to D- are included in
the bottom portfolio. For all screening strategies, long-only and long-short portfolios
are constructed, as for the positive and negative screening strategies. See table below
for the industry classes, and number of stocks included in each industry per year:
Table 1: GICS Industries
GICS Industry 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
Basic Material 11 11 9 11 10 10 13 13 10 12 12
Consumer Discretionary 5 5 5 5 5 6 6 7 12 13 13
Consumer Staples 2 2 2 2 2 2 2 2 4 4 4
Financials 15 15 15 15 17 18 19 18 21 21 22
Health Care 3 3 3 3 4 4 4 4 8 10 14
Industrial 13 13 13 12 14 14 18 16 27 29 35
Real Estate 3 3 5 5 5 6 5 5 11 17 13
Communication Services 6 6 6 6 6 6 6 6 6 6 7
SUM 58 58 58 59 63 66 73 71 99 112 120
3.2.5 GICS Industry and Classification
The grouping of the selected firms for the best-in-class screening strategy is based on
the GICS Classifaction, Global Industry Classification Standard (MSCI n.d.) and in-
clude 11 industries. The taxonomy was created in 1999 by MSCI and Standard &
Poor’s (SP) in order to sort firms into specific categories, enabling an easier user expe-
rience for the financial community. Apart from the Energy and Technology sectors that
are excluded from this thesis as explained under the limitations section, the Utility sec-
tor is excluded as this sector only provides rating, and returns, for the last year (2020)
of the sample period.
The SSE index is heavily tilted towards the Industrial, Financials and Real Estate sec-
tors, which is reflected in this thesis sample. These three industries are the sectors
which have performed best with respect to the stock returns during the sample period
(Refinitiv Eikon n.d.), which is unsurprising considering the overall importance of these
sectors for the export oriented Swedish economy. The significance of the financial sec-
tor on the SSE, is expected considering its size for the Swedish economy. The academic
report by Weis, Woltering and Sebastian (2021) investigate how the financial and real
19
estate sector benefits from low interest rates as there valuation increases. The authors
explain how interest rates affect various real estate stocks depending och stocks and
whether it is a value or growth firm (Weis et al. 2021), which is of significance for
considering the size of the financial sector on SSE.
For other sectors used in the thesis, such as the Health Care sector, has heterogeneous
composition as it has global actors and growth companies within the GICS classifica-
tion. Basic Material shows different patterns in composition and returns in comparison
to the Health Care due to commodity prices. The Basic Material and Communications
sector have been stable or decreasing, during the last decade (Sweden Exports n.d.),
subsequently affecting the revenue of the firms in the industry and the valuations of the
stocks. Consumer products, both Staples and Discretionary are not a large sector of the
Swedish economy, or stock market. The Consumer Staples industry in Sweden has for
long been dominated by ICA, Axfood, and Coop, making it difficult for new entries,
which is reflected in the number of firms listed on SSE (Market share of selected gro-
cery retailers in Sweden n.d.). The same does not hold for Consumer Discretionary as
the gambling and gaming sector is grouped in the same sector as retailing, dominated
by the H&M stock. The SSE has seen large increase in gambling and gaming related
firms listed as of 2015, when the sector grew rapidly in size (Refinitiv Eikon n.d.).
20
4 Results
The first part of this section presents the results of each screening strategy (positive,
negative and best-in-class) of alpha for long-only and long-short, value-weighted port-
folios of each model (CAPM, Fama-French three factor, and Carhart). In the next
section, the full-period is divided in two periods with respect to the sustainability law
implementation in 2017 for long-only and long-short, value-weighted portfolios of each
screening strategy. Lastly, equally-weighted long-only and long-short portfolios are
constructed for all models for the full period. All factors of the long-only (high-rated),
short-only (low-rated) and long-short portfolios can be seen in Appendix A.1. The out-
line of this section follows the structure of Kempf and Osthoff (2007) report, but for this
thesis all returns of alpha for each screening strategy are based on stocks Environmental
rating.
4.1 Screening Strategies
In Table 2, the results of the alphas for the positive and negative screening are pre-
sented for the value-weighted, long-only and long-short portfolio of each model. For
the long-only portfolio, the positive and negative screening strategies of each regres-
sion generate positive and significant results at a 0.1% level. For the positive screening
strategy, the strongest result is generated by Carhart with a monthly return of 0.0143%,
and for the negative screening strategy, the strongest result is also generated by Carhart
with a monthly return of 0.0156%. For the long-short portfolio there are no significant
results and the only positive return is generated with the positive screening strategy
with Carhart.
Table 2: Positive and negative screening of long-only and long-short portfolios (value-
weighted)
CAPM Fama-French 3 Carhart
Positive Negative Positive Negative Positive Negative
long-only 0.0112∗∗∗ 0.0134∗∗∗ 0.0117∗∗∗ 0.0138∗∗∗ 0.0143∗∗∗ 0.0156∗∗∗
(3.88) (4.44) (4.10) (4.63) (4.38) (4.74)
long-short -0.00230 -0.00509 -0.00202 -0.00506 0.000807 -0.005900
(-0.81) (-1.52) (-0.70) (-1.50) (0.26) (-1.38)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
21
The results in Table 3 show the alphas for the best-in-class screening strategy of the
long-only and long-short, value-weighted portfolios. For the long-only portfolios there
are no significant results with CAPM for the Basic Material, Financial, and Consumer
Discretionary, Communication sectors, but all returns are positive. For the other sec-
tors, the alphas are significant, where the strongest result is generated from the Indus-
trial sector with a monthly return of 0.00969% and a significant level of 1%. For the
long-short portfolios, there are no significant results and all sectors generate negative
returns apart from the Consumer Staples sector with CAPM.
With Fama-French three factor, also shown in Table 3, the long-only portfolios gener-
ate fewer significant results compared to CAPM. The Consumer Staples, Health Care,
Industrial, Communications and Real Estate sectors continue to be significant. The In-
dustrial and Real Estate sectors generate the strongest positive results with a significant
level of 1%, where the Real Estate sector generates the strongest monthly return of
0.0144%. As with CAPM, there are no significant results for the long-short portfolios
and all sectors generate negative returns apart from Consumer Staples and Real Estate
sector.
With Carhart (Table 3) the long-only portfolios continue to generate significant results
for the Industrial and Health Care sectors, whereas the Consumer Staples and Real
Estate sectors no longer show significant results. The Financial and Basic Material
sectors generate significant results for the first time with Carhart with a monthly return
of 0.0115% and 0.0012% respectively. For the long-short portfolio there are still no
significant results and all sectors generate negative returns apart from Basic Material,
Financials, Consumer Staples and Real Estate.
22
Table 3: Best-in-class of long-only and long-short portfolios (value-weighted)
CAPM
Basic Material Financials Consumer Staples Consumer Discretionary
long-only 0.00636 0.00492 0.0155∗ 0.00372
(1.42) (1.31) (2.11) (0.60)
long-short 0.0000704 -0.00412 0.0132 -0.00981
(0.00) (-1.05) (1.71) (-1.11)
Health Care Industrial Real Estate Communications
long-only 0.00872∗ 0.00969∗∗ 0.0133∗ 0.000250
(2.10) (2.82) (1.99) (0.06)
long-short -0.00828 -0.00241 -0.000371 -0.00753
(-0.95) (-0.56) (-0.06) (-1.25)
Fama-French 3
Basic Material Financials Consumer Staples Consumer Discretionary
long-only 0.00623 0.00543 0.0156∗ 0.00422
(1.41) (1.46) (2.09) (0.70)
long-short -0.00292 -0.00417 0.0134 -0.00894
(-0.12) (-1.04) (1.70) (-1.02)
Health Care Industrial Real Estate Communications
long-only 0.00899∗ 0.0100∗∗ 0.0144∗∗ 0.00110
(2.14) (2.99) (2.11) (0.29)
long-short -0.00795 -0.00221 0.000287 -0.00789
(-0.88) (-0.52) (0.04) (-1.39)
Carhart
Basic Material Financials Consumer Staples Consumer Discretionary
long-only 0.0112∗ 0.0115∗∗ 0.0180 0.00674
(2.23) (2.74) (1.68) (0.96)
long-short 0.00325 0.000295 0.0143 -0.00208
(0.16) (0.06) (1.23) (-0.20)
Health Care Industrial Real Estate Communications
long-only 0.0118∗ 0.0108∗∗ 0.0123 0.00120
(2.54) (2.73) (1.45) (0.31)
long-short -0.00609 -0.00170 0.00330 -0.00922
(-0.56) (-0.38) (0.38) (-1.38)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
23
4.2 Alternative Portfolio Construction
4.2.1 Sub-Periods
In Table 4, a sensitivity test for each screening strategy and model for the value-
weighted, long-only and long-short portfolios is conducted with inspiration from Kempf
and Osthoff (2007). The full sample period (2010-2020) is divided in two periods, one
before the sustainable law implementation (2010-2016), and one after (2017-2020).
For the long-only portfolio between 2010-2016 with CAPM, the positive and nega-
tive screening strategies generate positive and significant alphas at a 0.1% significance
level. After the implementation of the law the alphas are positive but not significant
with CAPM. For the long-short portfolio only the negative screening strategy between
2010-2016 generates a significant alpha at a 5% level.
Between 2010-2016, both Fama-French and Carhart generate positive and significant
results with the positive and negative screening strategies for the long-only portfolios.
Between 2017-2020 the negative screening strategy generates positive and significant
alphas, for Fama-French and Carhart, with the strongest alpha with Carhart. The pos-
itive screening generates a significant alpha with Carhart, but not with Fama-French,
in contrast to previous long-only period. The long-short portfolio generates significant
results for the negative screening strategy between 2010-2016 for Fama-French and
Carhart, but only before the law implementation at a 5% significance level.
24
Table 4: Positive and negative with sub-periods for long-only and long-short portfolios
(value-weighted)
2010-2016 2017-2020
Positive Negative Positive Negative
CAPM
long-only 0.0135∗∗∗ 0.0157∗∗∗ 0.00700 0.00891
(3.50) (4.07) (1.68) (1.91)
long-short -0.000428 -0.00907∗ -0.00519 0.00203
(-0.11) (-2.15) (-1.71) (0.38)
Fama-French 3
long-only 0.0139∗∗∗ 0.0160∗∗∗ 0.00767 0.00980∗
(3.58) (4.13) (1.92) (2.12)
long-short -0.000138 -0.00843 -0.00509 -0.0000351
(-0.03) (-1.97) (-1.87) (-0.01)
Carhart
long-only 0.0143∗∗ 0.0154∗∗∗ 0.0146∗∗ 0.0167∗∗
(3.21) (3.72) (3.11) (2.80)
long-short 0.00301 -0.0104∗ -0.00367 -0.000839
(0.66) (-2.07) (-1.26) (-0.12)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
In Table 5, the same sensitivity test have been applied to the best-in-class screening
strategy for value-weighted, long-only and long-short portfolios. There are few sta-
tistically significant results in this table as six out of 96 alphas generate significant
results and of those six, five are between 2010-2016. The strongest alpha for the long-
only portfolio is generated with Carhart for the Financial sector with a 1% significance
level. Other sectors showing significant results are more widespread and significant at
a 5% level. For the long-short portfolio there are no significant results for any industry
or model.
25
Table 5: Best-in-class with sub-periods for long-only and long-short portfolios (value-
weighted)
CAPM Fama-French 3 Carhart
2010-2016 2017-2020 2010-2016 2017-2020 2010-2016 2017-2020
Basic Material
long-only 0.0101 -0.000496 0.0103 -0.00168 0.0124 0.00562
(1.70) (-0.08) (1.71) (-0.29) (1.74) (0.81)
long-short -0.00132 0.00126 -0.00388 0.000574 -0.00730 0.00188
(-0.04) (0.14) (-0.11) (0.08) (-0.23) (0.19)
Consumer Discretionary
long-only 0.000208 0.00955 -0.000251 0.0125 -0.000358 0.0172
(0.03) (0.79) (-0.04) (1.14) (-0.04) (1.55)
long-short -0.00751 -0.0140 -0.00782 -0.00989 0.00297 -0.00901
(-0.65) (-1.02) (-0.68) (-0.75) (0.22) (-0.54)
Consumer Staples
long-only 0.0160 0.0150 0.0162 0.0143 0.0182 0.0154
(1.50) (1.90) (1.48) (1.87) (1.14) (1.33)
long-short 0.0160 0.00887 0.0162 0.00859 0.0182 0.00879
(1.50) (0.86) (1.48) (0.81) (1.14) (0.50)
Financials
long-only 0.0103∗ -0.00476 0.0106∗ -0.00376 0.0134∗∗ 0.0118
(2.45) (-0.70) (2.50) (-0.55) (2.82) (1.28)
long-short 0.00110 -0.0131 0.00121 -0.0138 0.00456 -0.00519
(0.26) (-1.63) (0.28) (-1.68) (0.95) (-0.45)
Health Care
long-only 0.00726 0.0111 0.00705 0.0129 0.00960 0.0183∗
(1.43) (1.53) (1.39) (1.70) (1.66) (2.20)
long-short -0.00297 -0.0167 -0.00364 -0.0132 -0.00136 -0.0118
(-0.26) (-1.30) (-0.30) (-1.08) (-0.10) (-0.64)
Industrials
long-only 0.0101∗ 0.00874 0.0103∗ 0.00952 0.00943 0.0106
(2.24) (1.66) (2.29) (1.94) (1.74) (1.78)
long-short -0.00390 0.000572 -0.00389 0.00122 -0.00107 -0.00504
(-0.73) (0.08) (-0.71) (0.17) (-0.19) (-0.65)
Real Estate
long-only 0.00975 0.0201 0.0102 0.0228 0.0674∗ 0.0394
(1.40) (1.44) (1.41) (1.65) (0.04) (1.89)
long-short -0.00126 0.00185 -0.000859 0.00273 -0.00157 0.0181
(-0.16) (0.17) (-0.10) (0.23) (-0.16) (0.97)
Communication Services
long-only 0.00000759 0.00159 0.000303 0.00372 0.00117 0.00387
(0.00) (0.26) (0.06) (0.61) (0.22) (0.60)
long-short -0.00875 -0.00370 -0.00885 -0.00511 -0.00959 -0.0117
(-1.40) (-0.33) (-1.38) (-0.53) (-1.30) (-1.09)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
26
4.2.2 Equally-Weighted Portfolios
In Table 6, the results of alpha for equally-weighted, long-only and long-short portfo-
lios, are presented for the positive and negative screening strategies for the full-sample
period. The long-only portfolio shows positive and significant returns for all regres-
sions. For the negative screening strategy the results are significant at a 0.1% level,
and for the positive screening strategy the results are significant at a 1% level, apart
from Fama-French which is significant at a 0.1% level. The long-short portfolio for the
positive screening strategy generates negative and significant alphas at a 0.1% level for
all models. The opposite can be seen for the negative screening where the alphas for
all regressions are positive, and not statistically significant.
Table 6: Positive and negative of long-only and long-short portfolios (equally-
weighted)
CAPM Fama-French 3 Carhart
Positive Negative Positive Negative Positive Negative
long-only 0.0111∗∗ 0.0111∗∗∗ 0.0117∗∗∗ 0.0116∗∗∗ 0.0132∗∗ 0.0130∗∗∗
(3.32) (3.84) (3.45) (3.99) (3.32) (3.74)
long-short -0.0315∗∗∗ 0.000227 -0.0300∗∗∗ 0.0000860 -0.0295∗∗∗ 0.00234
(-6.34) (0.06) (-6.29) (0.02) (-5.73) (0.41)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
The equally-weighted portfolio in Table 7 generates significant results for all models.
All models for the long-only portfolios generate significant results for the Financial,
Consumer Staple, Health Care, Industrial and Real Estate sectors. The strongest alpha
is generated with Fama-French for the Real Estate sector. The long-short portfolio gen-
erates only significant results with CAPM and Fama-French for the Consumer Staple
sector at a 5% significance level.
27
Table 7: Best-in-class of long-only and long-short portfolios (equally-weighted)
CAPM
Basic Material Financials Consumer Staples Consumer Discretionary
long-only 0.00518 0.00701∗ 0.0132∗∗ 0.00475
(1.23) (2.20) (3.14) (1.25)
long-short -0.00351 0.00165 0.0119∗ -0.00489
(-0.62) (0.45) (2.40) (-0.66)
Health Care Industrial Real Estate Communications
long-only 0.0988∗ 0.00883∗∗ 0.0169∗ -0.00103
(2.05) (2.67) (3.84) (-0.21)
long-short -0.00260 0.000914 0.00498 -0.0111
(-0.32) (0.30) (0.65) (-1.65)
Fama-French 3
Basic Material Financials Consumer Staples Consumer Discretionary
long-only 0.00549 0.00747∗ 0.0135∗∗ 0.00545
(1.31) (2.29) (3.21) (1.35)
long-short -0.00323 0.00172 0.0123∗ -0.00369
(-0.59) (0.45) (2.45) (-0.51)
Health Care Industrial Real Estate Communications
long-only 0.00995∗ 0.00907∗∗ 0.0177∗ -0.000971
(2.05) (2.79) (2.42) (-0.19)
long-short -0.00239 0.0009311 0.00566 -0.0113
(-0.28) (0.32) (0.73) (-1.71)
Carhart
Basic Material Financials Consumer Staples Consumer Discretionary
long-only 0.00928 0.0116∗∗ 0.0126∗ 0.00687
(1.76) (3.08) (2.50) (1.40)
long-short -0.00108 0.0103 0.0103 0.00142
(-0.15) (1.03) (1.50) (0.17)
Health Care Industrial Real Estate Communications
long-only 0.0128∗ 0.00967∗ 0.0171∗ 0.000664
(2.37) (2.44) (2.03) (0.13)
long-short 0.0000113 0.000202 0.0107 -0.0123
(0.00) (0.07) (1.18) (-1.60)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
28
5 Analysis
5.1 Screening Strategies
All regressions for the positive and negative screening strategies for the long-short port-
folio shows different results for the value-weighted portfolio, but none of the alphas
are statistically significant, and therefore no conclusions can be drawn with respect to
the research question. This finding is in line with earlier research such as Kempf and
Osthoff (2007), Derwall et al. (2007), Halbritter and Dorfleitner (2015), Hartzmark
and Sussman (2019), Limkriangkrai, Koh and Durand (2015), and Diaz et al. (2020),
who also obtained similar results, with no significant alphas for the long-short portfo-
lio. Moreover, the negative screening strategy in Table 2 shows negative returns for
all long-short regressions, where Carhart generates the highest return annually when
analyzing the alphas. This result is in line with Kempf and Osthoff (2007), as they also
obtain a negative alpha from the negative screening strategy, which was not statistically
significant.
The long-only portfolio generates positive and significant returns for all regressions
and for the positive and negative screening strategy, in stark contrast to the long-short
portfolio. The result by Carhart for the positive screening is in line with Kempf and
Osthoff (2007) report who also obtained a significant alpha, but this thesis generates
stronger alpha in comparison to them since it is significant at a 0.1% level compared to
a 10% level as Kempf and Osthoff. The result of the negative screening strategy with
Carhart is not in line with Kempf and Osthoff, where no significant results is generated
in their report. The findings of this thesis suggests that there could exist a premium on
the Stockholm stock exchange market, which a potential investor could exploit. More-
over, Carhart generates the strongest and highest alpha, suggesting that a momentum
strategy is rewarding. The findings for the positive and negative screening strategy are
in contrast to the efficient market hypothesis which argues that it should not be possible
to gain advantage by excluding controversial or low-rated environment stock.
The best-in-class screening strategy (Table 3) generates no significant results for the
three models when applying the long-short portfolio, continuing the poor performance
of the value-weighted, long-short portfolios. With long-only portfolios, there are two
industries, Industrial and Health Care, showing significant and positive results for all
models. The strongest return is generated for the Industrial sector throughout all mod-
els. The alpha increases as more factors are included in the model, suggesting that there
29
is a strong momentum factor with the Carhart model, and for the investor to consider.
This result is expected since this is one of the three sectors that has performed best on
the Stockholm stock market between 2010-2020. The finding assert that there is an
abnormal positive return for the Industrial sector when applying a best-in-class strategy
for the highest rated Environmental stocks. For the investor it will generate abnormal
returns of 0.0108% monthly when applying the momentum strategy on the Industrial
sector. This result is not in line with Diaz et al. (2021) who generate a monthly return
of 0.167%, but is not statistically significant. Therefore, it is hard to tell if this high re-
turn is generated because of high Environmental rating, of the underlying sector return
on the Stockholm stock market, or if it is because of unknown factors which are not
included in the models.
The Consumer Staple and Real Estate sectors generate significant and positive results
with CAPM and Fama-French, but not for Carhart. The most likely explanation for this
finding is that the momentum factor does not work for these industries where one buy
winners and sell losers. Another interesting result in Table 3 is that the Financial sector
does not generate significant result for CAPM and Fama-French but for Carhart. This
result could be randomized, but it also suggests that the momentum factor works strong
for the Financial sector.
A shareholder approach, where all prices are incorporated in accordance with EMH,
no abnormal performance should be possible to acquire, does seem to hold when an-
alyzing the sample for the long-short portfolios. When expanding the analysis to the
long-only portfolio the results are more aligned with significant and positive results
for both screening strategies in Table 2. It could be that the efficient market hypoth-
esis is not as strong in the sample as for the long-short, and other factors that are not
included which can significantly affect the results. A stakeholder approach, taken by
fund mangers in Van Duuren et al. (2016) report, does appear to generate abnormal
positive returns for this thesis in some part, as Fama-French (1993) show when apply-
ing a factor based model where he concludes that it could.
A selective best-in-class screening approach, where the investor take stakeholders and
environmental rating into account, does generate abnormal performance on the SSE as
the results from the the Industrial sector clearly demonstrates. The results from the
three screenings strategies generate mixed results, in contrast to Kempf and Osthoff
(2007), which demonstrated that it is possible to achieve abnormal performance. The
30
only sector which generates positive returns for all three models for the long-short port-
folios is the Consumer Staples sector, but as the results are not statistically significant,
no conclusions can be drawn.
The market does not appear to premier the stocks with high rating as it on aggregate
does not generate a positive abnormal alpha for the same long-only and long-short
portfolios for the best-in-class screening strategy. The findings contradict the previous
studies on the European market as Oesterich and Tsiakas (2015) argue that there is a
premium, though on the German market. The finding of this thesis could be due to
randomized effects, but also due to the the fact that risk should rewarded with higher
return. As Kaiser (2020) argues, a low Environmental rating is equal to higher risk, it is
not surprising that is premiered by the market by higher returns. The shareholder theory
argues that the only criterion for the executive should be to generate the highest returns
for the shareholder, whilst the stakeholder theory argue that a more holistic approach
where other interest play a role in the firm should be considered and more valued. The
stakeholder theory does not explain the results in the thesis as it can be argued that the
shareholder theory is more aligned with the results; that the executives should focus on
the value of the firm in the best interest of the shareholders. It is possible that there is
a lack if interest in high Environmental rating from the executive managing the low-
rated firms and that they focus more on generating shareholder return, but this is not a
research question for this thesis.
5.2 Alternative Portfolio Construction
5.2.1 Sub-Periods
When dividing the sample in two periods (Table 4), to evaluate if the implementation
of the new sustainability law had any effect on the returns, the results show that the pe-
riod before the implementation generate more significant results. The patterns holds for
both the long-only, and for the long-short portfolios for the negative screening strategy.
It appears to be an anomaly in the market before the implementation, suggesting that
an investor can generate excess return. The new law indicates that the Environmental
rating is incorporated in the price after the implementation for the positive screening,
long-only portfolios. The Efficient Market theory especially holds with CAPM, for
both positive and negative screening, and with Fama-French for the positive screening.
However, with Carhart the results are significant before and after the law implementa-
tion with both screening strategies. This indicates that buying high-rated winner stocks
31
is a winning concept to generate excess return even after the law implementation.
In a sense, the regulation has worked for the long-short strategy (Table 4) as it has
removed the inefficiency which is in line with the Efficient Market Hypothesis but not
with the research question. The returns after the law was implemented deviate from the
full-sample period analysis, that a trading strategy based on the negative and positive
portfolio generate abnormal returns. A possible explanation can be, as Wang and Sar-
gis (2020) argue, that larger firms tend to invest more in achieving a good rating. As
the sample size between 2010-2016 is significantly smaller then the 2017-2020 sample,
the results could have been affected by this relationship. The smaller size of the sample
between 2010-2016 most likely affected the results for the whole sample period. As
the sample becomes larger, the alpha drifts closer to zero, demonstrating the law of
large numbers importance. As future sampling will include more data, this deviation
will disappear and continue to drift closer to zero. Comparing the results with Kempf
and Osthoff (2007), who divided the data in two periods, this thesis results is similar to
theirs. The alphas does deviate, as expected considering that the period is divided.
The same approach (Table 5) as for the positive and negative screening strategies is
conducted for long-only and long-short portfolios, where the results are more spread
for industries and within industries. There are only six of 96 alphas which generate
significant results when dividing the sample in two periods. For the long-only strat-
egy the Financial sector is the only sector generating significant results for all models
before the law implementation. Since none of the other sectors generate continuously
significant results this result is most likely randomized. In general, the results for the
long-short are in line with the full-sample period which strengthens the results of the
thesis. For the long-only the results differ when dividing them in two periods as the
strength is not the same as in the results as for the full sample.
5.2.2 Equally-Weighted Portfolios
Table 6 reports the alphas of the long-only and the long-short portfolios for positive
and negative screening strategies of equally-weighted portfolios for the full-period. In
general, the results of alpha for the equally-weighted portfolios is stronger than for the
value-weighted portfolios which is not unsurprising since smaller stocks have higher
weight in equal-weighted portfolios than for value-weighted portfolios. However, for
Carhart the returns are less for equal-weighted portfolios with both screening strategies,
than for value-weighted. This is most likely because stocks with larger market cap have
32
performed better. For the long-only portfolios, the equally-weighted portfolios follow
the same pattern as for the value-weighted portfolios where all returns are statistically
significant for both the positive and negative screening strategies.
For the long-short portfolios with the negative screening strategy, the alphas are posi-
tive and not significant in contrast to the value-weighted portfolios, where the alphas are
negative and significant. One explanation for this observation could be that the exclu-
sion of controversial stocks are less for smaller stocks, and since those smaller stocks
have higher weight in the equal-weighted portfolio the alphas turns positive. With
the positive long-short screening strategy, the alphas are negative for CAPM, Fama-
French and Carhart, which is in general in line with the value-weighted portfolios of
these models. With the equal-weighted portfolio with Carhart the alpha is still negative
whereas for the value-weighted Carhart model alpha is positive. It could be that the
momentum factor is stronger for small stocks than for large stocks and can possibly
explain the differences. Noticeably is that the all alphas for all models are significant
at a 0.1% level when forming equally-weighted portfolios for the positive screening
strategy. The results of the equally-weighted portfolios is therefore in general better
than for value-weighted portfolios. Therefore, it could be argued afterwards that the
equally-weighted, long-short portfolios should have been part of the main results of the
thesis. However, this thesis follows the outline of Kempf and Osthoff (2007) where
value-weighted portfolios is constructed for the main results and the equally-weighted
portfolios is constructed as a sensitivity test of the weighting of stocks.
Best-in-class (Table 7) follow the expected path for the long-short, as the returns be-
comes stronger when applying the stock returns for equal-weighted portfolios. The best
result for the long-short portfolio is with the Consumer Staples sector, which generates
significant alphas at a 5% level with CAPM and Fama-French. This is also the only
sector which is statistically significant with the best-in-class screening strategy for the
long-short, equally-weighted portfolios. The same does not hold for the long-only as
the results are more scattered but with the same degree of significance. The Industrial
and Health Care sector continues to generate significant results for all models. There
appears to be a sensitivity with respect to the weighting scheme for the portfolio when
applying the long-only, best-in-class strategy. The phenomenon is surprising consid-
ering that the norm is for the equally-weighted to generate higher return than for the
value-weighted. The most possible explanation for this is the choice of only including
the mid cap and large cap companies, where the larger companies tend to outperform
33
the smaller companies. The surprising results from the change in weighting scheme
contradicts both Kempf and Osthoff (2007) and Halbritter and Dorfleitner (2005) find-
ings.
6 Conclusion
This thesis contributes to the literature by deepening the knowledge about the possibil-
ity to invest in Environmental high-rated stocks listed on the SSE, a knowledge which
is useful to both the academia and the investor community, as no study on this topic
have been conducted previously. An amateur investor which preferences are to invest
in Environmental high-rated stocks listed on the SSE can invest in a a selective strategy
where the investor go long-only, a strategy which requires low knowledge, is easy to
execute, and has a low transaction cost. The more sophisticated and professional in-
vestor can add the long-short strategy with a momentum factor to add extra alpha to
their portfolio.
In general across the regressions, the Carhart model generates the best returns. The
long-only strategy generates the best results for the potential investor whom wishes
to invest in Environmental high-rated stocks. The results are both statistically signifi-
cant and positive for all three regressions with Carhart as the highest return (0.0156%)
is achieved by the negative screening. The phenomenon holds when changing from
value-weighting to equally-weighting as the results are still positive and significant,
strengthening the results. The best-in-class screening strategy generates mixed results
across the board but statistically significant results for the Industrial sector for both the
value-weighted and equally-weighted portfolios.
The best significant result for the long-short was generated for the positive screen-
ing strategy for equally-weighted portfolios where the alphas are significant at a 0.1%
level for all models. The alphas are negative indicating that the strategy does not gen-
erate a positive abnormal performance and therefore contradicts the research question
of the thesis. This result was not in line with Kempf and Osthoff (2007) as their results
of the equally-weighted portfolios were similar to the value-weighted portfolios. The
best-in-class screening strategy generates few significant results when dividing them in
sub-periods, and no conclusion can be drawn from the results in general for either the
long-only or long-short strategies.
34
A possible avenue for further research is to examine if the results from this thesis
hold if the rating agency changes. Several other databases provide ESG rating, such
as Bloomberg, SRI, and MSCI, and including those will provide reliability for this the-
sis. Secondly, is to investigate how the sustainability law implementation in 2017 affect
the results, and rating, long-term. As more data is needed to verify the results, the au-
thors of this thesis does not believe that it is possible to examine this in the near-term.
Finally, one could broaden the scope and include more stock markets, i.e., the Nordics,
and evaluate if the results hold as it increases the data.
The mixed results for the thesis contradicts Kempf and Osthoff (2007), and is more
aligned with Nofsinger and Varma (2014), Hartzmark and Sussman (2014) and Halbrit-
ter and Dorfleinter (2015); that the results are mixed depending on strategy, model,
and portfolio construction. The potential investor do wisely to invest in a strategy in
the positive and negative screening, both for the value weighted and equally-weighted
portfolio strategy, if the investor choose a long-only approach.
35
References
A Friedman doctrine - The Social Responsibility Of Business Is to Increase Its Profits
(n.d.).
URL: https://www.nytimes.com/1970/09/13/archives/a-friedman-doctrine-the-
social-responsibility-of-business-is-to.html
AQR (n.d.), ‘Bab’.
URL: https://www.aqr.com/Insights/Datasets/Betting-Against-
Beta-Equity-Factors-Monthly?fbclid=IwAR3Ovxhumz1vqQ7tVKcfY-
aRMl4Q1i2wY3lmR2JO5gR0htFBFfSLbz0OztQ
Berman, L., Wicks, C., Kotha, S. & Jones, M. (2017), ‘Does stakeholder orientation
matter? the relationship between stakeholder management models and firm financial
performance’, Academy of Management Journal 42(5).
Blackrock (n.d.a), ‘Esg integration’.
URL: https://www.blackrock.com/us/individual/investment-ideas/sustainable-
investing/esg-integration
Blackrock (n.d.b), ‘Investing in energy transition’.
URL: https://www.blackrock.com/us/individual/insights/energy-transition-investing
Bodie, Z., Kane, A. & Marcus, A. (2021), Investments.
Carhart, M. M. (1997), ‘On persistence in mutual fund performance’, The Journal of
finance 52(1), 57–82.
Clarkson, B. M. (n.d.), ‘A stakeholder framework for analyzing and evaluating corpo-
rate social performance’, The Academy of Management Review 20(1), 92–117.
Derwall, J., Guenster, N., Bauer, R. & Koedijk, K. (2005), ‘The eco-efficiency premium
puzzle’, Financial Analysts Journal 61(2), 51–63.
Diaz, V., Ibrushi, D. & Zhao, J. (2021), ‘Reconsidering systematic factors during
the covid-19 pandemic–the rising importance of esg’, Finance Research Letters
38, 101870.
Fama, E. F. & French, K. R. (1993), ‘Common risk factors in the returns on stocks and
bonds’, Journal of financial economics 33(1), 3–56.
Freeman, R. (1984), Strategic Management: A Stakeholder Approach.
36
Friedman, M. (1962), Capitalism and Freedom, University of Chicago Press.
Friedman, M. & Friedman, R. (1980), Free To Choose: A Personal Statement, Harcourt.
Government, S. (n.d.), ‘Swedish government’.
URL: https://www.riksdagen.se/en/
Halbritter, G. & Dorfleitner, G. (2015), ‘The wages of social responsibility—where are
they? a critical review of esg investing’, Review of Financial Economics 26, 25–35.
Hartzmark, S. & Sussman, A. B. (2014), ‘Do investors value sustainability? a natural
experiment examining ranking and fund flows’, The Journal of Finance 76(6), 2789–
2837.
Hong, H. & Kacperczyk, M. (2007), ‘The price of sin: The effects of social norms on
markets’, Journal of Financial Economics 93(1), 15–36.
Kaiser, L. (2020), ‘Esg integration: value, growth and momentum’, Journal of Asset
Management 21(1), 32–51.
Kempf, A. & Osthoff, P. (2007), ‘The effect of socially responsible investing on port-
folio performance’, European financial management 13(5), 908–922.
Limkriangkrai, M., Koh, S. & Durand, R. B. (2017), ‘Environmental, social, and gov-
ernance (esg) profiles, stock returns, and financial policy: Australian evidence’, In-
ternational Review of Finance 17(3), 461–471.
Market share of selected grocery retailers in Sweden (n.d.).
URL: https://www.statista.com/statistics/565650/market-share-of-selected-grocery-
retailers-in-sweden/
Miles, S. (2012), ‘Stakeholder: Essentially contested or just confused?’, Journal of
Business Ethics 108, 185–198.
MSCI (n.d.).
URL: https://www.msci.com/documents/1296102/11185224/GICS+Methodology+2020.pdf/9caadd09-
790d-3d60-455b-2a1ed5d1e48c?t=1578405935658
Nofsinger, J. & Varma, A. (2014), ‘Socially responsible funds and market crises’, Jour-
nal of Banking Finance 48(1), 180–193.
OECD (2021), ‘Esg investing and climate transition’.
37
Oestreich, A. M. & Tsiakas, I. (2015), ‘Carbon emissions and stock returns: Evidence
from the eu emissions trading scheme’, Journal of Banking Finance 58(1), 294–308.
Perold, A. F. (2004), ‘The capital asset pricing model’, Journal of economic perspec-
tives 18(3), 3–24.
Plyakha, Y., Uppal, R. & Vilkov, G. (2021), Financial Risk Management and Modeling:
Chapter Equal or Value Weighting? Implications for Asset-Pricing Tests, Springer
Cham, Editors: Zopounidis, Constantin and Benkraiem, Ramzi and Kalaitzoglou,
Iordanis.
Refinitiv (2020), ‘Governance (esg) scores from refinitiv’.
Refinitiv Eikon (n.d.).
URL: https://eikon.thomsonreuters.com/index.html
Rozeff, M. S. & Zaman, M. A. (1988), ‘Market efficiency and insider trading: New
evidence’, Journal of Business pp. 25–44.
Sargis, M. & Wang, P. (n.d.), ‘Better minus worse’,
https://www.morningstar.com/insights/2020/02/19/esg-companies, .
Schwab, K. (1971), Modern Company Management in Mechanical Engineering.
SHoF (n.d.), ‘House of finance’.
URL: https://www.hhs.se/en/houseoffinance/
Smith, A. (1776-1790), An Inquiry into the Nature and Causes of the Wealth of Nation,
W. Strahan and T. Cadell,.
Sweden Exports (n.d.).
URL: https://tradingeconomics.com/commodity
Weis, C., Woltering, R.-O. & Sebastian, S. (2021), ‘Which stocks are driven by
which interest rates? evidence from listed real estate’, Journal of property research
38(3), 175–197.
Who Cares Wins (n.d.).
URL: https://www.unepfi.org/fileadmin/events/2004/stocks/whocareswinsglobalcompact2004.pdf
38
A Appendix
A.1 Screening Strategies
Table 8: Positive screening
Alpha MKT SMB HML UMD
CAPM
Equal Weighted
high-rated 0.0111∗∗ 0.631∗∗∗
(3.32) (12.96)
low-rated 0.0115∗∗ 0.684∗∗∗
(3.33) (12.10)
long-short -0.0315∗∗∗ 0.507∗∗∗
(-6.34) (-0.81)
Value Weighted
high-rated 0.0112∗∗∗ 0.451∗∗∗
(3.88) (4.33)
low-rated 0.0131∗∗∗ 0.642∗∗∗
(4.33) (13.02)
long-short -0.00230 -0.190∗∗∗
(-0.81) (-4.42)
Fama-French 3
Equal Weighted
high-rated 0.0117∗∗∗ 0.621∗∗∗ -0.135 0.219
(3.45) (12.60) (-1.54) (1.77)
low-rated 0.0114∗∗ 0.684∗∗∗ 0.0591 0.00302
(3.25) (12.22) (0.51) (0.02)
long-short -0.0300∗∗∗ 0.479∗∗∗ -0.113 0.570∗
(-6.29) (5.24) (-0.81) (2.45)
Value Weighted
high-rated 0.0117∗∗∗ 0.446∗∗∗ -0.228∗∗ 0.131
(4.10) (10.74) (-2.88) (1.40)
low-rated 0.0132∗∗∗ 0.642∗∗∗ -0.176 0.0298
(4.50) (12.54) (-1.58) (0.25)
long-short -0.00202 -0.194∗∗∗ -0.0517 0.0995
(-0.70) (-4.58) (-0.59) (0.96)
Carhartl
Equal Weighted
high-rated 0.0132∗∗ 0.614∗∗∗ -0.143 0.208 -0.103
(3.32) (11.91) (-1.67) (1.68) (-0.68)
low-rated 0.0122∗∗ 0.680∗∗∗ 0.0551 -0.00242 -0.0542
(3.14) (12.34) (0.48) (-0.02) (-0.49)
long-short -0.0295∗∗∗ 0.477∗∗∗ -0.116 0.566∗ -0.0382
(-5.73) (5.08) (-0.82) (2.38) (-0.22)
Value Weighted
high-rated 0.0143∗∗∗ 0.434∗∗∗ -0.242∗∗ 0.112 -0.185
(4.38) (11.09) (-3.33) (1.27) (-1.93)
low-rated 0.0131∗∗∗ 0.642∗∗∗ -0.175 0.0311 0.0127
(4.72) (12.08) (-1.62) (0.25) (0.09)
long-short 0.000807 -0.207∗∗∗ -0.0662 0.0798∗ -0.196
(0.26) (-4.73) (-0.78) (0.76) (-1.93)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
39
Table 9: Negative screening
Alpha MKT SMB HML UMD
CAPM
Equal Weighted
high-rated 0.0111∗∗∗ 0.603∗∗∗
(3.84) (11.62)
controversial 0.0118∗∗ 0.630∗∗∗
(2.71) (11.15)
long-short 0.000227 0.0280
(0.06) (0.46)
Value Weighted
high-rated 0.0134∗∗∗ 0.542∗∗∗
(4.44) (13.65)
controversial 0.00873∗∗ 0.492∗∗∗
(2.74) (9.34)
long-short -0.00509 -0.0499
(-1.52) (-0.94)
Fama-French 3
Equal Weighted
high-rated 0.0116∗∗∗ 0.595∗∗∗ -0.0298 0.171
(3.99) (11.74) (-0.35) (1.42)
low-rated 0.0121∗∗ 0.624∗∗∗ -0.00529 0.122
(2.73) (10.29) (-0.04) (0.73)
long-short 0.0000860 0.0303 0.0252 -0.0504
(0.02) (0.47) (0.21) (-0.31)
Value Weighted
high-rated 0.0138∗∗∗ 0.538∗∗∗ -0.198∗ 0.131
(4.63) (13.56) (-2.29) (1.46)
low-rated 0.00921∗∗ 0.487∗∗∗ -0.235∗ 0.132
(2.99) (8.71) (-2.27) (0.93)
long-short -0.00506 -0.0496 -0.0363 0.00000645
(-1.50) (-0.91) (-0.35) (0.00)
Carhart
Equal Weighted
high-rated 0.0130∗∗∗ 0.588∗∗∗ -0.0373 0.161 -0.102
(3.74) (11.35) (-0.45) (1.36) (-0.98)
low-rated 0.0159∗∗ 0.607∗∗∗ -0.0244 0.0957 -0.260
(2.65) (10.65) (-0.19) (0.57) (-1.50)
long-short 0.00234 0.0200 0.0137 -0.0661 -0.157
(0.41) (0.30) (0.12) (-0.40) (-0.81)
Value Weighted
high-rated 0.0156∗∗∗ 0.529∗∗∗ -0.207∗ 0.118 -0.127
(4.74) (13.64) (-2.57) (1.37) (-1.12)
low-rated 0.0102∗∗ 0.482∗∗∗ -0.240∗ 0.125 -0.0708
(2.64) (8.96) (-2.39) (0.85) (-0.54)
long-short -0.005900 -0.0458 -0.0320 0.00583 0.0580
(-1.38) (-0.85) (-0.32) (0.04) (0.36)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
40
Table 10: Best-in-class screening of long-short portfolios (CAPM)
Equal weighted Value Weighted
Alpha MKT Alpha MKT
Basic Material
Long-Short -0.00351 0.434∗∗∗ 0.0000704 -0.190
(-0.62) (4.04) (0.00) (-0.59)
Consumer Discretionary
long-short -0.00489 -0.321∗ -0.00981 -0.749∗∗∗
(-0.66) (-2.21) (-1.11) (-4.52)
Consumer Staples
long-short 0.0119∗ 0.128 0.0132 0.0593
(2.40) (1.82) (1.71) (0.61)
Energy
long-short -0.0141 0.832∗∗∗ -0.0140 -0.866∗∗∗
(-1.70) (-4.12) (-1.61) (-4.11)
Financials
long-short 0.00165 -0.0150 -0.00412 0.0203
(0.45) (-0.26) (-1.05) (0.31)
Health Care
long-short -0.00260 -0.115 -0.00828 -0.361∗
(-0.32) (-0.69) (-0.95) (-2.10)
Industrial
long-short 0.000914 -0.0133 -0.00241 -0.182∗
(0.30) (-0.22) (-0.56) (-2.26)
Real Estate
long-short 0.00498 -0.0628 -0.000371 0.135
(0.65) (-0.46) (-0.06) (1.10)
Technology
long-short -0.0183 -0.451∗ -0.0610 -1.240∗∗
(-1.22) (-2.57) (-1.20) (-2.84)
Communications
long-short -0.0111 -0.0325 -0.00753 -0.103
(-1.65) (-0.23) (-1.25) (-0.62)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
41
Table 11: Best-in-class screening with high and low rated portfolios (CAPM)
Equal weighted Value Weighted
Alpha MKT Alpha MKT
Basic Material
high-rated 0.00518 0.737∗∗∗ 0.00636 0.724∗∗∗
(1.23) (10.72) (1.42) (9.37)
low-rated 0.0140∗ 0.746∗∗∗ 0.00647 2.284∗∗
(2.16) (6.26) (0.14) (2.72)
Consumer Discretionary
high-rated 0.00475 0.674∗∗∗ 0.00372 0.326∗∗∗
(1.25) (7.34) (0.60) (4.26)
low-rated 0.00849 0.997∗∗∗ 0.0129 1.070∗∗∗
(1.25) (7.34) (1.85) (7.44)
Consumer Staples
high-rated 0.0132∗∗ 0.150∗ 0.0155∗ 0.0917
(3.14) (2.44) (2.11) (0.99)
low-rated 0.00627 0.218 0.0119 0.332
(0.49) (0.71) (1.10) (1.45)
Energy
high-rated N/A N/A N/A N/A
(N/A) (N/A) (N/A) (N/A)
low-rated 0.0135 0.833∗∗∗ 0.0131 0.868∗∗∗
(1.62) (4.12) (1.51) (4.11)
Financials
high-rated 0.00701∗ 0.716∗∗∗ 0.00492 0.728∗∗∗
(2.20) (12.97) (1.31) (10.63)
low-rated 0.00577 0.730∗∗∗ 0.00857∗∗ 0.708∗∗∗
(1.92) (14.46) (2.63) (11.91)
Health Care
high-rated 0.00988∗ 0.293∗∗∗ 0.00872∗ 0.120
(2.05) (3.55) (2.10) (1.96)
low-rated 0.0106 0.413∗ 0.0165 0.482∗∗
(1.39) (2.61) (1.94) (2.89)
Industrial
high-rated 0.00883∗∗ 0.675∗∗∗ 0.00969∗∗ 0.646∗∗∗
(2.67) (13.16) (2.82) (12.43)
low-rated 0.00914 -0.0133 0.0116∗ 0.829∗∗∗
(0.30) (-0.22) (2.34) (9.81)
Real Estate
high-rated 0.0169∗ 0.513∗∗ 0.0133∗ 0.802∗∗∗
(3.84) (11.62) (1.99) (5.72)
low-rated 0.0115 0.576∗∗ 0.0134∗ 0.670∗∗∗
(1.92) (5.52) (2.19) (6.33)
Technology
high-rated 0.0125 0.504∗∗ 0.00530 0.382∗∗∗
(1.34) (3.32) (1.20) (4.78)
low-rated 0.0291∗ 0.959∗∗∗ 0.0651 1.542∗∗∗
(2.16) (6.66) (1.29) (3.55)
Communications
high-rated -0.00103 0.463∗∗∗ 0.000250 0.381∗∗∗
(-0.21) (5.97) (0.06) (4.93)
low-rated 0.00907 0.498∗∗∗ 0.00732 0.485∗∗
(1.74) (4.08) (1.26) (3.26)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
42
Table 12: Best-in-class screening of long-short portfolios (Fama-French 3)
Equal weighted Value Weighted
Alpha MKT SMB HML Alpha MKT SMB HML
Basic Material
long-short -0.00323 0.434∗∗∗ -0.2666 0.0471 -0.00292 -0.161 1.464 -0.834
(-0.59) (4.33) (-1.34) (0.18) (-0.12) (-0.50) (1.70) (-0.97)
Consumer Discretionary
long-short -0.00369 -0.332∗ -0.592∗ 0.333 -0.00894 -0.748∗∗∗ -0.890∗∗ 0.125
(-0.51) (-2.39) (-2.44) (1.14) (-1.02) (-4.66) (-2.98) (0.39)
Consumer Staples
long-short 0.0123∗ 0.123 -0.147 0.131 0.0134 0.0585 -0.141 0.0404
(2.45) (1.75) (-0.97) (0.69) (1.70) (0.61) (-0.79) (0.18)
Energy
long-short -0.0141 -0.842∗∗∗ 0.481 0.124 -0.0136 -0.886∗∗∗ 0.548 0.279
(-1.75) (-4.22) (1.57) (0.29) (-1.63) (-4.23) (1.67) (0.60)
Financials
long-short 0.00172 -0.0152 -0.0533 0.0137 -0.00417 0.0201 -0.0537 -0.00493
(0.45) (-0.26) (-0.41) (0.10) (-1.04) (0.30) (0.43) (-0.04)
Health Care
long-short -0.00239 -0.116 -0.141 0.0482 -0.00795 -0.365∗ 0.132 0.101
(-0.28) (-0.70) (-0.65) (0.11) (-0.88) (-2.15) (-0.56) (0.24)
Industrial
long-short 0.000931 -0.00781 -0.279∗ -0.0627 -0.00221 -0.180∗ -0.309 0.00345
(0.32) (-0.14) (-2.51) (-0.48) (-0.52) (-2.45) (-1.98) (0.02)
Real Estate
long-short 0.00566∗ -0.0739 -0.124 0.242 0.000287 0.126 -0.221 0.210
(0.73) (-0.55) (-0.49) (0.80) (0.04) (1.02) (-1.03) (0.82)
Technology
long-short -0.0154 -0.500∗∗ -0.374 1.046 -0.0647 -1.139∗∗ -1.272 -1.816
(-1.05) (-2.66) (-0.89) (1.75) (-1.22) (-2.79) (-1.90) (-1.45)
Communications
long-short -0.0113 -0.0221 -0.246 -0.167 -0.00789 -0.0920 -0.203 0.193
(-1.71) (-0.16) (-1.09) (-0.54) (-1.39) (-0.59) (-0.99) (0.54)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
43
Table 13: Best-in-class screening with high and low rated portfolios (Fama-French 3)
Equal weighted Value Weighted
Alpha MKT SMB HML Alpha MKT SMB HML
Basic Material
high-rated 0.00549 0.734∗∗∗ -0.116 0.0949 0.00623 0.732∗∗∗ -0.267 -0.117
(1.31) (10.57) (-0.79) (0.60) (1.41) (9.47) (-1.78) (0.66)
low-rated 0.0136∗ 0.747∗∗∗ 0.0795 -0.0646 0.0193 2.337∗∗ -3.255∗ 0.878
(2.04) (6.42) (0.38) (-0.24) (0.44) (2.70) (-2.10) (0.62)
Consumer Discretionary
high-rated 0.00545 0.666∗∗∗ -0.305 0.204 0.00422 0.340∗∗∗ -0.761∗∗∗ -0.0482
(1.35) (9.39) (-2.30) (1.10) (0.70) (4.67) (-4.95) (-0.24)
low-rated 0.00801 1.00∗∗∗ 0.295 -0.115 0.0124 1.077∗∗∗ 0.146 -0.158
(1.16) (7.48) (1.28) (-0.43) (1.74) (7.43) (0.59) (-0.57)
Consumer Staples
high-rated 0.0135∗∗ 0.151∗ -0.318 0.0245 0.0156∗ 0.0968 -0.291 -0.0533
(3.21) (2.58) (-2.38) (0.14) (2.09) (1.04) (-1.70) (-0.24)
low-rated -0.00126 -0.119 -1.511∗ -1.877∗ 0.00611 0.0718 -1.174∗∗ -1.424∗
(-0.10) (-0.38) (-2.74) (-2.36) (0.66) (0.36) (-3.32) (-2.44)
Energy
high-rated N/A N/A N/A N/A N/A N/A N/A N/A
(N/A) (N/A) (N/A) (N/A) (N/A) (N/A) (N/A) (N/A)
low-rated 0.0134 0.844∗∗ -0.483 -0.132 0.0127 0.888∗∗∗ -0.546 -0.282
(1.67) (4.22) (-1.58) (-0.31) (1.52) (4.24) (-1.67) (-0.61)
Financials
high-rated 0.00747∗ 0.707∗∗∗ 0.00154 0.185 0.00543 0.716∗∗∗ 0.0500 0.218
(2.29) (13.21) (0.01) (1.65) (1.46) (11.22) (0.42) (1.72)
low-rated 0.00612∗ 0.722∗∗∗ 0.0463 0.151 0.00913∗∗ 0.697∗∗∗ -0.00309 0.222
(2.04) (15.34) (0.48) (1.35) (2.83) (12.18) (-0.03) (1.72)
Health Care
high-rated 0.00995∗ 0.297∗∗ -0.256 -0.0376 0.00899∗ 0.118∗∗∗ -0.160 0.0689
(2.05) (3.36) (-1.71) (-0.19) (2.14) (1.87) (-1.31) (0.44)
low-rated 0.0106 0.416∗ -0.100 -0.0575 0.0165 0.484∗∗ -0.0278 -0.0331
(1.32) (2.64) (-0.53) (-0.15) (1.84) (2.93) (-0.13) (-0.09)
Industrial
high-rated 0.00907∗∗ 0.674∗∗∗ -0.205 0.0455 0.0100∗∗ 0.644∗∗∗ -0.242 0.0791
(2.79) (12.84) (-1.65) (0.33) (2.99) (12.57) (-1.92) (0.66)
low-rated 0.000931 -0.00781 -0.279∗ -0.0627 0.0118∗ 0.824∗∗∗ 0.0671 0.0743
(0.32) (-0.14) (-2.51) (-0.48) (2.34) (10.05) (0.40) (0.34)
Real Estate
high-rated 0.0177∗ 0.488∗∗ 0.383 0.0421 0.0144∗∗ 0.775∗∗∗ 0.298 0.506
(2.42) (3.20) (1.84) (1.33) (2.11) (5.68) (1.40) (1.85)
low-rated 0.0117 0.563∗∗∗ 0.506∗ 0.175 0.0140∗ 0.651∗∗∗ 0.525∗ 0.302
(1.97) (5.77) (2.04) (0.77) (2.29) (6.73) (2.38) (1.23)
Technology
high-rated 0.0142 0.473∗∗ -0.130 0.646∗ 0.00572 0.380∗∗∗ -0.132 0.103
(1.49) (3.26) (0.48) (2.17) (1.29) (4.58) (-0.84) (0.56)
low-rated 0.0280∗ 0.976∗∗∗ 0.257 -0.375 0.0692 1.437∗∗∗ 1.171 1.914
(2.13) (6.32) (0.68) (-0.71) (1.32) (3.55) (1.76) (1.53)
Communications
high-rated -0.000971 0.464∗∗∗ -0.0932 -0.000971 0.00110 0.367∗∗∗ -0.162 0.299
(-0.19) (5.84) (-0.53) (-0.00) (0.29) (4.84) (-1.38) (1.89)
low-rated 0.00941 0.487∗∗∗ 0.158 0.176 0.00852 0.460∗∗∗ 0.0417 0.0491
(1.87) (4.24) (0.84) (0.66) (1.55) (3.39) (0.20) (1.52)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
44
Table 14: Best-in-class screening of long-short portfolios (Carhart)
Equal weighted Value Weighted
Alpha MKT SMB HML UMD Alpha MKT SMB HML UMD
Basic Material
long-short -0.00108 0.424∗∗∗ -0.277 0.0321 -0.149 0.00325 -0.189 1.432 -0.877 -0.428
(-0.15) (4.17) (-1.43) (0.12) (-0.77) (0.16) (-0.57) (1.63) (-1.03) (-0.58)
Consumer Discretionary
long-short 0.00142 -0.356∗∗ -0.618∗ 0.297 -0.355 -0.00208 0.779∗∗∗ -0.925∗∗ 0.0770 -0.477
(0.17) (-2.63) (-2.57) (1.00) (-1.38) (-0.20) (-4.97) (-3.07) (0.24) (-1.63)
Consumer Staples
long-short 0.0103 0.132 -0.137 0.145 0.143 0.0143 0.0546 -0.145 0.0345 -0.0593
(1.50) (1.82) (-0.88) (0.76) (0.68) (1.23) (0.52) (-0.82) (0.16) (-0.18)
Energy
long-short -0.0128 0.848∗∗∗ 0.475 0.115 -0.0898 -0.0130 -0.889∗∗∗ 0.544 0.276 -0.0412
(-1.26) (-4.13) (1.55) (0.26) (-0.25) (-1.20) (-4.14) (1.67) (0.58) (-0.11)
Financials
long-short 0.00461 -0.0284 -0.0681 -0.00646 -0.201 0.000295 -0.000176 0.0309 -0.0360 -0.310∗
(1.03) (-0.47) (-0.52) (-0.05) (-1.76) (0.06) (0.00) (0.25) (-0.26) (-2.46)
Health Care
long-short 0.0000113 -0.127 -0.153 0.0315 -0.167 -0.00609 -0.373∗ -0.141 0.0877 -0.129
(0.00) (-0.76) (-0.71) (0.08) (-0.70) (-0.56) (-2.22) (-0.61) (0.21) (-0.50)
Industrial
long-short 0.000202 -0.00449 -0.275∗ -0.0576 0.0507 -0.00170 -0.182∗ -0.311∗ 0.000114 -0.0355
(0.07) (-0.08) (-2.42) (-0.44) (0.45) (-0.38) (-2.45) (-1.98) (0.00) (0.24)
Real Estate
long-short 0.0107 -0.0965 -0.150 0.207 -0.347 0.00330 0.112 -0.237 0.189 -0.210
(1.18) (-0.72) (-0.61) (0.70) (-1.13) (0.38) (0.89) (-1.13) (0.73) (-0.82)
Technology
long-short -0.0152 -0.501∗∗ -0.375 1.044 -0.0172 -0.0856 -1.044∗ -1.166 -1.671 1.452
(-0.89) (-2.77) (-0.90) (1.71) (-0.05) (-1.18) (-2.17) (-1.88) (-1.43) (1.03)
Communications
long-short -0.0123 0.0179 -0.241 -0.160 0.0648 -0.00922 -0.0860 -0.196 -0.184 0.0921
(-1.60) (-0.13) (-1.04) (-0.51) (0.24) (-1.38) (-0.56) (-0.93) (-0.51) (0.46)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
45
Table 15: Best-in-class screening with high and low rated portfolios (Carhart)
Equal weighted Value Weighted
Alpha MKT SMB HML UMD Alpha MKT SMB HML UMD
Basic Material
high-rated 0.00928 0.716∗∗∗ -0.135 0.0684 -0.264 0.0112∗ 0.709∗∗∗ -0.292∗ -0.152 -0.343∗
(1.76) (10.60) (-0.97) (0.45) (-1.80) (2.23) (9.41) (-2.07) (-0.91) (-2.35)
low-rated 0.0143 0.739∗∗∗ 0.0840 -0.0653 0.0526 -0.0145 2.694∗ -3.437∗ 0.894 2.439
(1.75) (6.38) (0.38) (-0.24) (0.19) (-0.36) (2.61) (-2.14) (0.62) (0.89)
Consumer Discretionary
high-rated 0.00687 0.660∗∗∗ -0.312∗ 0.194 -0.0985 0.00674 0.329∗∗∗ -0.774∗∗∗ -0.0657 -0.175
(1.40) (9.23) (-2.39) (1.03) (0.58) (0.96) (4.51) (-5.04) (-0.33) (-0.95)
low-rated 0.00458 1.016∗∗∗ 0.312 -0.0910 0.239 0.00802 1.097∗∗∗ 0.168 -0.128 0.303
(0.59) (7.81) (1.35) (-0.34) (1.05) (0.98) (7.78) (0.67) (-0.45) (1.25)
Consumer Staples
high-rated 0.0126∗ 0.155∗ -0.314∗ 0.0307 0.0622 0.0180 0.0857 -0.304 -0.0703 -0.169
(2.50) (2.61) (-2.33) (0.17) (0.39) (1.68) (0.86) (-1.82) (-0.32) (-0.54)
low-rated -0.00605 -0.122 -1.770∗ -1.882∗ 0.416 0.00493 0.0711 -1.238∗∗ -1.426∗ 0.103
(-0.40) (-0.43) (-2.51) (-2.47) (0.81) (0.42) (0.36) (-2.90) (-2.39) (0.26)
Energy
high-rated N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A
(N/A) (N/A) (N/A) (N/A) (N/A) (N/A) (N/A) (N/A) (N/A) (N/A)
low-rated 0.0120 0.850∗∗ -0.475 -0.122 0.1000 0.0120 0.892∗∗∗ -0.543 -0.278 0.0447
(1.17) (4.13) (-1.56) (-0.28) (0.28) (1.11) (4.15) (-1.66) (-0.59) (0.12)
Financials
high-rated 0.0116∗∗ 0.688∗∗∗ -0.0198 0.156 -0.290∗∗ 0.0115∗∗ 0.689∗∗∗ 0.0188 0.175 -0.423∗∗∗
(3.08) (13.64) (-0.19) (1.42) (-2.98) (2.74) (11.71) (0.17) (1.42) (-3.62)
low-rated 0.00701∗ 0.718∗∗∗ 0.0417 0.145 -0.0623 0.0107∗∗ 0.690∗∗∗ -0.0113 0.210 -0.112
(1.94) (15.50) (0.44) (1.30) (-0.58) (2.84) (12.41) (-0.11) (1.63) (-1.04)
Health Care
high-rated 0.0128∗ 0.284∗∗ -0.271 -0.0576 -0.199 0.0118∗ 0.105 -0.175 0.0493 -0.195
(2.37) (3.19) (-1.83) (-0.30) (-1.29) (2.54) (1.70) (-1.46) (0.32) (-1.56)
low-rated 0.0116 0.412∗ -1.05 -0.0645 0.0699 0.0174 0.480∗∗ -0.0326 0.0396 -0.0646
(1.23) (2.60) (-0.57) (-0.18) (-0.34) (1.59) (2.90) (-0.15) (0.11) (-0.27)
Industrial
high-rated 0.00967∗ 0.672∗∗∗ -0.207 0.0420 -0.0347 0.0108∗∗ 0.640∗∗∗ -0.246 0.0736 -0.0542
(2.44) (13.13) (-1.67) (0.30) (-0.28) (2.73) (12.63) (-1.96) (0.61) (-0.42)
low-rated 0.000202 -0.00449 -0.275∗ -0.0576 0.0507 0.0120∗ 0.823∗∗∗ 0.0659 0.0725 -0.0171
(0.07) (-0.08) (-2.42) (-0.44) (0.45) (2.12) (10.13) (0.40) (0.32) (-0.09)
Real Estate
high-rated 0.0171∗ 0.491∗∗ 0.386 0.0425 0.0411 0.0123 0.784∗∗∗ 0.308 0.520 0.143
(2.03) (3.17) (1.85) (1.38) (0.16) (1.45) (5.62) (1.45) (1.97) (0.56)
low-rated 0.00600 0.589∗∗∗ 0.535∗ 0.215 0.394 0.00884 0.674∗∗∗ 0.551∗ 0.338 0.356
(1.00) (6.58) (2.15) (0.93) (1.67) (1.37) (7.26) (2.51) (1.35) (1.70)
Technology
high-rated 0.0185 0.454∗∗∗ -0.152 0.616∗ -0.301 0.00593 0.378∗∗∗ -0.133 0.102 -0.0161
(1.37) (3.42) (-0.59) (2.06) (-0.84) (1.10) (4.59) (-0.87) (0.53) (-0.08)
low-rated 0.0326∗ 0.955∗∗∗ 0.234 -0.407 -0.317 0.0911 1.337∗∗ 1.059 1.761 -1.523
(2.16) (6.23) (0.62) (-0.75) (0.96) (1.27) (2.82) (1.72) (1.51) (-1.07)
Communications
high-rated 0.000664 0.456∗∗∗ -0.102 -0.0122 -0.114 0.00120 0.366∗∗∗ -0.162 0.299 -0.00723
(0.13) (5.64) (-0.59) (-0.07 (-0.62) (0.31) (4.82) (-1.37) (1.86) (-0.06)
low-rated 0.0122 0.475∗∗∗ 0.144 0.157 -0.191 0.00992 0.453∗∗∗ 0.0345 0.481 -0.0978
(2.01) (4.23) (0.74) (0.58) (-1.07) (1.47) (3.36) (0.16) (1.46) (-0.47)
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
46
A.2 Total Return of GICS Sectors
Table 16: Gics Industries and total return in percentage
GICS Industry Total Return
Basic Material 21.2%
Consumer Discretionary 34.3%
Consumer Staples 11.0%
Energy -52.0%
Financials 110.2%
Health Care 43%
Industrial 105.8%
Real Estate 77.8%
Technology 12.2%
Communication Services 44.4%
47
A.3 Score Grade
Table 17: The methodology for converting a percentile score to a letter grade score.
Source: Refinitiv, 2022
Score range Grade Description
0.0 ⇐ score ⇐ 0.08333 D- ”D” score indicates poor relative ESG perfor-
mance and insufficient degree of transparency
in reporting material ESG data publicly
0.08333 < score ⇐ 0.1666 D
0.1666 < score ⇐ 0.2500
D+
0.2500 < score ⇐ 0.333 C- ”C” score indicates satisfactory relative to
ESG performance and moderate degree of
transparancy in reporting material ESG data
publicly
0.333 < score ⇐ 0.4166 C
0.4166 < score ⇐ 0.5000 C+
0.5000 < score ⇐ 0.5833 B- ”B” score indicates good relative ESG per-
formance and above average degree of trans-
parency in reporting material ESG data publicly
0.5833 < score ⇐ 0.6666 B
0.6666 < score ⇐ 0.7500 B+
0.7500 < score ⇐ 0.8333 A- ”A” score indicates excellent relative to ESG
performance and high degree of transparency in
reporting material ESG data publicly
0.8333 < score ⇐ 0.91666 A
0.91666 < score ⇐ 1 A+
48