Rewarding sustainability : analyzing the effect of the 2018 global 100 on stock returns

(1)

Faculty of Economics and Business, Amsterdam School of Economics Bachelor Thesis, BSc Econometrics and Operations Research

REWARDING SUSTAINABILITY

ANALYZING THE EFFECT OF THE 2018 GLOBAL 100 ON

STOCK RETURNS

Tom Kapteijn (11058471) June 2018 Supervisor Derya G¨uler Abstract

This study analyses the effect of the 2018 Global 100 by Corporate Knights on the stock returns of the firms that are put on or taken off the 2018 Global 100. It is concluded that there is no evidence of a significant effect on stock returns, however a significant effect on stock volatility is found. It is also concluded that both the effect on stock returns as the effect on volatility are smaller for US traded firms than for non-US traded firms. This is in line with the results of earlier research and economic theory.

(2)

Statement of originality

This document is written by Tom Kapteijn who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1 Introduction

In recent years more and more companies are making decisions based on sustainability. Reed (2017) recently stated that BP, an oil company from origin, is investing $200 million in solar energy. Meaning not only that firms are starting to care about sustainability but are taking action to make their conduct of business more sustainable as well. However, in general sustainability tends to be costly, hence companies only have an incentive to involve sustainability into their decision-making if investors do indeed value the striving for sus-tainability. Therefore the question arises whether investors value corporate sustainability or not.

The first complication of answering this question is how the concept of corporate sustainability is defined and how it is measured. In general corporate sustainability is most-commonly defined as a business strategy that considers more dimensions, such as social and environmental, than just profit-maximization. This means that a corporate sustainable firm is willing to make lower profits if that means the impact that firm has on the environment is lower. There are multiple indexes which attempt to measure corpo-rate sustainability, including the Dow Jones Sustainability World Index, the Global 100 by Corporate Knights and the Global 100 Green Rankings by Newsweek which are all discussed in this paper. These indexes calculate their ranking based on a few criteria such as energy-, carbon- and waste productivity. Because these indexes use a defined formula the rankings are fully objective. This means that these rankings can be used in scientific research into whether investors value corporate sustainability.

This study aims to examine to what extent the 2018 Global 100 list influences stock returns. This research consists of an event study of the announcement day of the 2018 Global 100 list to measure the effect on stock returns, a cross-sectional analysis to measure whether the effect is different for US traded and non-US traded stocks and an analysis regarding the effect of the Global 100 on stock volatility. For the event study this paper uses the methodology discussed by MacKinlay (1997). Companies that have moved in or out of the Global 100 list are taken into the sample and their stock price movements are examined and then compared.

Earlier research into whether investors value corporate sustainability has found mixed results. Cheung (2011) concludes that being on the Dow Jones Sustainability World Index only has an effect on the stock price on short term, and that there is no significant effect in the long term. Murgaia and Lence (2015) conclude that being on the Global 100 Green

(5)

is smaller, although not significantly, for US-traded stocks. They state that this might be because sustainable US-traded stocks are already included in the Dow Jones Sustainability US Index, this means that upon the announcement of the Global 100 less information is released to the market.

This paper contributes to the current literature because of two reasons. First, it examines the effect of the Global 100 list by Corporate Knights, a list of which the effects have not been studied yet. Second, this paper uses more recent data. The awareness of corporate sustainability is growing fast and this is reason to evaluate to what extent results are different for the year 2018. This paper also tries to determine whether there is a significant difference in effect between US- and non-US traded stocks, which Murgaia and Lence (2015) failed to find but economic theory implies there could be.

This paper consists out of multiple sections. The paper begins with the discussion of the theoretical background in section 2. In section 3 the data that is used in this paper is discussed. Section 4 consists out of defining the econometric model that is used for the event study, the cross-sectional analysis and the volatility analysis. In section 5 the results of the study are shown and compared to the results of other studies. Finally, the conclusions of the research are drawn in section 6.

(6)

2 Theoretical background

This section consists out of four subsections. In order to better understand what the Global 100 by Corporate Knights is, the index and the way it is calculated is discussed in the first subsection. Earlier research into the performance of socially responsible investments is shortly reviewed in the second subsection. In the third subsection studies about event study methodology are briefly discussed, so that difference in methodology can be discussed in the fourth subsection. Subsequently, in the fourth subsection a literature overview of earlier related studies into the influence of sustainability indexes on stock returns is given.

2.1 The global 100 index by Corporate Knights

The Global 100 index is created by Corporate Knights Magazine, and the first edition was announced in 2005. The index aims to list the 100 most sustainable corporations. Companies of all industries and countries are eligible, but only large corporations (com-panies with at least 2 billion dollar revenue) are taken into account. The ranking is based on fourteen key performance indicators, which can be divided into resource-, financial-, and employee management KPIs. Only public data is used in the calculations, meaning that submissions from corporations are not required. If data of a certain key performance indicator1 (KPI) is not disclosed the KPI is set to zero resulting in a punishment for non-disclosure (Corporate Knights, 2013).

Companies are evaluated on four universal KPIs and priority KPIs. A KPI is marked as a priority KPI for a certain industry if it is disclosed by at least 10 percent of the companies in that industry, and therefore priority KPIs differ per industry. The four universal KPIs that are used in the calculations are leadership diversity, clean capitalism pay link, pension fund status and percentage tax paid (Corporate Knights, 2013). It is notable that the Global 100 by Corporate Knights index only uses fully objective data, unlike other sustainability indexes that use the opinion of a panel of experts as a variable. This makes the Global 100 by Corporate Knights suitable for scientific research.

2.2 Socially responsible investing

In order to be able to evaluate to what extent investors value sustainability it is important to know whether investing in SRI funds entails a sacrifice in financial returns. Therefore, several studies in to the historic performance of SRI funds are discussed in this subsection.

1

(7)

Socially responsible investments have become more popular over the last years and therewith the total size of SRI funds has grown greatly ((Jones et al., 2008)). However, economic theory suggests that SRI funds should perform worse than conventional funds, because social responsibility adds a constraint to profit maximization and there are addi-tional costs for screening the possible investments on social responsibility.

Jones et al. (2008) examine the performance of SRI funds compared to four Australian benchmarks for the period 1986-2005. In this study a sample of 89 SRI funds is used, and the benchmarks that are used are the ASX100, ASX200, ASX500 and the All Ordinaries Accumulation Index. Jones et al. (2008) use both a single-factor and a multi-factor CAPM model, which means that under-performance can be measured by Jensen’s alpha. They conclude that SRI funds significantly under-perform compared to the market benchmarks. However, Jones et al. (2008) also put a few remarks to their research. First, they mention that the Australian market benchmarks performed exceptionally well in the sample period and significantly better than foreign market benchmarks. The consequence of this is that funds with foreign stocks in their portfolio will tend to perform worse than the Australian market benchmarks by definition. They document that SRI funds with a high domestic exposure under-perform less than SRI funds with a low domestic exposure. Secondly, they mention that the Australian oil- and metal industry performed well in the sample period. This means that avoidance of these industries by SRI funds could be the reason of the under-performance of SRI funds. Although there may be similar patterns, it does not necessarily follow that this result is representative for the rest of the world (Jones et al., 2008).

Curto and Vital (2014) also examine the performance of SRI indexes by comparing the performance of SRI indexes to that of traditional indexes. This study focuses on the period 2001 to 2011 and uses a sample of 14 indexes of which 10 are SRI indexes. They conclude that SRI funds outperform the traditional funds, but that the difference in performance is not statistically significant. Curto and Vital (2014) test for significance using parametric and non-parametric tests, but does not use the CAPM model, therefore the results are hard to compare to the results of Jones et al. (2008).

Jones et al. (2008) mention that studies prior to theirs have failed to use the right models or the needed sample size to find any significant difference in the performance of SRI funds compared to the market benchmarks. They do use appropriate models and sample size, and do find a significant result that corresponds with economic theory. Therefore it is concluded that the economic theory holds, and prior research failed to

(8)

support this because of misspecified models or a shortcoming sample size.

2.3 Event studies

Event study methodology is most commonly used to investigate the effect of sustainability indexes on stock returns. MacKinlay (1997) documents how an event study should be performed. He describes that in order to perform an event study the expected daily returns in the case that there was no event should be estimated, and then be compared to the observed daily returns. The difference between these returns will then be called the abnormal returns, and will also be daily. By taking the sum of these abnormal returns over the period of the event study the cumulative abnormal returns (CAR) is then obtained. It can then be tested whether the cumulative abnormal returns significantly differ from zero. If they do the event does have an effect on stock returns. In order to estimate the expected returns a model, such as the capital asset pricing model (CAPM), is used (MacKinlay, 1997). Studies that examine the effect of sustainability indexes use different models to do this and these differences are discussed in the next subsection.

Sitthipongpanich (2011) puts a few remarks to this by discussing the limitations of an event study. First of all, she states that the abnormal returns might not just be influenced by the examined event, but might also be influenced by market inefficiencies and coexisting events. This could cause a bias in the results. Secondly, the chosen estimation period can influence the results greatly. Thirdly, the model that is chosen to estimate the expected returns might also influence the results. This means that the estimation period and estimation model should be chosen with great care, as also documented by MacKinlay (1997). Overall an event study is a good method to measure the effect of an event but choices made in the event study should be made with care and always be explained well.

2.4 Effect of sustainability indexes on stock returns

There are multiple indexes which aim to evaluate how sustainable firms are and list the most sustainable firms. However, these indexes only have use if they influence the stock market, therefore it is important to analyze whether any effects occur when these indexes are announced. This can be done by studying the period around the announcement day using the event study methodology discussed by MacKinlay (1997). There are multiple studies that have attempted to do this and their results and any differences in methodology are discussed in this subsection.

(9)

Cheung (2011) analyzes the impact of being put on or taken of the Dow Jones Sustain-ability World Index (DJSWI) (inclusion or exclusion) on the stock prices, for the period 2002 to 2008. Because this study attempts to estimate the effect of inclusion or exclu-sion, only firms that are included or excluded from the DJSWI in the sample period are taken into the sample. He uses event study methodology and uses a characteristic-based benchmark model to estimate expected returns, in contrast to earlier studies which use a CAPM model. Cheung (2011) finds that there are temporary significant changes in stock price when an announcement is made, but there is no significant effect on the long term. This means that investors only value corporate sustainability in a temporary way.

Karlsson and Chakarova (2008) perform a similar study, investigating the effect of in-clusion and exin-clusion from the Dow Jones Sustainability Index in the time period between 2002 and 2007. This results in a total sample size of 343 in- or exclusions. Again, event study methodology is used and a single factor CAPM model is used to estimate the ex-pected returns. Karlsson and Chakarova (2008) conclude that there is no significant effect of the DJSI on stock returns, and also specifically state that there is no negative effect of the DJSI on stock returns. Geographical differences are discussed, but these differences are not tested on significance and therefore can not be compared to the results of other studies.

Tsai (2008) (as cited by Cheung (2011)) also analyzes the impact of both inclusions and exclusion from the Dow Jones Sustainability Index for US-traded stocks in the time period 2002-2006. He concludes that exclusion from the DJSI does have a significant negative impact on stock returns, but that inclusion does not have a significant effect on stock returns.

Murgaia and Lence (2015) carry out a slightly different research and examine how the market reacts to the release of Newsweek’s ”Green Ranking” Global 100. Firms that were on the 2009 edition of this index are taken into the sample and their returns in the period around the announcement date are analyzed. They use event study methodology and an extended version of the Fama-French Four Factor Model (FFFM) to estimate the expected returns. Murgaia and Lence (2015) conclude that the relative value of firms on the list increases significantly after announcement of the list, and that this effect is significantly stronger for non US-traded stocks than for US-traded stocks. They state that a possible reason for the difference in effect is that some of the US-traded stocks that were in Newsweek’s Green Ranking had already been in the ”US 500 List”, which is a different index that lists the 500 most sustainable US-traded firms and was released

(10)

earlier in 2009. Besides that they state that the European regulatory history could also explain the difference in effect. The reason for this is that investors might fear further environmental regulations in the future, and such regulation will have a lower impact on firms that already are sustainable to some extent.

Prior research into the effect of sustainability indexes on stock returns resulted in various conclusions. Cheung (2011) and Karlsson and Chakarova (2008) conclude that in-clusion or exin-clusion from the DJSI does not have a significant effect on stock returns. Tsai (2008) (as cited in Cheung (2011)) concludes only a significant negative effect of exclusion from the DJSI on stock returns, but finds no significant effect of inclusion from the DJSI on stock returns. Murgaia and Lence (2015) conclude that inclusion from Newsweek’s ”Green Ranking” Global 100 has a positive effect on stock returns, but use a slightly dif-ferent methodology and do not examine the effect of inclusion or exclusion but the effect of inclusion for all firms on the list. This makes it difficult to compare their results to the results of other studies. Murgaia and Lence (2015) are the only study to compare the effect of sustainability indexes on stock returns for US-traded and non US-traded stocks, and conclude that this effect is stronger, but not significantly stronger, for non-US traded stocks than for US-traded stocks. Because of the conflicting results in earlier studies this paper assumes the hypothesis that there is no significant effect of sustainability indexes on stock returns.

(11)

3 Data

This paper uses data originating from the website of Corporate Knights, the WRDS2 database and the CRPS3 database. Data about the Global 100 index is obtained from the website of Corporate Knights, data concerning the stocks is obtained from the WRDS database and Fama French data is extracted from the CRPS database.

3.1 Corporate Knights

The website of Corporate Knights is used to extract data concerning both the announce-ment day and the Global 100 index itself. Regarding the announceannounce-ment day the original press release of the Global 100 is used, on this press release the announcement date and announcement time are found. Concerning the index both the 2018 Global 100 index and the 2017 Global 100 index are obtained. These two editions are compared in order to find inclusions and exclusion of the 2018 Global 100 index. This results in 47 inclusions and 47 exclusions.

3.2 WRDS database

Daily data of the closing price of the stocks of the firms that were included or excluded in the 2018 Global 100 index is extracted from the Wharton database. This is done for a total period of 310 trading days, of which 265 before the announcement day and 45 after the announcement day. For stocks that are traded on multiple stock exchanges data about the main stock exchange is used. Data about one specific firm that was included in the 2018 Global 100 could not be obtained and that firm is therefore not taken in account in this study. Some observations show extremely high or low returns due to mergers or errors in the data. These observations are set to 0 (return of 0). This is done for a total of 5 observations. Descriptive statistics of the returns are displayed in table 2.

3.3 Fama French variables

Daily data of five Fama French variables for the same period as the stock data is obtained from the CRPS database. The five variables of which data is obtained are: the excess market return, small minus big, high minus low, robust minus weak and conservative minus aggressive. The excess market return is defined as the global market return5 minus the

2_{Wharton Research Data Services} 3_{Center for Research in Security Prices} 4_{Average standard deviation in group}

(12)

Sample n Mean Std. Dev.4 Min. Max.

Inclusion 46 0.0006 0.0130 -0.2586 0.1783

Exclusion 47 0.0005 0.0116 -0.1741 0.1297

Inclusion US traded 9 0.0006 0.0143 -0.2586 0.1783 Inclusion non-US traded 37 0.0006 0.0127 -0.2043 0.1300 Exclusion US traded 13 0.0005 0.0121 -0.1252 0.1297 Exclusion non-US traded 34 0.0005 0.0115 -0.1741 0.1086

Table 1: Descriptive statistics returns (WRDS database)

risk free return. This variable is used to explain the returns by estimating the correlation with the global market. The small minus big variable is used to explain any difference in returns between small and large firms. Small minus big is calculated with the following formula ((Fama and French, 1993)):

SM B = 1/3(Small Value + Small Neutral + Small Growth)

− 1/3(Big Value + Big Neutral + Big Growth) (1) In the same manner, high minus low is used to explain any difference in returns between firms with a high book to market value and firms with a small book to market value. High minus low is calculated as follows (Fama and French, 1993):

HM L = 1/2(Small Value + Big Value)

− 1/2(Small Growth + Big Growth). (2) Robust minus weak and conservative minus aggressive are calculated in the same way as high minus low. These variables are used to explain the difference in returns for robust (high quality) stocks and weak (low quality) stocks and the difference in returns for firms with an aggressive strategy and firms with conservative strategy.

(13)

4 Methodology

This study aims to analyze the returns of the firms that have been included or excluded in the 2018 Global 100 list. This leads to a total sample of 46 inclusions and 47 exclusions, because stock data of one inclusion could not be obtained. Of the 46 firms included 9 are US-traded and of the 47 firms excluded 13 are US-traded. In this subsection event study methodology is discussed, and it is explained how abnormalities in the returns are calculated and tested for statistical significance. It is also briefly described how these abnormalities are compared and it is discussed how abnormalities in stock volatility are measured and tested. Finally, it is explained how the robustness of the findings is verified.

4.1 Event study

This paper uses event study methodology, as thoroughly described by MacKinlay (1997). This methodology compares the real returns to the expected returns in the absence of an event, and is used to measure the effect of an announcement on the stock returns. In order to do this an estimation window and an even window is chosen. The defined estimation period is used to understand how the firms react to market movements. Based on these results the returns of the firms in the event window in the absence of an event given the market movements are estimated. The abnormal returns are then defined as the difference between the estimated returns and the actual returns:

ARi,t = Ri,t− E[Ri,t|Xt] (3)

The course and extent of these abnormal returns are analyzed and can also be tested for significance. In this way it is possible to examine the effect of the announcement and also test whether this effect is statistically significant.

4.2 Choosing the estimation- and event window

The choices of estimation and event window are made based on the windows used by Cheung (2011). Cheung (2011) uses an estimation window that ranges from t = −250 to t = −16 and an event period that ranges from t = −15 to t = 60 (where t = 0 is the announcement day (AD)). However, because data from the CRPS database was only available up to t = 45 the event period ranges up to t = 45 instead of up to t = 60. This event window is longer than the event window that MacKinlay (1997) describes. The result of this is that long term effects are measured rather than short term effects,

(14)

meaning that it is possible to measure whether the value of a firm that is included or excluded from the 2018 Global 100 changes instead of just measuring short term market movements. In this study the 5 trading days before the announcement are excluded to prevent any disturbances in the results caused by the leaking of information in the week before the announcement (MacKinlay, 1997). This results in an total event period of 56 days. 26/01/2017 t = -250 29/12/2017 t = -15 22/01/2018 (AD) t = 0 27/03/2018 t = 45

4.3 Measuring and testing abnormal returns

In this study abnormal returns are calculated and it is tested if these returns are sig-nificantly different from zero for individual firms. Based on these individual effects, a conclusion about the total or mean effect is drawn.

4.3.1 Calculating abnormal returns

Abnormal returns are defined as the difference between the estimated results and the actual returns. The estimated returns while using the CAPM model are the estimated alpha plus the estimated beta times the excess market return. This means that if the CAPM model is used the abnormal returns of firm i at time t are defined as follows:

ARi,t= Ri,t− ( ˆαi+ ˆβi(Rmf,t− Rrf,t)) (4)

Fama and French (2003) conclude that the CAPM fails to estimate returns properly, especially for large stock with a high correlation with the market. Because the firms in this study are only firms with a value of over $2 billion, using the CAPM model might lead to misspecifications. Therefore, this study uses the Fama French five-factor model instead of the CAPM model to estimate expected returns. The Fama Frech five-factor model is specified better for companies with a high correlation with the market and also estimates the returns of companies with an aggresive strategy better. In this study the Fama Frech five-factor model returned a 6.4 percent higher R-squared than the CAPM model for fitting the returns in the estimation window, meaning that it much provides a better fit of the returns. Based on the Fama Frech five-factor model, the abnormal returns can be written as follows:

(15)

ARi,t = Ri,t− ( ˆαi+ ˆβi(Rmf,t− Rrf,t) + ˆsiSM Bt

+ ˆhiHM Lt+ ˆwiRM Wt+ ˆciCM At) (5)

The cumulative abnormal returns are then defined as the sum of abnormal returns in the event period (which ranges from t = −15 to t = 45 in this study):

CARi = 45

X

t=−15

ARi,t (6)

The cumulative abnormal returns represent the abnormalities in returns in the event window. Positive cumulative abnormal returns mean that a firm performed better than would be expected under normal circumstances and the converse holds for negative cumu-lative abnormal returns.

4.3.2 Testing abnormal returns

It is assumed that the abnormal returns are normally, independent and identically dis-tributed with mean 0 and variance σ2

i. Under this assumption cumulative abnormal returns

are distributed as follows (where 56 days is the length of the event period):

CARi∼ N (0, 56σ2i) (7)

This distribution can be rewritten to a Student’s t distribution. This makes it easier to test effects for significance and show test results.

t = CARi σi

√

56 ∼ N (0, 1) (8)

Using this distribution (8) it can be tested whether the individual cumulative abnormal returns are significantly different from zero. The test uses the null-hypothesis that there is no significant effect, and the testing for significance is done on a 5% significance level in this study. This means that the null hypothesis that there is no significant effect will be rejected if t < −1.960 for a negative effect and if t > 1.960 for a positive effect.

4.4 Comparing cumulative abnormal returns

After calculating cumulative abnormal returns, the results of traded and non US-traded stocks are compared and it is tested whether the cumulative abnormal results differ significantly between US traded and non-US traded stocks. This is done by performing

(16)

a linear regression with the cumulative abnormal returns per firm as dependent variable and a dummy that is 1 for US traded stocks as independent variable.

CARi = β0+ β1US.tradedi+ (9)

The regression coefficient is tested for significance using a t-test under the null-hypothesis that β1 = 0. If the coefficient differs from 0 significantly then the cumulative abnormal

returns, and therefore the effect of the announcement on stock returns, is significantly different for US traded firms than for non-US traded firms.

4.5 Measuring and testing abnormal volatility

In addition to the analysis on the effect of the Global 100 on stock returns, the effects of the Global 100 on volatility are also analyzed. This is done using the GARCH6-based approach described by Bia lkowski et al. (2008). In this approach the expected volatility in the absence of an event is compared to the actual volatility. Savickas (2006) concludes that for the purpose of testing event induced volatility the GARCH-based test performs better than the traditional test, the standardized cross-sectional test and the mean rank test. This means that the GARCH based approach should always be used to test event induced volatility in order to prevent any misspecifications.

4.5.1 Measuring abnormal volatility

First, a GARCH(1,1) model is fitted to the returns. This is done by estimating equation (10) and (11) jointly by optimizing the Maximum Likelihood. For estimating of the model a total sample of 294 trading days is used.

Ri,t = α + β(Rmkt,t− Rrf,t) + i,t, i,t ∼ N (0, hi,t) (10)

hi,t = γ0+ γ1hi,t−1+ γ2i,t−1 (11)

After the GARCH model is estimated the expected volatility in the absence of an event is calculated. For this calculation, only information that was available before the event window is used (represented by Ω). This means that the expected volatility can be written as (Bia lkowski et al., 2008) (where t∗ is the last day of the estimation window):

6

(17)

E[hi,t∗_+k|Ω∗_t] = ˆγ₀ k−1

X

j=0

( ˆγ1+ ˆγ2)j+ ( ˆγ1+ ˆγ2)k−1γˆ1hi,t∗+ ( ˆγ₁+ ˆγ₂)k−1γˆ₂2_i,t∗ (12)

The multiplicative abnormal volatility λtis then calculated using the expected

volatil-ity. This variable can be interpreted as the abnormal volatility the stocks are exposed to per day. The multiplicative abnormal volatility is calculated using equation (13) (Bia lkowski et al., 2008), where N is the number of firms included in the sample.

ˆ λt= N X i=1 (N ˆi,t−PNj=1ˆj,t)2 N (N − 2)E[hi,t|Ωt∗] +PN j=1E[hi,t|Ωt∗] (13) The cumulative abnormal volatility is then defined as the sum of the multiplicative abnormal volatility, subtracted by the number of days in the event window (Bia lkowski et al., 2008). In this study the event window ranges from t = 15 to t = 45 and has a total length of 61 days, meaning that the cumulative abnormal returns can be written as:

CAV =

45

X

t=−15

λt− 61 (14)

The cumulative abnormal volatility represents the abnormal volatility of the sample in the event window. This means that it can be both positive and negative. Positive cumulative abnormal volatility imply that the stocks in the sample behaved more volatile in the event window, and the converse holds for negative cumulative abnormal volatility.

4.5.2 Testing abnormal volatility

The null hypothesis for testing abnormal volatility is that the cumulative abnormal volatil-ity is equal to zero. Under this null hypothesis λt represents the variance of a standard

normal distribution. This means that λt follows is distributed as: λt(N − 1) ∼ χ2_{(N −1)}.

The cumulative abnormal volatility, here represented by test-statistic θ, is then distributed as: θ = 45 X t=−15 (N − 1) ˆλt∼ χ2(N −1)(61) (15)

Distribution (15) can be used to test the cumulative abnormal volatility for significance for a certain sample. This is done for six different groups: inclusion, exclusion, inclusion US traded, inclusion non-US traded, exclusion US traded and exclusion non-US traded. If the null hypothesis that there is no abnormal volatility is rejected for a certain group, it means that the stock volatility of that group was affected by the even.

(18)

4.6 Robustness check

In order to check the robustness of the results two additional analyses are performed. In the first extra analysis the CAPM model is used to estimate returns instead of the Fama French factor model, this is done to check how the choice for the Fama French five-factor influences the results. In the second extra analyses the used event window is rages from t = −15 to t = 15 instead of from t = −15 to t = 45, meaning the event window is shorter. This means that this analysis also measures any short term effects of the event.

(19)

5 Results

This study aims to measure the effect of inclusion and exclusion from the Global 100 list on stock returns, the results of this are presented in this subsection. In the first subsection individual results are discussed and an overview of significant effects on the stock returns of individual firms is given. In the second subsection the total effect of the index is discussed, meaning that it is analyzed whether inclusion and exclusion have a significant effect on stock returns on average. In the third subsection the difference in results between US traded and non-US traded firms is discussed. And finally the results of the additional analyses is given in the fourth subsection.

5.1 Measuring individual effects

Table 2 shows the results of the study by displaying the amount of significant individual reactions to both inclusion and exclusion. Inclusion has a significantly positive effect on the returns of four of the included firms, and a negative effect on the returns of only one of the included firms. Exclusion has a significantly positive effect on the returns of two of the excluded firms and a negative effect on the returns of four of the excluded firms.

Sample n Sign. positive Sign. negative % positive CAR

Inclusion 46 4 (8.70%) 1 (2.17%) 43.48%

Exclusion 47 2 (4.26%) 4 (8.51%) 50.00%

Inclusion US traded 9 0 (0.00%) 0 (0.00%) 33.33%

Inclusion non-US traded 37 4 (10.81%) 1 (2.70%) 45.95% Exclusion US traded 13 0 (0.00%) 1 (7.69%) 38.46% Exclusion non-US traded 34 2 (5.88%) 3 (8.82%) 52.94%

Table 2: Individual effects main analysis

The individual results imply that inclusion or exclusion from the Global 2018 does have an effect on stock returns. This conclusion is drawn because of two reasons. First, there are more positive than negative reactions to inclusion and more negative than positive reactions to exclusion. This means that based on the individual results it seems that the effects of inclusion are mainly positive and the effects of exclusion are mainly negative. Second, the reactions differ between inclusion and exclusion. If the reactions would be caused by external factors it would be expected that the reactions of the firms that are included and excluded are the same. This is not the case.

(20)

5.2 Measuring the total effect

In this subsection the individual effects and the course of the mean cumulative abnormal returns are combined in order to draw a conclusion on the total effect of the 2018 Global 100 on stock returns.

Figure 1 displays the course of the mean cumulative abnormal returns for both inclu-sions and excluinclu-sions in the event window, and table 3 shows the mean cumulative abnormal returns for inclusion and exclusion. The mean cumulative abnormal return and the course of the mean cumulative abnormal returns are only marginally different for inclusion com-pared to exclusion. This insinuates that abnormalities are likely caused by external factors and not by the event itself, because effects caused by the index are expected to be different for inclusion and exclusion. This suggests that the effect of the Global 100 on stock returns is either small or non-existent, contradicting the results obtained by measuring individual effects.

For both inclusion and exclusion the abnormal returns are positive in the days before the announcement and negative in the days after the announcement. This could be caused by traders anticipating on the announcement by buying relevant stocks and selling them right before the announcement. If this is the case it means that traders are speculating on the announcement and there is likely a short term effect. However, because this effect is similar for inclusion and exclusion this could also be caused by external factors. The latter is analyzed further in the second additional analysis of this paper.

(21)

Sample n Mean Std. Dev. Min. Max. Inclusion 46 0.0136 0.1318 -0.2383 0.4316 Exclusion 47 0.0134 0.1139 -0.2482 0.3122 Table 3: Descriptive statistics cumulative abnormal returns

subsection it is hard to draw a definitive conclusion of the effect of the Global 100 on stock returns. Although individual firms react mainly positive to inclusion and mainly negative to exclusion, the course of the cumulative abnormal returns do not show any evidence for a effect. Taken together this means that there is no overwhelming evidence for an effect of inclusion or exclusion from the Global 100 on stock returns.

This result is in line with the results of earlier studies. Cheung (2011) investigated the effect of the Dow Jones Sustainability Index (DJSI) on stock returns and also did not conclude any significant effect on stock returns for inclusion or exclusion. Tsai (2008) (as cited in (Cheung, 2011)) did however conclude a significant negative effect for exclusion from the DJSI. The results of this study also imply that such effect could exist, but is very small. A possible explanation for the difference in results could be that Tsai (2008) (as cited in (Cheung, 2011) examined the effect of the DJSI and not the Global 100. The 2017 DJSI index was announced in October 2017, a few months before the 2018 Global 100. This could mean that the 2018 Global 100 released less information to the market and the effects are therefore smaller.

5.3 Comparing between US traded and non-US traded companies In total 4.55% of the US traded firms reacted significantly (positive or negative) to in-clusion or exin-clusion from the 2018 Global 100, for non-US traded firms this number is 14.08%. From these percentages it can be concluded that US traded firms tend to react more moderate to inclusion or exclusion than non-US traded firms. This is confirmed by the mean cumulative abnormal returns which are displayed in table 4. The mean cumu-lative abnormal returns of US traded are closer to zero than the mean of non-US traded stocks, and this means that the effects of the announcement are on average smaller for US traded stocks.

It is also statistically tested whether the reaction of US traded is significantly different than that of non-US traded firms (table 3). This is done by performing a linear regression and using the null-hypothesis that the reaction of US traded firms is the same as that of non-US traded firms. Out of this test is it concluded that there is no significant difference

(22)

Sample n Mean Std. Dev. Min. Max.

Inclusion 46 0.0136 0.1318 -0.2383 0.4316

Exclusion 47 0.0134 0.1139 -0.2482 0.3122

Inclusion US traded 9 0.0005 0.1280 -0.1077 0.3230 Inclusion non-US traded 37 0.0168 0.1342 -0.2383 0.4316 Exclusion US traded 13 0.0121 0.1090 -0.1771 0.1980 Exclusion non-US traded 34 0.0139 0.1173 -0.2482 0.3122

Table 4: Descriptive statistics cumulative abnormal returns in reaction between US traded firms and non-US traded firms.

Test results Inclusion Exclusion Coefficient US traded -0.01621 -0.00174

p-value 0.745 0.693

Table 5: Regression results US vs non-US

Overall it is concluded that non-US traded firms show a stronger, but not significantly stronger, effect of inclusion or exclusion. However even when US traded firms are put aside there is still no evidence for a significant effect of inclusion or exclusion for non-US traded firms.

This result is in line with the results of Murgaia and Lence (2015), who found that US traded stocks react different, but not significantly different, than non-US traded stocks to inclusion or exclusion from Newsweek’s green ranking. An explanation for the more moderate reaction of US traded firms could be that investors fear regulations from the European Union (which only apply to European companies) (Murgaia and Lence, 2015). This would mean that sustainability would be more viable for European firms because investors expect a punishment for non-sustainable European firms. Another explanation could be the announcement of the DJSW (which only features US traded companies) a few months before the announcement of the 2018 Global 100. This could mean that the 2018 Global 100 releases less information to the market about US traded firms than about non-US traded firms. However, the concluded differences in the reaction US traded firms compared to non-US traded firms are very small and not statistically significant. This means that the found differences could also be caused by coincidence, due to the small

(23)

5.4 Effect on volatility

The effect of the Global 100 on volatility is tested for the full event period for six different groups. Of these six groups five groups show a significant change in volatility in the event window (table 6). The only group that does not show a significant change in volatility is the group of US-traded exclusions. This result is important because of two reasons. First, this means that although there is no overwhelming evidence of an effect on stock prices there is statistical evidence that the Global 100 affects the volatility of stocks that are included and excluded. Secondly, this supports the conclusion drawn in the previous subsection that effects of the Global 100 are more moderate for US traded firms.

Sample n CAV p-value

Inclusion 46 31.401 0.000

Exclusion 47 71.205 0.000

Inclusion US traded 9 39.761 0.000 Inclusion non-US traded 37 79.648 0.000 Exclusion US traded 13 -6.673 0.984 Exclusion non-US traded 34 86.820 0.000

Table 6: Cumulative abnormal volatility

The course of the abnormal volatility (displayed in figure 2) is stable in the days before the announcement, meaning it does not show any evidence of anticipation on the announcement as was discussed in section 5.2. However, there is a peak in abnormal volatility in the days after the announcement for inclusion. This could by investors buying stocks that are included in the Global 100 after the announcement. There is also a high peak in volatility for exclusion about 25 days after the announcement. Because it is not logical that stocks are affected by the event so heavily so long after the announcement, it is likely that this peak is caused by external factors.

Although the statistic tests do show a significant effect of inclusion and exclusion from the 2018 Global 100 on the volatility of stocks, the course of the abnormal volatility (displayed in figure 2) is similar for inclusion and exclusion. There could be two possible explanations for this. First, it could be that volatility of stocks that are included is affected in the same as the volatility of stocks that are excluded, and therefore show similar patterns. Second, it could be that the abnormalities are caused by external factors, and the effects of the 2018 Global 100 on stock volatility is actually smaller.

(24)

Figure 2: Abnormal volatility’s (displayed in percentages)

Taken together this means that it can be concluded that inclusion and exclusion from the 2018 Global 100 does have an effect on volatility, although further studies should investigate in what way the Global 100 affects volatility exactly. No earlier research into the effect of sustainability indexes on stock volatility is found, and therefore the results can not be directly compared.

5.5 Robustness check

Two additional analyses are performed to check the robustness of the results. Both of the additional analyses find no overwhelming evidence for an effect of inclusion or exclusion on stock returns, this is in line with the results of the main analysis.

5.5.1 Analysis with shortened event window

In the first additional analysis the event window is shorter than the event window in the main analysis and ranges from t = −15 to t = 15. Because the event window is shorter this analysis will aim to examine the short term effects instead of the long term effects and will therefore be able to determine whether there are any temporary changes in firm value.

Compared to the main analysis four less firms reacted significantly positive to inclu-sion, one more firm reacted significantly negative to incluinclu-sion, the same amount of firms reacted significantly positive to exclusion and two less firms reacted significantly negative

(25)

Inclusion 46 0 (0.00%) 2 (4.34%) 45.65%

Exclusion 47 2 (4.26%) 2 (4.26%) 46.81%

Table 7: Individual effects additional analysis 1

reactions to inclusion is lower, and the amount of negative reactions to exclusion is also lower. This implies that the found effects are more moderate in the short term. This could mean that as a result of inclusion or exclusion from the Global 100 a difference in firm value develops in the long term.

In total 0 of the 22 US traded firms reacted significantly, in the main analysis this was 1 of the 22 firms. This means that compared to the main analysis the US traded firms reacted less significantly, supporting the conclusion that US-traded firms react more moderate to inclusion and exclusion from the Global 100.

5.5.2 Analysis with CAPM model

In the second additional analyses the capital asset pricing model (CAPM) is used to calculate the expected returns instead of the Fama French five-factor model. The CAPM model fits the returns in the estimation window worse, the R-squared of the CAPM model is 6.4 percent lower than the R-squared of the Fama French five factor model on average. However, a additional analysis using the CAPM is still useful to check whether the choice of estimation model influenced the results.

In comparison to the main analysis there are three less firms that reacted significantly positive to inclusion and one more firms that reacted negative to inclusion, the amount of both positive and negative reactions is the same for exclusion. This means that compared to the main analysis a less positive effect is found for inclusion and the effect for exclusion is the same. This means that the choice for the Fama French five-factor model to estimate returns might have influenced the results in a way that the effects of inclusion seem more positive than they are in reality. In the second additional analysis it is found that 1 of the 22 US firms reacted significantly. This is the same amount as in the main analysis.

(26)

Inclusion 46 1 (2.17%) 2 (4.34%) 45.65%

Exclusion 47 2 (4.26%) 4 (8.51%) 44.68%

Table 8: Individual effects additional analysis 2

In both additional analyses a significant positive effect is found in a smaller fraction of the inclusions. This could mean that the choices of the event window length and estimation model influenced the results in a way that lead to a more positive effect for inclusion. The other results are in line with the results of the main analysis. Because the overall differences are very small it is concluded that the results of the main analysis are robust in general.

(27)

6 Conclusion

In this paper the effect of inclusion or exclusion from the 2018 Global 100 on stock returns is analyzed. This is done by analyzing the effect on stock returns, examining to what extent these effects are different for US traded firms compared to non-US traded firms, and measuring whether the 2018 Global has an effect on stock volatility.

Although a significant effect is found for a number of individual firms, no overwhelming evidence for a significant total effect is found. The found abnormalities in stock returns are very similar for inclusion and exclusion meaning that it is likely that these were caused by external factors and not the 2018 Global 100 index itself.

It is concluded that US traded firms react differently to inclusion or exclusion from the 2018 Global 100 but that these differences are not statistically significant. Gener-ally speaking it is found that US traded firms react more moderate, less significantly, to inclusion or exclusion than non-US traded companies.

Statistical evidence is found that the 2018 Global 100 affects the volatility of the stocks that are included and excluded. It is concluded that this effect is smaller for US traded stocks, and the effect is not significant for excluded US traded stocks.

These results are generally in line with the conclusions of previous studies and also correspond with economic theory on the matter. Murgaia and Lence (2015) and Karlsson and Chakarova (2008) also find no significant effect inclusion or exclusion from the Dow Jones Sustainability Index (DJSI). Tsai (2008) (as cited in Cheung (2011) does however conclude a significant negative effect for exclusion from the DJSI. An explanation for this could be that he examined the effects of the DJSI and not the 2018 Global 100. Both Cheung (2011) and Karlsson and Chakarova (2008) find a non-significant difference in reaction between US traded and non-US traded firms, that result was also found in this study and is also in line with economic theory.

Because the results of the additional analyses are similar to the results of the main analysis, it is concluded that the results of the main analysis are generally robust. However, in the additional analyses it is concluded slightly less firms react significantly positive to inclusion. Although the difference in reaction is minor this could mean that the positive effect of inclusion is actually smaller.

Future studies to the subject should use a larger sample size than this study by taking more years than just 2018 into the study. This way it becomes possible to distinguish between different types of firms and test if a significant effect does exist for certain types of companies. For example, in this study it is concluded that stock returns of non-US

(28)

traded companies are affected more heavily by inclusion and exclusion from the Global 100 than stock returns of US traded companies. However, the sample size used in this study is too small to properly test whether these differences are significant. A study with a larger sample size might be able to investigate this further.

Taken together, it is concluded that the 2018 Global 100 index does not have a large impact on stock returns. Because this is in line with the conclusions of other studies, it seems that investors do not value sustainability on the short term. This corresponds with the conflicting economic theories that on one hand sustainable firms tend to perform worse and on the other hand future environmental regulations might bring sustainable firms an advantage. However, because this is the case it might be that if the extra regulations are implemented in the future, investors will start to value sustainability.

(29)

References

Bia lkowski, Jedrzej, Gottschalk, K., and Wisniewski, T. P. (2008). Stock market volatility around national elections. Journal of Banking Finance, 32(9):1941–1953.

Cheung, A. (2011). Do stock investors value corporate sustainability? evidence from an event study. Journal of Business Ethics, 99(2):145–165.

Corporate Knights (2013). The 2017 global 100: Overview of methodology. Retrieved from http://www.corporateknights.com/wp-content/uploads/2017/01/ 2017-Global-100_Methodology-Final.pdf.

Curto, J. and Vital, C. (2014). Socially responsible investment: A comparison between the performance of sustainable and traditional stock indexes. Journal of Reviews on Global Economics, 3:349–363.

Fama, E. F. and French, K. R. (1993). Common risk factors in the returns on stocks and bonds. Journal of financial economics, 33(1):3–56.

Fama, E. F. and French, K. R. (2003). The value premium and the capm. The Journal of Finance, 26(2):165–178.

Jones, S., van der Laan, S., Frost, G., and Loftus, J. (2008). The investment performance of socially responsible investment funds in australia. Journal of Business Ethics, 80:181– 203.

Karlsson, J. and Chakarova, Y. (2008). Does corporate social responsibility pay off?-an event study of the impact of corporate entry and exit from the dow jones sustainability world index on the market value of a company. Industriell och finansiell ekonomi, 07/08 (7).

MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of Economic Literature, 35(1):13–39.

Murgaia, J. M. and Lence, S. H. (2015). Investors reaction to environmental performance: Global perspective of the newsweeks green rankings. Environmental and Resource Eco-nomics, 60(4):583–605.

Reed, S. (December 15, 2017). Bp, once a renewables leader, bets $200 million on solar. The New York Times. Retrieved from https://www.nytimes.com/2017/12/15/business/ energy-environment/bp-lightsource-solar.html.

(30)

Savickas, R. (2006). Event-induced volatility and tests for abnormal performance. Journal of Financial Research, 61(5):2163–2185.

Sitthipongpanich, T. (2011). Understanding the event study. Journal of Business Admin-istration., 34(130):60–68.

Tsai, C.-C. (2008). The reputation effect and value in corporate social responsibility. Unpublished Master Thesis, Department of Finance, Yuan Ze University, Taiwan.

(31)

Appendix A

Company name alpha mkt.rf SMB HML RMW CMA R2 n

1 AAREAL BANK AG -0.00 0.01 -0.01 0.02 -0.00 -0.01 0.27 241 2 ABB LTD -0.00 0.01 -0.01 -0.01 -0.00 0.01 0.25 241 3 ACCIONA SA -0.00 0.01 -0.01 -0.01 0.01 0.01 0.15 241 4 AKZO NOBEL NV 0.00 0.00 -0.01 -0.01 -0.02 0.01 0.11 241 5 ALLIANZ SE 0.00 0.01 -0.01 0.01 0.00 -0.00 0.29 241 6 AMUNDI SA 0.00 0.02 -0.01 0.01 0.00 -0.00 0.23 241 7 ASTRAZENECA PLC 0.00 0.01 0.00 -0.02 -0.02 0.00 0.08 241 8 AXA SA -0.00 0.01 -0.01 0.01 -0.01 -0.01 0.40 241 9 BANCO DO BRASIL -0.00 0.02 -0.00 0.01 0.00 0.01 0.09 241 10 BIOMERIEUX 0.00 0.01 -0.01 0.00 0.01 -0.00 0.16 241

11 CANADIAN IMPERIAL BANK 0.00 0.01 0.00 -0.00 -0.01 0.01 0.23 234

12 CAPITALAND LTD -0.00 0.01 0.00 -0.00 0.00 -0.00 0.04 241

13 CHR.HANSEN HOLDINGS AS 0.00 0.01 -0.00 -0.01 -0.00 -0.00 0.14 241 14 CIA ENERGETICA DE MINAS -0.00 0.02 0.01 0.01 -0.00 0.01 0.09 241

15 DEUTSCHE BOERSE AG -0.00 0.01 -0.00 -0.00 0.01 -0.00 0.10 241

16 ENGIE BRASIL ENERGIA SA -0.00 0.01 -0.00 -0.00 -0.00 0.00 0.13 241

17 GLAXOSMITHKLINE PLC -0.00 0.01 -0.00 -0.01 -0.00 0.01 0.09 241

18 HALMA PLC 0.00 0.01 -0.01 -0.00 0.00 0.00 0.14 241

19 HEWLETT PACKARD ENTERPRISE -0.00 0.01 -0.01 -0.00 -0.01 -0.01 0.04 234

20 HONDA MOTOR CO LTD -0.00 0.01 0.00 0.01 0.01 -0.00 0.19 241

21 INGERSOLL-RAND PLC -0.00 0.01 -0.01 -0.01 -0.01 0.00 0.22 234

22 ITRON INC -0.00 0.01 0.01 -0.01 -0.00 -0.00 0.11 234

23 LILLY (ELI) & CO 0.00 0.00 -0.01 -0.01 -0.01 0.00 0.11 234

24 METLIFE INC -0.00 0.01 -0.01 0.02 -0.01 -0.02 0.34 234 25 NESTLE SA/AG -0.00 0.01 -0.01 -0.01 0.00 0.01 0.20 241 26 NISSAN MOTOR CO LTD -0.00 0.01 0.00 0.01 0.02 0.00 0.11 241 27 NORDEA BANK AB -0.00 0.01 -0.01 0.01 -0.01 -0.00 0.16 241 28 NVIDIA CORP 0.00 0.02 -0.00 -0.03 -0.03 -0.02 0.21 234 29 ORKLA ASA -0.00 0.01 -0.00 -0.00 0.00 0.00 0.09 241 30 ORSTED A/S 0.00 0.01 -0.00 -0.00 0.01 0.01 0.05 241 31 RENAULT SA -0.00 0.01 -0.01 0.01 0.00 0.00 0.13 241 32 SAMSUNG SDI CO LTD 0.00 0.00 0.00 -0.01 -0.00 0.01 0.01 241 33 SANDVIK AB -0.00 0.02 -0.00 -0.00 -0.00 0.00 0.24 241 34 SANOFI -0.00 0.01 -0.01 -0.01 -0.01 0.01 0.16 241 35 SCA-SVENSKA CELLULOSA AB 0.00 0.00 0.00 -0.01 -0.01 0.00 0.02 241 36 SCOTTISH & SOUTHERN ENERGY -0.00 0.00 -0.00 -0.01 0.00 0.00 0.02 241 37 SEKISUI CHEMICAL CO LTD -0.00 0.01 0.00 0.02 0.02 -0.01 0.14 241

38 SUEZ SA -0.00 0.01 -0.01 -0.00 0.01 0.00 0.12 241

39 TAIWAN SEMICONDUCTOR MFG CO 0.00 0.00 0.01 -0.00 -0.00 0.00 0.03 241

40 TELUS CORP 0.00 0.01 0.00 0.00 -0.00 0.00 0.10 234

41 TEXAS INSTRUMENTS INC 0.00 0.01 -0.01 -0.00 -0.00 -0.01 0.36 234

42 UMICORE SA 0.00 0.01 0.00 -0.02 -0.00 0.02 0.09 241

43 VALEO SA -0.00 0.02 -0.01 0.00 0.01 0.00 0.23 241

44 VERBUND AG 0.00 0.01 -0.01 0.01 0.01 -0.01 0.09 241

45 VESTAS WIND SYSTEMS A/S -0.00 0.01 -0.01 -0.01 -0.01 -0.00 0.06 241

46 WARTSILA OYJ ABP -0.00 0.01 -0.01 -0.01 -0.00 0.00 0.20 241

(32)

Company name alpha mkt.rf SMB HML RMW CMA R2 n 1 ACCENTURE PLC 0.00 0.01 -0.01 -0.01 0.00 0.01 0.23 234 2 ADIDAS AG -0.00 0.01 -0.01 0.00 0.00 -0.01 0.07 241 3 APPLE INC 0.00 0.01 -0.01 0.00 -0.00 -0.03 0.39 234 4 ASML HOLDING NV -0.00 0.01 -0.00 -0.01 -0.01 -0.00 0.27 241 5 ASSICURAZIONI GENERALI SPA -0.00 0.01 -0.01 0.01 -0.01 -0.00 0.20 241 6 ASTELLAS PHARMA INC -0.00 0.01 -0.00 0.00 0.01 0.01 0.09 241 7 BANK OF MONTREAL 0.00 0.01 -0.00 0.01 -0.01 -0.00 0.35 234 8 CAMECO CORP -0.00 0.01 -0.00 0.00 -0.01 0.02 0.12 234 9 CENTRICA PLC -0.00 0.00 -0.00 0.00 0.01 0.00 0.01 241 10 COLGATE-PALMOLIVE CO -0.00 0.01 -0.00 -0.00 0.02 0.02 0.15 234 11 CREDIT AGRICOLE SA 0.00 0.01 -0.02 0.01 -0.01 -0.01 0.38 241 12 DANSKE BANK AS 0.00 0.01 -0.01 -0.00 -0.01 0.01 0.25 241 13 DNB ASA 0.00 0.01 -0.01 0.00 -0.01 0.01 0.20 241 14 EDWARDS LIFESCIENCES CORP -0.00 0.01 -0.00 -0.00 0.00 -0.01 0.11 234 15 ENAGAS SA -0.00 0.01 -0.01 -0.00 0.01 0.00 0.18 241 16 ENI SPA -0.00 0.01 -0.01 0.00 -0.01 0.01 0.31 241 17 FRAPORT AG 0.00 0.01 -0.01 0.00 0.00 -0.00 0.09 241 18 GENERAL MILLS INC -0.00 0.01 -0.00 -0.00 0.01 0.02 0.13 234 19 HANG SENG BANK LTD 0.00 0.00 0.00 0.01 0.00 -0.00 0.08 241 20 HENKEL AG & CO KGAA -0.00 0.01 -0.01 -0.01 0.01 0.01 0.23 241 21 HESS CORP -0.00 0.01 0.01 -0.01 -0.03 0.03 0.28 234 22 HOLMEN AB 0.00 0.01 -0.00 -0.01 0.01 0.01 0.05 241 23 IBERDROLA SA -0.00 0.01 -0.01 -0.00 0.01 0.00 0.13 241 24 KONINKLIJKE DSM NV 0.00 0.01 -0.01 0.00 0.00 -0.00 0.16 241 25 LG ELECTRONICS INC 0.00 0.00 0.01 -0.01 -0.01 0.00 0.01 241 26 MARKS & SPENCER GROUP PLC -0.00 0.01 0.00 0.01 0.01 -0.00 0.05 241 27 MICROSOFT CORP 0.00 0.01 -0.01 -0.01 -0.01 -0.02 0.51 234 28 NEC CORP -0.00 0.01 -0.00 0.02 0.03 -0.01 0.11 241 29 NOVOZYMES A/S 0.00 0.01 0.00 -0.01 0.00 0.01 0.09 241 30 PEUGEOT SA -0.00 0.01 -0.01 0.00 -0.01 0.00 0.16 241 31 PROLOGIS INC 0.00 0.01 0.00 -0.00 -0.00 0.00 0.14 234 32 RECKITT BENCKISER GROUP PLC -0.00 0.01 -0.01 -0.01 0.01 0.01 0.14 241 33 RELX PLC 0.00 0.01 -0.01 -0.01 0.00 0.01 0.16 241 34 REXEL SA -0.00 0.01 -0.00 0.00 -0.00 0.00 0.11 241 35 ROYAL BANK OF CANADA 0.00 0.01 -0.00 0.00 -0.01 0.00 0.39 234 36 SKANDINAVISKA ENSKILDA BANK -0.00 0.01 -0.01 0.00 -0.00 0.00 0.26 241 37 SKY PLC -0.00 0.00 0.00 0.00 0.00 -0.00 0.01 241 38 SMITHS GROUP PLC -0.00 0.01 -0.00 -0.00 -0.01 0.00 0.15 241 39 SONOVA HOLDING AG -0.00 0.01 -0.00 -0.01 -0.00 0.01 0.16 241 40 STARHUB LTD 0.00 -0.00 0.00 -0.00 -0.01 0.00 0.02 241 41 STATOIL ASA 0.00 0.01 -0.00 0.00 -0.01 0.01 0.22 241 42 SYSMEX CORP 0.00 0.01 0.00 0.00 0.01 0.00 0.07 241 43 TELENOR ASA 0.00 0.01 0.00 0.01 0.01 0.00 0.07 241 44 TORONTO DOMINION BANK 0.00 0.01 0.00 0.01 -0.01 -0.00 0.23 234 45 TOTAL SA -0.00 0.01 -0.01 0.00 -0.00 0.01 0.38 241 46 VARIAN MEDICAL SYSTEMS INC 0.00 0.01 -0.00 -0.01 -0.01 0.00 0.09 234 47 WOLTERS KLUWER NV 0.00 0.01 -0.00 -0.01 0.01 0.01 0.15 241

(33)

Company name CAR Significantly positive Significantly negative US traded 1 AAREAL BANK AG 0.08 FALSE FALSE FALSE 2 ABB LTD -0.09 FALSE FALSE FALSE 3 ACCIONA SA -0.07 FALSE FALSE FALSE 4 AKZO NOBEL NV 0.04 FALSE FALSE FALSE 5 ALLIANZ SE -0.03 FALSE FALSE FALSE 6 AMUNDI SA -0.03 FALSE FALSE FALSE 7 ASTRAZENECA PLC -0.04 FALSE FALSE FALSE 8 AXA SA -0.10 FALSE FALSE FALSE 9 BANCO DO BRASIL 0.35 TRUE FALSE FALSE 10 BIOMERIEUX -0.04 FALSE FALSE FALSE 11 CANADIAN IMPERIAL BANK -0.08 FALSE FALSE TRUE 12 CAPITALAND LTD -0.00 FALSE FALSE FALSE 13 CHR.HANSEN HOLDINGS AS -0.16 FALSE TRUE FALSE 14 CIA ENERGETICA DE MINAS 0.43 TRUE FALSE FALSE 15 DEUTSCHE BOERSE AG 0.17 TRUE FALSE FALSE 16 ENGIE BRASIL ENERGIA SA 0.10 FALSE FALSE FALSE 17 GLAXOSMITHKLINE PLC 0.11 FALSE FALSE FALSE 18 HALMA PLC -0.08 FALSE FALSE FALSE 19 HEWLETT PACKARD ENTERPRISE 0.32 FALSE FALSE TRUE 20 HONDA MOTOR CO LTD -0.02 FALSE FALSE FALSE 21 INGERSOLL-RAND PLC 0.00 FALSE FALSE TRUE 22 ITRON INC 0.02 FALSE FALSE TRUE 23 LILLY (ELI) & CO -0.11 FALSE FALSE TRUE 24 METLIFE INC -0.01 FALSE FALSE TRUE 25 NESTLE SA/AG -0.07 FALSE FALSE FALSE 26 NISSAN MOTOR CO LTD 0.07 FALSE FALSE FALSE 27 NORDEA BANK AB -0.05 FALSE FALSE FALSE 28 NVIDIA CORP -0.02 FALSE FALSE TRUE 29 ORKLA ASA 0.01 FALSE FALSE FALSE 30 ORSTED A/S 0.13 FALSE FALSE FALSE 31 RENAULT SA 0.19 TRUE FALSE FALSE 32 SAMSUNG SDI CO LTD -0.09 FALSE FALSE FALSE 33 SANDVIK AB 0.08 FALSE FALSE FALSE 34 SANOFI -0.06 FALSE FALSE FALSE 35 SCA-SVENSKA CELLULOSA AB -0.24 FALSE TRUE FALSE 36 SCOTTISH & SOUTHERN ENERGY 0.04 FALSE FALSE FALSE 37 SEKISUI CHEMICAL CO LTD -0.16 FALSE TRUE FALSE 38 SUEZ SA -0.16 FALSE TRUE FALSE 39 TAIWAN SEMICONDUCTOR MFG CO -0.02 FALSE FALSE FALSE 40 TELUS CORP -0.08 FALSE FALSE TRUE 41 TEXAS INSTRUMENTS INC -0.04 FALSE FALSE TRUE 42 UMICORE SA 0.15 FALSE FALSE FALSE 43 VALEO SA -0.02 FALSE FALSE FALSE 44 VERBUND AG 0.08 FALSE FALSE FALSE 45 VESTAS WIND SYSTEMS A/S 0.05 FALSE FALSE FALSE 46 WARTSILA OYJ ABP 0.07 FALSE FALSE FALSE

(34)

Company name CAR Significantly positive Significantly negative US traded 1 ACCENTURE PLC -0.03 FALSE FALSE TRUE 2 ADIDAS AG 0.10 FALSE FALSE FALSE 3 APPLE INC -0.04 FALSE FALSE TRUE 4 ASML HOLDING NV 0.04 FALSE FALSE FALSE 5 ASSICURAZIONI GENERALI SPA 0.04 FALSE FALSE FALSE 6 ASTELLAS PHARMA INC 0.22 TRUE FALSE FALSE 7 BANK OF MONTREAL -0.05 FALSE FALSE TRUE 8 CAMECO CORP 0.18 FALSE FALSE TRUE 9 CENTRICA PLC 0.17 FALSE FALSE FALSE 10 COLGATE-PALMOLIVE CO -0.06 FALSE FALSE TRUE 11 CREDIT AGRICOLE SA 0.02 FALSE FALSE FALSE 12 DANSKE BANK AS -0.01 FALSE FALSE FALSE 13 DNB ASA 0.03 FALSE FALSE FALSE 14 EDWARDS LIFESCIENCES CORP 0.20 TRUE FALSE TRUE 15 ENAGAS SA -0.03 FALSE FALSE FALSE 16 ENI SPA 0.09 FALSE FALSE FALSE 17 FRAPORT AG -0.18 FALSE TRUE FALSE 18 GENERAL MILLS INC -0.18 FALSE TRUE TRUE 19 HANG SENG BANK LTD -0.06 FALSE FALSE FALSE 20 HENKEL AG & CO KGAA 0.05 FALSE FALSE FALSE 21 HESS CORP 0.16 FALSE FALSE TRUE 22 HOLMEN AB -0.03 FALSE FALSE FALSE 23 IBERDROLA SA -0.07 FALSE FALSE FALSE 24 KONINKLIJKE DSM NV -0.02 FALSE FALSE FALSE 25 LG ELECTRONICS INC -0.16 FALSE FALSE FALSE 26 MARKS & SPENCER GROUP PLC -0.06 FALSE FALSE FALSE 27 MICROSOFT CORP 0.02 FALSE FALSE TRUE 28 NEC CORP 0.14 FALSE FALSE FALSE 29 NOVOZYMES A/S -0.13 FALSE FALSE FALSE 30 PEUGEOT SA 0.19 TRUE FALSE FALSE 31 PROLOGIS INC -0.03 FALSE FALSE TRUE 32 RECKITT BENCKISER GROUP PLC -0.03 FALSE FALSE FALSE 33 RELX PLC -0.16 FALSE TRUE FALSE 34 REXEL SA 0.02 FALSE FALSE FALSE 35 ROYAL BANK OF CANADA -0.06 FALSE FALSE TRUE 36 SKANDINAVISKA ENSKILDA BANK -0.03 FALSE FALSE FALSE 37 SKY PLC 0.31 TRUE FALSE FALSE 38 SMITHS GROUP PLC 0.07 FALSE FALSE FALSE 39 SONOVA HOLDING AG 0.00 FALSE FALSE FALSE 40 STARHUB LTD -0.25 FALSE TRUE FALSE 41 STATOIL ASA 0.10 FALSE FALSE FALSE 42 SYSMEX CORP 0.04 FALSE FALSE FALSE 43 TELENOR ASA -0.02 FALSE FALSE FALSE 44 TORONTO DOMINION BANK -0.02 FALSE FALSE TRUE 45 TOTAL SA 0.11 TRUE FALSE FALSE 46 VARIAN MEDICAL SYSTEMS INC 0.07 FALSE FALSE TRUE 47 WOLTERS KLUWER NV -0.04 FALSE FALSE FALSE

Rewarding sustainability : analyzing the effect of the 2018 global 100 on stock returns