Name and ID number: Ingrid Marie Zeiner

Marie Therese Johansen

Master Thesis

- The Effect of Corporate Social

Responsibility Announcements
on a Company’s Stock Returns
and Market Efficiency -

Due-date of delivery:
02.09.2013

Campus:
BI Norwegian Business School- Oslo
Thesis

Examination code and name:

GRA 19003 - Master Thesis
Programme:
Master of Science in Business & Economics, Major in Finance

Supervisor:
Ilan Cooper

“This thesis is a part of the MSc programme at BI Norwegian Business School. The school takes no
responsibility for the methods used, results found and conclusions drawn."
Master Thesis GRA 19003 02.09.2013

Table of Content
ABSTRACT ................................................................................................................................ II

1. INTRODUCTION .............................................................................................................. 1

1.1 CORPORATE SOCIAL RESPONSIBILITY ............................................................................... 1

1.2 MARKET EFFICIENCY ....................................................................................................... 2
1.3 DISCUSSING THE THESIS ................................................................................................... 3
1.4 THE STRUCTURE OF THIS THESIS ....................................................................................... 5

2. RELATED LITERATURE ................................................................................................ 6

3. METHODOLOGY ............................................................................................................. 9

4. DATA ................................................................................................................................ 15

5. EMPIRICAL RESULTS .................................................................................................. 18

5.1 RESULTS OF ABNORMAL RETURNS .................................................................................. 18

5.2 RESULTS OF CUMULATIVE ABNORMAL RETURNS ............................................................. 23
5.3 DISCUSSING THE RESULTS OF AAR AND CAAR .............................................................. 30
5.4 ROBUSTNESS TESTS ........................................................................................................ 35
5.5 COMPARING AAR AND CAAR BEFORE, DURING AND AFTER THE 2008 CRISIS .................. 36

6. CONCLUSION ................................................................................................................. 41

7. BIBLIOGRAPHY ............................................................................................................ 45

8. APPENDIX: EXHIBITS .................................................................................................. 48

9. APPENDIX: PRELIMINARY MASTER THESIS ........................................................ 51

Page i
Master Thesis GRA 19003 02.09.2013

Abstract
In this thesis we study the effects of Corporate Social Responsibility (CSR)
announcements on a company’s stock returns. We focus on announcements
among American corporations from 2005-2012. We perform an event study where
we use the Market Model, the Fama & French Model and the Carhart Model, and
we look for abnormal returns on a firm’s stock returns. Furthermore, we
investigate whether the potential excess returns can be considered an anomaly to
market efficiency. We conclude that CSR announcements result in negative
abnormal returns, and result in negative cumulative abnormal returns.
Furthermore, we conclude that CSR announcements may represent an anomaly to
market efficiency.

Page ii
Master Thesis GRA 19003 02.09.2013

1. Introduction
Over the last decade(s), companies all over the world have increased their
Corporate Social Responsibility (CSR) efforts. Many companies take the
responsibility of contributing to creating a better environment, reducing global
warming, improving health- and social situations for poor countries, improving
the working conditions for employees, etc., very seriously. Therefore, these
companies have established comprehensive CSR programs to combat with many
of the serious issues and problems that the world is dealing with; both on a local
and global level. However, these CSR plans do not only help improving the
world; oftentimes they also increase the value of the respective companies. When
a large company has made an announcement about a significant prospective CSR
investment, it has been reported that these announcements have led to increasing
stock returns for the company (Arx and Ziegler, 2009; Cellier and Chollet, 2011).
However, is this really the case? Do CSR announcements actually increase stock
returns of a company, even though these CSR programs oftentimes cost the
business a large amount of money without generating directly related revenues? In
this thesis we will investigate the relationship between companies’ CSR
announcements and the development of the stock returns of these companies.
More specifically, we will examine whether the firms experience abnormal returns
which can be related to the CSR announcement. We will see whether historic data
from 2005-2012 can show that CSR announcements create abnormal returns, and
if they may be considered an anomaly to market efficiency. Thus, the research
question for this thesis is as follows:
“Does a Corporate Social Responsibility announcement among American
multinational companies over the last seven years indicate abnormal returns?
If so, can this be considered an anomaly to market efficiency?”

1.1 Corporate Social Responsibility

There are many definitions of CSR, several of which include the fact that CSR is
considered to be a firm’s responsibility and commitment to the society and the
firm’s stakeholders. This commitment is also seen as a contribution which
exceeds what is actually required from a firm by different legal aspects and
expectations form regulators (Investopedia 2012). We will use the following

Page 1
Master Thesis GRA 19003 02.09.2013
definition of CSR in our thesis, as we feel that this statement comprehensively
involves important aspect of corporate social responsibility: “A process to
integrate social, environmental, ethical and human rights concerns into their
business operations and core strategy in close collaboration with their
stakeholders” (EU Commission 2012).

CSR can be divided into five different dimensions (Dahlsrud 2006), where each of
the dimensions will define CSR regarding different criteria and thus make distinct
areas of CSR. The five dimensions are of environmental, social, economic,
stakeholder and voluntariness characters. We choose to focus on the two
dimensions of environment and social CSR efforts only. This is due to the fact
that these are the most common CSR efforts, and we want to research the effects
of what is “most commonly done” in the market place. Furthermore, we have an
interest in focusing on these two types due to the fact that two previous studies on
the US market have found different effects of CSR investments in these two
dimensions. Bird et al. (2007) found that investments above minimum
requirements in social and environmental CSR efforts were punished in the
American market in the time period of 1991-2003, while Arx and Ziegler (2009)
found that investments into these two dimensions were positively valued in the
American market in the time period of 2003-2006. Thus, it will be interesting to
compare and contrast our findings to these two studies. The two dimensions of
environmental and social CSR consider CSR regarding “the natural environment”
and “the relationship between the business and the society”, respectively. The
environmental dimension contributes to a better environment by for example
introducing recyclable products, while the social dimension contributes to a better
society by for example providing school books for children in primary schools.

1.2 Market Efficiency

An efficient market is a market where stock prices reflect all available
information. One can separate between weak, semi-strong and strong form market
efficiency. Weak form means that stock prices reflect all information available by
examining market trading data, a semi-strong form means that prices additionally
reflect all public information about the prospect of a company, while strong form
means that stock prices reflect all information relevant to the company; including

Page 2
Master Thesis GRA 19003 02.09.2013
inside information. We will refer to market efficiency as a semi-strong form in
this thesis. New information is immediately embedded into the prices, and the
only way to obtain higher expected returns is to take on higher risk. Thus, neither
technical nor fundamental analysis should be able to give investors abnormal
returns, as abnormal returns represent returns above what is justified by risk
(Bodie, Kane and Marcus 2011, 371-402).

Anomalies to market efficiency represent patterns of returns that seem to

contradict market efficiency. Easily accessible statistics such as a stock’s P/E
ratio, dividends announcements and book-to-market size sometimes seem to give
abnormal returns above what is justified by risk, and can represent anomalies to
the efficient market hypothesis (Bodie, Kane and Marcus 2011, 371-402). For
example, when a firm has a dividend announcement, the stock prices may rise and
the investors may be able to get increased returns when investing in this company.
This increase in returns is not caused by increased risk of the company, but by the
announcement of a dividend distribution. Hence, investors experience abnormal
returns above what is justified by the risk of a certain company, and the dividend
announcement acts an anomaly to market efficiency. In this study we are going to
examine whether we may claim that a company’s CSR announcements provide a
company with abnormal returns, and also if CSR announcements represent an
anomaly to market efficiency.

1.3 Discussing the Thesis

CSR has become a central theme for most American corporations, and many
executives, investors and researchers question the effects of CSR on a company’s
financial performance. It is important to get an estimate of the financial effects of
CSR, as this is likely to have an impact on future CSR efforts by various
corporations. CSR efforts are undoubtedly important around the globe, and
contribute to improving or solving numerous issues that the world is facing. If
researchers and experts are able to prove that CSR efforts not only improve the
world, but also have a positive impact on a firm’s financial performance, it is
more likely that firms will continue to maintain, and hopefully expand, their CSR
programs. Thus, the importance of this research topic cannot be underestimated,

Page 3
Master Thesis GRA 19003 02.09.2013
as positive results and valid evidence may contribute to improving the world
through increased CSR efforts.

In this thesis we will proceed with an event study to examine whether CSR
announcements create abnormal returns and may be an anomaly to market
efficiency. There are 68 announcements in the data set of this thesis. Our
contribution to existing literature lies in the way that we will combine the various
specifications in our study; meaning how we will combine the time frame of data
collection, methods and data. Our main specifications are as follows: 1) we will
look at the effects of CSR for American companies from 2005-2012, 2) we will
analyse the effect of the companies’ CSR announcements themselves; and not the
overall rating of companies’ CSR programs; and 3) we will use the CAPM, the
Fama & French Model and the Carhart Model. While some of these
specifications have been applied in previous studies, the way that we combine
them in this study makes our thesis unique. Thus, it is interesting to see the results
from this study, and how they coincide with existing literature on the topic of
CSR’s effect on financial performance.

Before performing all the analyses of this thesis, we expected that CSR
announcements would have a positive effect on a company’s abnormal returns,
and that CSR announcements would be an anomaly to market efficiency. There
were several reasons why we expected to make this observation. First, it is
because firms can create positive publicity and reputation by contributing to
society. This is likely to improve people’s impression of the company, which can
lead to increased sales, more investors, improved supplier deals, etc. Overall, this
can lead to higher earnings for the company, which furthermore can have a
positive influence on the stock return of the company. Second, performing CSR
efforts may have a positive signaling effect towards the investment community.
The fact that a firm is able to invest in CSR programs signals that the firm has
sufficient funding for investment into projects that do not generate directly related
returns. This furthermore signals financial stability which will be positively
valued by the market, and this may increase the stock return of the company.
Additionally, the fact that a company performs CSR may make people wanting to
associate themselves with the company, which furthermore may help firms recruit
new and talented employees. Thus, CSR activities might attract more employees

Page 4
Master Thesis GRA 19003 02.09.2013
with higher qualifications, which increases the overall quality of the workforce
and hence the company as well.

Although there are many reasons why CSR investments can be positively related
to stock returns, there are also some reasons why there might be a negative
relationship. First and foremost, CSR efforts are generally expensive. They do not
generate any directly related income, which means that CSR investments may be
seen as pure expenses for a company. This can decrease the profitability of the
company, and thus the investment community might see CSR investments as
something that lowers the financial value of a company, and the stock returns of a
company may thereby decrease. Furthermore, CSR efforts can in some situations
create negative publicity for a company if the company is accused of doing CSR
“only for creating positive publicity and higher returns” rather than for the “good
will”. This negative publicity can lead to loss of customers and investors, and this
is likely to have a negative effect on a company’s stock returns.

After having performed all the analyses in the thesis, we observed that our
conclusions are opposite from our expectations; we mostly found a negative
relationship between CSR announcements and abnormal returns. Also, the Market
Model and the Carhart Model have very similar results, yet surprisingly provide
different results than the Fama & French Model. Overall, we find mostly negative
abnormal returns and negative cumulative abnormal returns. Therefore, we
conclude that CSR announcements result in negative abnormal returns for a
company, and may be considered an anomaly to market efficiency.

1.4 The Structure of this Thesis

This thesis will have the following structure: in section two we will introduce the
main findings of related literature on this topic. In the third section we will present
the methodology of our research, and will go into the details of the event study. In
section four we will present the data that we use in our research. Furthermore, in
section five we will present and analyse the empirical results of our study, and we
will proceed to conclude the thesis in section six. We will also include a detailed
bibliography and appendix in the end of the thesis.

Page 5
Master Thesis GRA 19003 02.09.2013

2. Related Literature
During the recent decades, the amounts of CSR efforts that companies have
commit to have grown significantly, and discussions about the wider effects of
these social actions have risen. Several claimed that such CSR programs
influenced the value of the companies in a positive manner, and this incentivised
researchers to further examine the relationship between companies’ CSR
programs and the value of the respective companies. Thus, previous research does
exist on the topic of this thesis, and the researches vary in their findings. Some
studies have results indicating that CSR affects returns positively; others claim
that CSR affects returns negatively, while some also claim that CSR does not
really have an effect on financial returns at all. Furthermore, some studies looked
at all CSR dimensions as one whole, while some studies separated between the
different dimensions of CSR. In this literature review we will present several
studies that have performed various studies and present various findings.

Brammer, Brooks and Pavelin (2006) researched the relationship between

corporate social performance and stock returns in the UK for the time period of
2002-2004. Their main finding was that firms with high scores on CSR (meaning
that they have invested quite significantly in CSR) have lower stock returns, while
companies with the lowest possible score on CSR outperformed the market and
experienced abnormal returns. Thus, this study found that CSR investment is
largely destructive on shareholder value. The study also differentiated between
social, environmental, employment and community CSR performance. It found
that social CSR efforts tend to perform the worst relative to stock returns,
sequentially followed by environment, employment and community CSR
activities.

Bird et al. (2007) researched what CSR activities are valued by the market in the
US, and had a time frame of data collection from 1991-2003. They researched the
five dimensions of community, diversity, employee relations, environment and
product both on the scale of strengths and concerns; implying ten dimensions all
together. Their main finding was that the market seems to value most firms that
satisfied only the minimum requirements of the dimensions of environment and
diversity (mostly as required by law), and that the market punished firms that

Page 6
Master Thesis GRA 19003 02.09.2013
exceeded minimum investment in environmental CSR efforts. They also found
that the social dimension was no longer being valued, and that the market is most
proactive towards CSR in the dimension of employee relations. Furthermore, they
found that the market’s attitude towards CSR activities change over time, and that
the activities recently being valued by the market appear to be diversity, and
employment. Finally, they also found evidence to suggest that companies being
identified in the market as having a wide spectrum of CSR activities are being
rewarded in the market place (and vice-versa), indicating that there are
reputational benefits (and costs) related to CSR programs.

Arx and Ziegler (2009) measured the effect of corporate social responsibility on
stock performance in the US and in Europe for the time period of 2003-2006.
Their analysis showed that financial markets do value environmental and social
activities of a firm compared with other firms within the same industry.
Furthermore, they found that the positive effects on average monthly stock returns
seemed to be more robust in the US rather than in Europe. Arx and Ziegler used
three different models in their study; including the CAPM, Fama & French Model
and the four-factor Carhart Model. They found that the results were more
significant for the simple CAPM than for the Fama & French and the Carhart
Models.

Cellier and Chollet (2011) measured the impact of CSR rating announcements on
stock prices on the European Market from 2004-2009 on short term European
stock returns. Their study showed that CSR really matters for financial markets,
and that different CSR components have different effects on the stock prices. CSR
announcements regarding human rights seem to have a positive effect,
environment and human resources seem to have a negative effect, while
community involvement has a mixed effect.

Mollet and Ziegler (2012) measured the impact of socially responsible investing
(SRI) on American and European stock markets in the time period of 1998-2009.
They used the four-factor Carhart Model, including risk factors for common
market return, size, value and momentum. The researchers found that SRI is
mostly related to large-sized firms. Furthermore, they also found that when all
four risk factors are included, there was no evidence suggesting that SRI was

Page 7
Master Thesis GRA 19003 02.09.2013
neither penalized nor rewarded by the stock markets. Thus, this recent study
suggested that CSR did not have an effect on stock returns in Europe and the US.

Research on this topic has also been performed in other parts of the world outside
the US and Europe. Several similar studies have been conducted on the Asian
market, in which the general consensus seems to be that CSR has a positive
impact on stock performance. A general study of the Asian market found
significant evidence that CSR has a positive effect (Cheung [Link] 2009), in
addition to more narrow studies of China and Taiwan (Chen and Wang 2011;
Wang 2011). A study from Pakistan claimed that CSR had no effect (Iqbal [Link]
2012). Studies from Australia also have mixed findings, where it is both suggested
that CSR does not have a significant impact of stock prices (Newell and Lee
2012) and that there is a weak link (Galbreath and Shum 2012). Furthermore, a
study from South Africa also suggested that CSR announcements have a positive
impact on stock performance; particularly announcements of substantive
monetary value (Arya and Zhang 2008).

The list of previous research could be far longer. Researchers have found different
results over the last decade, and as can be observed CSR activities are valued
either positively, negatively or not valued at all by various markets over various
time periods. As Bird et al. (2007) found, the market’s attitude towards CSR
changes over time. It will be interesting to see how the American market has
valued CSR announcements in the time interval of 2005-2012 in this thesis,
applying the CAPM, the Fama & French Model as well as the Carhart Model.

Page 8
Master Thesis GRA 19003 02.09.2013

3. Methodology
The purpose of this thesis is to investigate whether a CSR announcement can
create abnormal stock returns, and if this can be considered an anomaly with
respect to market efficiency. In order to examine this we have applied an event
study approach, which is a widely used and accepted research methodology in
finance. The actual event in this thesis is a public CSR announcement from a
company, in which a company announces that it will invest in a CSR project. We
look at companies which have had CSR announcements during the period of 2005
and up to 2012, which will be the overall event window. The collected data of the
announcements that we use in our study had to meet certain criteria that we define
in section four of this thesis.

First, we studied if the new information of a CSR announcement triggered an

immediate response in the market, which lead the stock returns to increase or
decrease, and thus create abnormal or excess returns in the short run. Hence, we
looked at a (-3, 3) day time interval around the event date, which gave us a 7 day
event window to check for the immediate response. We also analyzed if one
could observe similar effects in a somewhat longer time horizon around the
announcement day, and thus performed the study at a (-90, 90) day interval as
well; giving us an additional event window of 181 days. Secondly, we
investigated if we could find evidence that the market happened to have learned
about the event before the actual event day, and thus suggesting leakage of
information. Furthermore, we wanted to see if the possible excess return
continued after the actual event, and hence could be considered to be an anomaly
to market efficiency. We studied both event windows for evidence of such results
by performing a CAR study. Additionally, we analyzed our results more in-depth
by making other event windows within these original windows of 7 and 181 days.

For all results that we obtained in this study, we use t- statistics to check for
significance at the 90%, 95% and a 99% significance level. We indicate the level
of significance with stars, where one star (*) relates to the 90% significance level,
two stars (**) for 95% and three stars (***) for the 99% level. We have 68
announcements, and the critical values for the t- statsistics that we apply is based
on the student t- distribution with T- 2 degrees of freedom (Brooks 2008; Exhibit

Page 9
Master Thesis GRA 19003 02.09.2013
1). We used the t- statistics we got from our analysis to check for significant
results, and the t- statistics are calculated in the following way:

(Barber and Lyon 1997)

When performing this event study, we have tested for abnormal returns and
market efficiency. We applied three different models; the single index market
model CAPM, the three factor market model of Fama & French, and finally the
four factor Carhart Model. As one adds independent factors to a model, the
explanation power (indicated with the R2) increases and hence, explains more of
the variation in the data (Brooks 2008). Thus, as the Carhart Model is the largest
model with most independent factors, we expected this model to give the highest
R2 and explanation power.

The CAPM takes into consideration that a stock’s expected return is affected by
one market factor and one firm-specific factor:

Eqtn 1: E(ri) = rf + βi[E(rM) – rf]

E(ri): expected return of stock i
rf: risk free rate
β: stock return’s sensitivity to market return
E(rM): expected market return
(Bodie, Kane and Marcus 2011, 321)

However, we used the index model representation of CAPM in our further

analysis:

Eqtn 2: rt = α + βrMt + εt
rMt: market rate of return in period t
β: the stock’s sensitivity to the market return
εt: part of the security’s return that comes from firm specific events
α: the average rate of return the stock would realize in a period with a zero
market return
(Bodie, Kane and Marcus 2011, 381)

Page 10
Master Thesis GRA 19003 02.09.2013
Fama & French also take into consideration that the return can be affected by one
market factor and one firm-specific factor, while also adding two factors for a
firm’s size and book to market value:

Eqtn 3: rit = αi + βiMrMt + βiSMBSMBt + βiHMLHMLt + εit

SMB = Small Minus Big; i.e. the return of a portfolio of small stocks in excess
of the return on a portfolio of large stocks
HML = High Minus Low; i.e the return of a portfolio of stocks with a high book
to- market ratio in excess of the return on a portfolio of stocks with a
low book to- market ratio.
(Bodie, Kane and Marcus 2011, 363)

The Carhart Model includes one additional independent factor that considers the
momentum effect in the market:
Eqtn 4: rit = αi + βiMrMt + βiSMBSMBt + βiHMLHMLt +βiMOMMOM t + εit

MOM = Momentum effect at the end of month t; i.e. the difference between
winner and loser portfolios in the past.
(Arx and Ziegler, 2009)

The betas in the different models will show how much the returns are affected by,
and correlated with, the market factor in all three models, the SMB and HML
factors in the Fama & French and the Carhart Model, and the momentum effect in
the Carhart Model. The betas are also an indicator of risk, as the assumption of the
market efficiency hypothesis is that higher risk gives higher expected return, as
one should be rewarded for taking upon more risk (Bodie, Kane and Marcus 2011,
371- 402).

The first step in our event study was to make an estimate for what one could
expect the normal stock return in the market (rM) to be if the CSR announcement
events never happened. These proxies can be made in several ways, and we
applied the three models that we included in this study to make such estimates for
each model. We made these proxies by collecting data from the Kenneth French
website. Furthermore, we calculated the stocks’ return by using the following
formula:

Page 11
Master Thesis GRA 19003 02.09.2013

To decrease the chances for the market estimations to be affected by the actual
CSR events that we are studying, we collected the data points for the proxies one
year (250 trading days) prior to the actual event window. Furthermore, we used
the statistical program Eviews to obtain the fitted values and hence, the
estimations for the normal stock returns.

In order to get a better assessment of the potential abnormal returns (AR) resulting
from the CSR announcement, we conducted a more in-depth analyses of the
residuals of the regression models. Thus, this was the second step in our analysis.
The residuals indicates and captures the unexpected; thus, being an indicator for
how much the stocks’ return may be affected by the CSR event that we are
studying. Hence, we had to investigate the residuals further. We did this by
calculating the stocks’ abnormal return by taking the actual return adjusted for
dividends (which we obtained from [Link]) minus the estimated
normal return from the market proxies:

Eqtn 5: MMOD:

Eqtn 6: FF:

(Bodie, Kane and Marcus 2011, 321)

Eqtn 7: CARH:

(Arx and Ziegler, 2009)

The abnormal returns for each of the companies’ announcements included in the
study was calculated, and then averaged (giving AAR). This is because it is more
interesting to see what one could expect the AR to be on average for the market,
rather than for an independent firm.

Page 12
Master Thesis GRA 19003 02.09.2013
We used a hypothesis test to indicate whether CSR announcements have an
impact on the average abnormal returns. In equations 5, 6 and 7 above, we
expected the epsilon to be equal to zero on the actual announcement day (Day 0).
Additionally, we also tested the other days in the event window because we
wanted to see if there were any other days showing abnormal returns. Thus, we
had the following hypothesis:

If ε was significantly different from zero, we rejected the null hypothesis. We

measured this by calculating the t- statistics as previously mentioned, and checked
whether the results were significant. A rejection of the null hypothesis suggests
that CSR announcements represent average abnormal returns (AAR). We
investigated if there was evidence of abnormal returns at the actual event date in
order to see if the market reacted instantly to the occurring event, and if this was
reflected in the stock returns. Furthermore, as said, we also checked for significant
AARs and immediate effects for the other days in the event windows.

To take into consideration the possibility that the market could learn about the
CSR announcement some time before the actual event through an information
leakage, and if the CSR announcement could be considered an anomaly to market
efficiency, the third step in this study was to obtain an indicator that accounted for
this. We did this by summing up the abnormal returns of all of the announcements
within the two time intervals, and got the cumulative abnormal return (CAR).
Once again, we averaged them across all the observations (giving CAAR), as it is
more interesting to see what one can expect the CAR to be on average for the
market, rather than for an independent firm. Making a graph with these results
showed us if the graph drifted before the event, which would suggest information
leakage. Furthermore, and more importantly, the graphs would indicate whether
there were continuous fluctuations of the stocks’ return after the actual event
(Bodie, Kane and Marcus 2011, 382-393). We had the following hypothesis:

H0: CAAR remains stable

HA: CAAR continues to drift

Page 13
Master Thesis GRA 19003 02.09.2013
If the CAAR graph continued to drift after the event, implying a rejection of the
null hypothesis; then the CSR announcements could be considered an anomaly to
market efficiency. We performed this study for both the time intervals of 7 and
181 days, but we also conducted a more in-depth analysis of the CAARs by
calculating the CAARs for different event windows within the two time intervals.
We checked whether the results were significant and if we could reject the H0 or
not. We did this by calculating the t- statistics as previously described, and
checked the different significant levels.

Page 14
Master Thesis GRA 19003 02.09.2013

4. Data
Data of proper quality is vital for most empirical researches, and is also
significantly important for this thesis. As described in the previous section, we
have performed an event study in order to investigate the relationship between
companies’ CSR announcements and their returns. In this section we will describe
the data that we use in the study.

There exist many variations of CSR announcements. CSR actions differ in size,
field, type and budget, and the companies also differ in size, industry, nationality,
etc. Thus, it was necessary to define certain criteria for the CSR announcements
that we have included in our study. These criteria concern the CSR action itself (in
section A below), as well as the actual company (in section B below). We have
used the following criteria for this thesis:

A. CSR projects may be of very different amounts; ranging from a few dollars up
to several hundred million dollars. In this thesis we have chosen to apply a
range of funding which includes some of the most commonly used amounts on
CSR projects. Thus, we have a range of $1,000,000-$50,000,000 on the CSR
announcements in our study. This range allowed us to include a wide range of
announcements, while simultaneously eliminating the smallest and largest
announcements such that we increase the chances of avoiding big outliers in
our data sample. Additionally, we want our results to be the most possible
applicable to what we commonly experience in the market place, and thus we
chose this common range of funding. Furthermore, the CSR announcements in
our study had to be of an environmental or social character, and the CSR
efforts could be domestic and/or international.

B. We also made certain criteria with regards to what firms we would include in
our study. First and foremost, we looked at American companies only.
Second, the companies had to be at least ten years old by the announcement
date, as we did not want to include younger companies that are in a natural
growth phase. Growth companies can expect highly fluctuating returns, and
thus it could be more difficult to find the actual effect of the CSR
announcements. Third, we included a maximum of three announcements per
company with at least a year and a half in between each of them. We only
Page 15
Master Thesis GRA 19003 02.09.2013
wanted to include a maximum of three per company, as we did not want single
companies to have too large of an impact on our results. For example, if we
had included many announcements for one firm, the overall results of our
analyses would have been relatively impacted by the individual performance
of this specific firm. Furthermore, we maintained at least a year and a half in
between each announcement per firm to ensure that the estimation windows
and event windows would not overlap. Fourth, the firms did not necessarily
have to be multinational. Many firms were multinational; however, since we
are looking at the effects of CSR announcements on the American stock
market, we also included firms that were merely of a domestic character. Fifth,
the firms had to be listed on an American stock exchange, and sixth, we did
not include not-for profit organizations. Lastly, the firms did not have to be
within a certain industry. We included CSR announcements from firms across
a wide specter of industries, for example finance, retail, technology, services,
food and drink, sports, etc.

The quality of the data is vital for an empirical study like this, and finding
sufficient data is oftentimes a challenge. There exist databases which contain
substantial information about companies’ CSR programs and announcements. For
this thesis we have used the webpage [Link] as the main database for
gathering information about companies’ CSR announcements. This webpage has
gathered public releases regarding CSR efforts from 2,800 companies worldwide,
dating back to approximately year 2000. Thus, we have researched numerous
press releases from various American firms regarding CSR, and have matched
these CSR announcements with the criteria that we have set for our study. We
have found 68 announcements to match the criteria for our study, and will thus be
able to make valid inferences.

We have used data from CSR announcements that have taken place between 2005
and 2012. When looking at companies’ press releases for their CSR
announcements, we were able to find the exact dates of the announcements.
Thereafter, to find data on company returns, we gather stock prices for each firm
from [Link]. When finding values for the SMB and HML factors in
order to apply the Fama & French model, as well as the momentum factor in the
Carhart Model, we retrieved information from the homepage of Kenneth French.

Page 16
Master Thesis GRA 19003 02.09.2013
We also collected data on market return from this page in order to make the
market proxies.

The 68 announcements included in this study are divided among 39 different

companies. We gathered some basic information about each of these companies,
to get an idea of what types of firms invest in CSR efforts matched to our criteria.
Exhibit 2.1 and 2.2. present descriptive statistics of the companies included in the
study, as well as CSR funding amounts and dates of announcements. As
presented, the companies represent a wide range of sectors and industries, with the
top three sectors represented being services (36%), technology (18%) and
consumer goods (13%). Most of the firms are relatively large companies, with
90% of the companies being listed in the S&P 500. Furthermore, most of the
companies are fairly old and well-established. Despite the fact that one of our
criteria was that the companies had to be at least ten years old by the
announcement day (implying that the firms had to be established by 2002 at the
latest), only 2 out of the 39 companies were established after 1990. Thus, a
general consensus is that the companies in our dataset are fairly large and well-
established, and represent a wide area of sectors and industries.

Page 17
Master Thesis GRA 19003 02.09.2013

5. Empirical Results
In this section we will present the results of our analyses. First, we will present the
results from the tests of abnormal returns and the tests of the cumulative abnormal
returns, and thereafter we will discuss the implications of these results.
Furthermore, we will perform robustness tests of the results in our analyses.
Finally, we will compare and contrast the abnormal returns and the cumulative
abnormal returns before, during and after the 2008 financial crisis.

5.1 Results of abnormal returns

In section 3 we stated that we expect the epsilon term to be zero on the
announcement day, and hence that the CSR announcements do not to have an
impact on the average abnormal returns (AAR). We also want to test potential
AARs on other days in our two event windows, and thus we presented following
hypothesis:

Evidence of significant AAR would imply that we could reject H0 and that the
epsilon term is different from zero as the alternative hypothesis states:

We will investigate these hypotheses by looking at the results from our analysis.
Table 1 presents the average abnormal returns (AAR) each day of the short time
horizon of the 7 days around the CSR announcements for all the companies,
hence, representing the excess returns in the 7 days perspective around the event.
Table 2, on the other hand, represents the AAR for the companies at different days
within the longer event window of 181 days. In this table the AARs are presented
for only a select number of days within the whole event window. The AARs are
given by the first line for every day, the t- statistics are given in the second line,
and the third line states whether the results are significant and at what level. All of
our data is tested for heteroscedasticity by applying the White’s Test,
autocorrelation by applying the Durbin Watson Test, and we also tested for
multicollinearity. We found no evidence of the latter two, but we found some
evidence of heteroscedasticity in the data for a few companies. We corrected for
this, but we found no significant change in the results.

Page 18
Master Thesis GRA 19003 02.09.2013
Both Table 1 and 2 show results which indicates that within a short and a longer
event window one could expect to get days with both positive and negative AARs
related to a CSR annnouncement, yet most negative, as there are altering signs for
the AARs during these particular days. However, the applied models give
somewhat different results and there are also only a few days within the event
windows that actually give statistically significant results. We will comment on
selected significant results. All three models in our analyses; the Market Model,
the Fama & French Model and the Carhart Model, are included in the tables. One
can observe from the tables that the R2 increases as additional factors are included
and hence, the larger model has a higher explanation [Link], the Carhart
Model has the highest R2 with 0,456 and 0,462 in Table 1 and 2 respectively. This
means that the model explains 45,6 % and 46,2% of the variation within the data,
which is more that what the other two models captures.

Table 1: AAR (-3,3) Interval, 7 days Event Window 0,0030

Table 1
0,0020
Event Fama & French
Statistic Market Model Carhart Model
Window Model
0,0010
R2 0,4200 0,4439 0,4560
AAR 0,0021 0,0016 0,0021 0,0000
-3 T-statistic 1,1723 0,8646 1,1903
Significance insign insign insign
-0,0010
AAR -0,0015 -0,0011 -0,0019
-2 T-statistic -1,1431 -0,8522 -1,3897
Significance insign insign insign -0,0020
AAR 0,0001 0,0003 0,0012
-1 T-statistic 0,0510 0,1752 0,6075 -0,0030
Significance insign insign insign
AAR -0,0032 -0,0028 -0,0033 -0,0040
0 T-statistic -2,0193 -1,7792 -1,9467
Significance ** * *
AAR 0,0023 0,0020 0,0022
1 T-statistic 1,5602 1,2451 1,4573
0,0030
Significance insign insign insign
AAR -0,0016 -0,0021 -0,0017
0,0020
2 T-statistic -1,1990 -1,5798 -1,3039
Significance insign insign insign
AAR -0,0005 -0,0001 -0,0011 0,0010
3 T-statistic -0,3836 -0,0384 -0,7909
Significance insign insign insign 0,0000

-0,0010

Table 1 shows that the analysis give insignificant results for most of the days in
the short event window. However, an important finding is that the actual day of
the announcement (Day 0) has t- statistics that gives statistically significant results
at the 95% level (indcated by two stars **) for the Market Model and at the 90%
level (indicated by one star *) for the other two models. The AARs are all slightly
Page 19
Master Thesis GRA 19003 02.09.2013
negative with -0.0032 , – 0.0028 and – 0.0033 for the Market Model, Fama &
French model and the Carhart Model, respectively. This implies that when the
CSR action of a company is announced, it will effect the firms’ stock returns
negatively with approximately -0.3 %, which furthermore could mean that the
market values such CSR activities negatively. Thus, this suggests that we can
reject H0 of having a zero epsilon on the announcement day, and claim that CSR
announcements lead to negative abormal returns. A company that attend to
announce an upcoming CSR action could therefore expect to underperform with -
0.3% relative to the market at the announcement day.

Page 20
Master Thesis GRA 19003 02.09.2013
Table 2: AAR (-90,90) Interval, 181 days Event Window
Table 2
Event Fama & French
Statistic Market Model Carhart Model
Window Model
2
R 0,4265 0,4524 0,4623
AAR 0,0027 0,0023 0,0025
-90 T-statistic 1,7139 1,5221 1,6579
Significance * insign insign
AAR 0,0004 0,0009 0,0002
-75 T-statistic 0,3004 0,5570 0,1167
Significance insign insign insign
AAR 0,0002 -0,0005 -0,0007
-50 T-statistic 0,1577 -0,4001 -0,6774
Significance insign insign insign
AAR -0,0015 -0,0010 -0,0017
-30 T-statistic -1,0133 -0,6949 -1,1642
Significance insign insign insign
AAR 0,0004 0,0010 0,0010
-20 T-statistic 0,2344 0,6333 0,5939
Significance insign insign insign
AAR -0,0018 -0,0017 -0,0019
-15 T-statistic -0,7612 -0,7546 -0,8309
Significance insign insign insign
AAR -0,0008 -0,0006 -0,0008
-10 T-statistic -0,5040 -0,3577 -0,5006
Significance insign insign insign
AAR 0,0009 0,0012 0,0018
-5 T-statistic 0,4799 0,6387 0,9687
Significance insign insign insign
AAR -0,0025 -0,0020 -0,0017
5 T-statistic -1,2019 -0,9669 -0,7864
Significance insign insign insign
AAR -0,0026 -0,0022 -0,0018
10 T-statistic -2,3196 -1,7150 -1,3534
Significance ** * insign
AAR 0,0028 0,0034 0,0019
15 T-statistic 1,2463 1,4719 0,8484
Significance insign insign insign
AAR -0,0026 -0,0023 -0,0025
20 T-statistic -1,7832 -1,5588 -1,6706
Significance * insign *
AAR -0,0002 -0,0006 -0,0014
30 T-statistic -0,1385 -0,4416 -1,0093
Significance insign insign insign
AAR -0,0017 -0,0018 -0,0021
50 T-statistic -1,1506 -1,3136 -1,3793
Significance insign insign insign
AAR 0,0017 0,0015 0,0023
75 T-statistic 0,8184 0,7378 1,1299
Significance insign insign insign
AAR -0,0168 -0,0165 -0,0170
90 T-statistic -1,1305 -1,1128 -1,1555
Significance insign insign insign

Table 2 shows that the Market Model in total gives the most significant results.
This is somewhat expected as the other two models are stricter and therefore it can
be harder to obtain significant results. However, one can also see that the Market

Page 21
Master Thesis GRA 19003 02.09.2013
Model only has two more significant results than the other models, and that the
AARs in this time interval of 181 days mostly are insignificant. Nevertheless, 90
days prior to the actual event the Market Model has a significant result at the
90%- level with positive AAR at 0,27%. Furthermore, at day 10 after the
announcement, the Market Model and the Fama & French Model have significant
AAR at the 95% and 90% level respectively. However, these ARRs are negative
with approximately – 0, 2%. One can also see similar results at day 20 after the
announcement for the Market Model and the Carhart Model, where there are
statistically significant negative AARs with around – 0,2% at the 90%- level.

According to the Market Model , the positive and signifcant result 90 days prior to
the event could imply that the market learns about the CSR activity some time
before the event actually happens, and at this time values it positively. On the
other hand, the significant negative results that we observe 10 and 20 days after
the event has occured, could signal that the firm is somewhat punished for its
actions. These findings could imply that the market players may value the thought
of a company being involved in CSR acivities (indicated by positive AAR at -90),
whereas when it actually becomes a reality it is no longer valued in a positive way
(indicated by negative AARs at 10 and 20). Another interpretation could be that
the market does learn about the event 90 days prior to the official announcement,
and values such activities positively which is reflected in the stock returns at that
point in time. Then, when the actual announcement is made, the positive reactions
are already included in the stock returns, hence, the negative returns are
essentially a result of the fact the company is spending money. However, one have
to keep in mind that Table 2 also shows that there are many days which do not
have evidence of sigfnificant AARs within the time interval of 181 days.
Furthermore, in order to make proper statements regarding the market reactions
and the possibility of information leakage, one has to investigate this further in a
CAR study. We have conducted such a CAR study, which will be analyzed in the
next section. Implications and a discussion of the results from the AAR and
CAAR studies will also be adressed more in-depth in section 5.3.

To summarize, we get somewhat altering results regarding the AARs both for the
short and the longer term event windows. Therefore, we have to interpret the
results of the AARs on the days around the event somewhat carefully. We get a

Page 22
Master Thesis GRA 19003 02.09.2013
positive and significant result before the event day, but most of the significant
results are negative. Even though many of the days in the event windows are
insignificant with respect to the AARs, the most important and interesting
evidence of the AAR study is the statistical significant results on the actual event
day (Day 0). All three models had significant results with a negative AARs with
approxiately – 0,3%. Based on this,we can say that the epsilon term is different
from zero on the announcement day, and thus reject H0: εt = 0. This implies that
CSR announcements could lead to significant abnormal returns.

5.2 Results of cumulative abnormal returns

By summing up the ARs of the announcements, we can obtain the cumulative
abnormal returns across all announcements, and furthermore make a graph of the
average CAR (CAAR). If the graph drifts before the event, this gives us an
impression that the market has some information about, and reaction to, the event
prior to the actual event date. Furthermore, and more importantly, the CAAR
graph can show us whether the CSR announcements could be an anomaly to
market efficiency theory. In section 3 we stated that to be in line with this theory,
the graph of the CAAR should be somewhat stable after the event day. Hence, we
got the following hypotheses:
H0: CAAR remains stable
HA: CAAR continues to drift

Evidence of drifting in the graphs after the event would imply that we could reject
H0 and have evidence for market anomalies. We will look into different event
windows within each of the original event windows of 7 and 181 days, in order to
check if there are different results regarding the significance.

Graph 1 presents the cumulative average abnormal returns (CAAR) for the
complete 7 days event window for all three models, while Graph 2 represents the
CAAR for the companies for the complete 181 days event window.

Page 23
Master Thesis GRA 19003 02.09.2013
Graph 1: CAAR, (-3, 3) Interval, 7 days Event Window

Graph 2: CAAR, (-90,90) Interval, 181 days Event Window

Both graphs depict that the CAAR drifts before as well as after the announcement
day (Day 0). This implies that the market does get some information about the
ocurring event some time before it is actually announced, and that there are
market reactions that affetcts the stock returns after the event has been announced
as well.

In the short event window represented in Graph 1 one can see that there is a drop
in the stock returns right before the event and at day 0, followed by an incline the
first day, until it continuses to drift donwards again. One can also see that the
three models give somewhat equal results within this time frame. Graph 2, on the
other hand, which presents the longer event window, shows that there are clearly
different results amongst the three models. The Market Model and the Carhart
Page 24
Master Thesis GRA 19003 02.09.2013
Model follow an equal distribution of the CAARs, where one can see that there is
a clear downward drift after the event. The Fama & French Model also continues
to have a declining CAAR after the event, however the decline is more modest in
this model. It is somewhat puzzling that the results of the Market Model and the
Carhart Model are more alike than the Fama & French and Carhart Model.
Therefore, to ensure that our analyses were correct, we recollected all the data and
checked the data several times, and then repeated the analyses again.

Regardless the model, the graphs indicate that we could reject the H0: CAAR
remains stable, and say that the CAAR continues to drift. However, we wanted to
investigate the CAAR results more in-depth and check for significance. Thus, we
divided the 7- and 181 days event windows into smaller event windows. Table 3
displays the CAARs for the different event windows within the 7 days original
event window, whereas Table 4 represents the results of the event windows within
the 181 day time frame. The CAARs are given by the first line for every day,
while the t- statistics are given in the second line and the third line states whether
the results are significant or not and at what level.

Table 3: CAAR;
Table 3 In-Depth analysis within 7 days Event Window
Event Fama & French
Statistic Market Model Carhart Model
Window Model
CAAR -0,0008 -0,0005 0,0000
-1,1 T-statistic -0,3432 -0,1958 0,0102
Significance insign insign insign
CAAR -0,0031 -0,0025 -0,0022
-1,0 T-statistic -1,5823 -1,2733 -0,9767
Significance insign insign insign
CAAR -0,0009 -0,0008 -0,0011
0,1 T-statistic -0,4330 -0,3920 -0,4857
Significance insign insign insign
CAAR -0,0024 -0,0016 -0,0019
-2,1 T-statistic -0,8319 -0,5810 -0,5886
Significance insign insign insign
CAAR -0,0047 -0,0037 -0,0041
-2,0 T-statistic -1,8344 -1,4597 -1,4615
Significance * insign insign
CAAR -0,0025 -0,0029 -0,0028
0,2 T-statistic -0,9534 -1,1244 -0,9710
Significance insign insign insign

Table 3 shows that there is only one event windows that is statistically significant,
which is the (-2, 0) event window. The CAAR is negative with -0,47% and it is
Page 25
Master Thesis GRA 19003 02.09.2013
significant at the 90%- level, meaning that the firms experience negative market
reactions related to the CSR announcements which is reflected in the stock
returns. However, this result is only significant for the Market Model, while the
other two models do not have any significant results within these event windows.

Page 26
Master Thesis GRA 19003 02.09.2013
Table 4: CAAR, In- Depth analysis within 181 days Event Window
Table 4
Event Fama & French
Statistic Market Model Carhart Model
Window Model
CAAR -0,1564 -0,0767 -0,1449
-90,90 T-statistic -2,6282 -0,8068 -2,4120
Significance ** insign **
CAAR -0,0288 -0,0066 -0,0290
-90,0 T-statistic -1,6391 -0,2534 -1,5622
Significance insign insign insign
CAAR -0,0418 -0,0176 -0,0441
0,90 T-statistic -1,8427 -0,5526 -1,9583
Significance * insign *
CAAR -0,0446 -0,0043 -0,0466
-75,75 T-statistic -2,2010 -0,1067 -2,1019
Significance ** insign **
CAAR -0,0185 0,0004 -0,0209
-75,0 T-statistic -1,3383 0,0174 -1,4935
Significance insign insign insign
CAAR -0,0294 -0,0074 -0,0293
0,75 T-statistic -1,9313 -0,3047 -1,6444
Significance * insign insign
CAAR -0,0369 -0,0095 -0,0398
-50,50 T-statistic -2,3828 -0,3411 -2,3384
Significance ** insign **
CAAR -0,0156 -0,0032 -0,0164
-50,0 T-statistic -1,5494 -0,1963 -1,5690
Significance insign insign insign
CAAR -0,0246 -0,0090 -0,0271
0,50 T-statistic -2,0663 -0,5427 -1,9822
Significance ** insign *
CAAR -0,0150 0,0007 -0,0187
-30,30 T-statistic -1,4733 0,0424 -1,6782
Significance insign insign *
CAAR -0,0065 0,0022 -0,0069
-30,0 T-statistic -0,8328 0,2096 -0,9113
Significance insign insign insign
CAAR -0,0118 -0,0042 -0,0155
0,30 T-statistic -1,2826 -0,3526 -1,4690
Significance insign insign insign
CAAR -0,0021 0,0084 -0,0025
-20,20 T-statistic -0,2422 0,6482 -0,2681
Significance insign insign insign
CAAR 0,0021 0,0068 0,0005
-20,0 T-statistic 0,3316 0,8619 0,0843
Significance insign insign insign
CAAR -0,0075 -0,0011 -0,0067
0,20 T-statistic -0,9842 -0,1137 -0,7678
Significance insign insign insign
CAAR -0,0162 -0,0128 -0,0147
-10,10 T-statistic -2,5013 -1,5944 -2,1666
Significance ** insign **
CAAR -0,0061 -0,0033 -0,0052
-10,0 T-statistic -1,2344 -0,6107 -1,0397
Significance insign insign insign
CAAR -0,0134 -0,0122 -0,0132
0,10 T-statistic -2,5666 -2,0275 -2,3999
Significance ** ** **

Table 4 somewhat confirms what one could observe from the graphs, namely that
the results of the Market Model and the Carhart Model are more similar to each
Page 27
Master Thesis GRA 19003 02.09.2013
other than the Fama & French Model. The table also shows that once again, we
get both significant and insignificant results with respect to the CAARs. The
Market Model and the Carhart Model have significantly negative CAARs for the
(-90, 90) window, whereas the Fama & French Model has an insignificant CAAR
for this window. This would also mean that the distribution of the Fama & French
CAAR as observed in Graph 2, which were completely different from the other
two models, is not significant. These results imply that for the time interval of 181
days one would expect to get negative CAARs with approximately -12% in
relation to a CSR announcement by applying the Market Model and the Carhart
Model, while the Fama & French Model does not give any significant
implications of these matters as it has insignificant results. Furthermore, the
significant results of the Market Model and the Carhart Model; and their
corresponding drifting graphs; suggest that CSR announcements could be
considered to be an anomaly to market efficiency.

In general, Table 4 shows that the Market Model and the Carhart Model seem to
have more significant results in the longer event windows than the Fama &
French Model; for instance, the first two models have significant negative CAARs
at the 95% -level for the event windows of (-75, 75) , (-50,50) and (- 10, 10).
However, it is important to point out that the event window of (0, 10) is
statistically significant for all the three models with t- statistics which are
significant at the 95% - level. Due to this, and the fact that the Market Model and
the Carhart Model were significant at the (-10,10) window, we analyzed this event
window more detailed. Therefore, we created different event windows within the
(-10, 10) time frame. The results are shown in Table 5 below:

Page 28
Master Thesis GRA 19003 02.09.2013
Table 5: CAAR, In-Depth Analysis within 181 days Event Window
Table 5
Event Fama & French
Statistic Market Model Carhart Model
Window Model
CAAR -0,0042 -0,0015 -0,0037
-10,1 T-statistic -0,8249 -0,2537 -0,6861
Significance insign insign insign
CAAR -0,0082 -0,0047 -0,0065
-10,5 T-statistic -1,3925 -0,6738 -1,0388
Significance insign insign insign
CAAR -0,0095 -0,0060 -0,0075
-10,7 T-statistic -1,5096 -0,8174 -1,0838
Significance insign insign insign
CAAR -0,0162 -0,0128 -0,0147
-10,10 T-statistic -2,5013 -1,5944 -2,1666
Significance ** insign **
CAAR -0,0118 -0,0038 -0,0076
-10,15 T-statistic -1,7641 -0,4279 -1,1073
Significance * insign insign
CAAR -0,0104 -0,0017 -0,0082
-10,20 T-statistic -1,3316 -0,1586 -0,9587
Significance insign insign insign
CAAR -0,0033 -0,0005 -0,0025
-5,5 T-statistic -0,6433 -0,0805 -0,4384
Significance insign insign insign
CAAR -0,0012 0,0009 -0,0011
-5,0 T-statistic -0,3245 0,2215 -0,2876
Significance insign insign insign
CAAR -0,0054 -0,0041 -0,0050
0,5 T-statistic -1,3735 -0,9449 -1,1869
Significance insign insign insign
CAAR -0,0101 -0,0085 -0,0089
0,8 T-statistic -2,0161 -1,5123 -1,6169
Significance ** insign insign
CAAR -0,0134 -0,0122 -0,0132
0,10 T-statistic -2,5666 -2,0275 -2,3999
Significance ** ** **

The results show that the Market Model clearly gives more significant results than
the other two models, and that the only event window that is still significant for all
three models is the event window (0, 10). The results show negative CAARs for
all models which are statistically significant at the 95%- level, where the Market
Model has a negative result of – 1,34% , the Fama & French model has a negative
CAAR at – 1,22%, and the Carhart Model has a negative CAAR at – 1,32%. This
implies that the market values CSR negatively and one would expect to get a
negative CAAR within the time interval of 10 days after the CSR announcement.
Additionally, according to the Market Model and the Carhart Model, there are
negative statistically significant results at the 95%- level in the event window of (-

Page 29
Master Thesis GRA 19003 02.09.2013
10,10). This indicates that the market learns about the occurring event some time
before the actual event has occurred, in addition to the fact that the market values
the CSR negatively after the event.

Our analyses give both significant and insignificant results; hence, one will have
to interpret the results carefully. However, we do find evidence of significant
negative CAARs throughout the event windows, both in the 7 days and the 181
days intervals. Within the 7 day time interval, though, there were only one event
window for the Market Model that was significant at the 90%- level, whereas
within the 181 day interval we got some more and stronger statistically significant
results. Checking for significance, the graphical presentations of the results also
show that that the graphs actually drift downwards before and after the event.
These results imply that the market learns about the announcement some time
before the event actually occurs and values the CSR activities negatively. The
drifting graphs also imply that it could be considered to be an anomaly to market
efficiency. The Market Model and the Carhart Model were the two models that
overall gave most results that were significant, both in the analysis of the different
event windows and for the graph itself. The Fama & French Model had less
significant results. Hence, based on the results from the Fama & French Model,
we are not able state clear conclusions. Furthermore, due to the low significance
within the event window of 7 days, we are not able to conclude that the CAAR
represent an anomaly within a short time aspect of the CSR announcement.
However, based on the stronger results from the Market Model and the Carhart
Model within the event window of 181 days, we are able to say that the CAAR
continues to drift after the event. Therefore, we reject the hypothesis H0: CAAR
remains stable, and say that within a longer time horizon of the event, the CAAR
continues to drift. Thus, this suggests that CSR announcements represent an
anomaly to market efficiency.

5.3 Discussing the results of AAR and CAAR

Following the discussions from the previous two sections, we observe that our
analyses show that we have one positive and some negative significant average
abnormal returns, and more importantly, that we have significant negative
cumulative average abnormal returns for the Market Model and the Carhart
Model. We also observed that some results were insignificant, in addition to the
Page 30
Master Thesis GRA 19003 02.09.2013
fact that the Fama & French Model had very different results from the Market
Model and the Carhart Model.

Overall, we conclude that our results suggest that CSR efforts had a negative
impact on companies’ stock returns in the period of 2005-2012. Also, since the
Fama & French Model mainly has insignificant results while the Market Model
and the Carhart Model have much significant results, we are going to focus on the
findings of the Market Model and the Carhart Model in this section. Therefore, in
this section we will discuss potential reasons for the negative average abnormal
returns and cumulative average abnormal returns that we found. For simplicity,
we will use the term “abnormal returns” as a conjoint term for both “average
abnormal returns” and “cumulative average abnormal returns” throughout the
remainder of this section.

As mentioned in section 1.3, before performing the analyses in this thesis we

expected that CSR announcements were going to result in positive abnormal
returns, and that CSR announcements would be a positive anomaly to market
efficiency. This was mainly due to the fact that CSR signals financial stability, it
could create positive publicity, and it may attract qualified personnel to the
company. However, as indicated, our analyses showed that CSR announcements
create negative abnormal returns for a company. At first, this surprised us.
However, after having discussed the results thoroughly, we found several
potential reasons why CSR announcements result in negative abnormal returns for
a company. These reasons will be discussed in the following paragraphs in this
section.

As described in section 2; Literature Review; previous studies have found varying

results regarding the effects of CSR announcements on stock returns. Some
studies have found positive connections, some found negative connections, while
some researchers concluded that CSR efforts have no significant impact on stock
returns. For example, Arx and Ziegler found that financial markets positively
value environmental and social activities of a firm in the US and Europe in the
time period of 2003-2006, Bird et al. found that the American market seemed to
value firms that satisfied only the minimum requirements (ie as required by law)
of the dimensions of environment and diversity in the period of 1991-2003, while

Page 31
Master Thesis GRA 19003 02.09.2013
Mollet and Ziegler did not find evidence suggesting that CSR had any impact on
stock returns in the period of 1998-2009. Additionally, some studies found
altering results on other markets as well. Thus, the market seems to value CSR
differently over time, as Bird et al. also mentioned in their study. On the American
market, Arx and Ziegler found a positive effect of CSR in the period of 2003-
2006, and Mollet and Ziegler found no effect in the period of 1998-2009. And
now, we have found a negative effect in the period of 2005-2012. Thus, our
results are confirming the finding of Bird et al.; that the effects of CSR on a
company’s abnormal returns change over time. Therefore, this may suggest that
our negative results essentially is a part of an overall change in the market
valuation of CSR; the negative market perception of CSR in the period of 2005-
2012 may basically be due to the fact that the perception of CSR has changed
from the findings of Arx and Ziegler in 2009 and Mollet and Ziegler in 2012.

Another potential explanation for the negative abnormal returns lies in the fact
that CSR efforts are very expensive. Firms make substantial investments into
various projects, while these investments do not generate any directly related
returns. Thus, it may be that the investment community does not appreciate that
firms spend much money on projects that do not directly contribute to increasing
the value of the firms, and hence the CSR efforts result in negative abnormal
returns. Additionally, in the period of 2008-2009, the American economy
experienced a significant crisis where many companies ran into financial
difficulties. It may be that the presence of this crisis heavily influenced the results
that we got in our analyses, and hence contributed to creating negative abnormal
returns. In order to get an idea of how this financial crisis impacted our sample
and our tests, we investigate our results in a manner which takes the crisis into
consideration in section 5.5.

Relating to the previous argument; today, after the financial crisis in the beginning
of the 2000s and the crisis of 2008-2009, several experts claim that the American
economy is heavily influenced by macroeconomic events. Thus, firm
performances are now more influenced by macroeconomic events than they used
to be, and it may be that firm specific events currently do not have the same effect
on companies’ stock returns as they used to. Thus, this may suggest that firm
specific events such as CSR announcements do not impact companies’ abnormal

Page 32
Master Thesis GRA 19003 02.09.2013
returns in the manner that we would expect them to, and it suggests that the
negative returns that the firms are experiencing are moreover related to the state of
the economy and general events in the firms’ market place. Thus, it may be that
we observe negative abnormal returns out of coincidence or relations to the macro
economy, rather in direct relations to the CSR announcements.

Another possible explanation for the negative abnormal returns may lie in the
types of projects that firms invest in. In general, firms invest in CSR projects of
numerous different types. Oftentimes, firms invest in projects that are directly
related to their own industry; for example when DELL Inc. donates $10 million to
education technology initiatives across the country. However, oftentimes we
observe that firms invest in CSR projects that are not directly related to their own
field at all; such as when Walt Disney donates $1 million to “various
environmental causes”. When the companies invest in CSR projects that are not
directly related to their own industry, they run the risk of being accused of doing
CSR “only for creating positive publicity and higher returns” rather than for the
“good will”. Investors may not value that the firms is spending money that do not
directly benefit the industry in which the company operates. This may cause the
CSR effort to have a negative effect on a company’s stock returns, rather than a
positive effect. Furthermore, firms also experience a tax relief when spending
money on CSR. Thus, if it seems that certain firms are doing CSR essentially to
experience tax benefits rather than purely doing it for “the greater good”, this is
also likely to cause the CSR efforts to have a negative impact on a company’s
abnormal returns.

The specific dimensions of CSR in which the companies invest may also
influence the impact that a company’s CSR efforts have on its stock returns. As
mentioned in section 1.1, there are essentially five main dimensions of CSR into
which companies can invest; these five being the environmental, social, economic,
stakeholder and voluntariness dimensions. As previously described, we chose to
focus on the two dimensions of environment and social CSR efforts. Essentially,
we chose to focus on these two types of dimensions, as these seem to be the two
most common types of CSR dimensions. However, the fact that we are looking at
the dimensions of environment and social may have a direct impact on the fact
that we observe negative abnormal returns in our analyses. Previous studies

Page 33
Master Thesis GRA 19003 02.09.2013
disagree on the effects of these two dimensions. Bird et al. (2007) found that the
market valued the various dimensions differently. Furthermore, they found that
for the time period of 1991-2003, the market punished firms that exceeded
minimum investment in environmental CSR efforts, while the social dimension
was no longer being valued. On the contrary, Arx and Ziegler (2009) found that
the market valued environmental and social CSR efforts in the period of 2003-
2006. Thus, these researchers found opposite results, and our results seem to
support the findings of Bird et al., which indicates that it may be that the trend of
“punishing” CSR efforts in environmental and social dimensions still remains.
Thus, it may be that we obtain negative abnormal returns in our analyses due to
the fact that we have focused on social and environmental efforts, and it may be
that we had observed different results if we had focused on CSR announcements
within the dimensions of economic, stakeholder and voluntary CSR efforts.

In our study, the event window for a “longer term effect” starts 90 days prior to,
and ends 90 days after the CSR announcements. However, this event window may
be too short in order to be able to observe the true effect of the CSR
announcement. As Arx and Ziegler (2009) mention, it may take longer time for
the true effect of a CSR project to reveal itself for a company. The full process of
a CSR project may last for a long period of time; it may take up to several years
from the day that the CSR project is first planned until the day that the project is
completed. Thus, when we look at the effect of the CSR announcements for the
following 90 days only, we are not able to include the effects of the completed
CSR activity. Perhaps the market has a negative reaction towards the CSR effort
at the day when the announcement is made and the time right after, since the
market essentially just realizes that the firm is going to spend a significant amount
of money on something that is not generating any direct returns. Furthermore;
perhaps the market instead values a CSR effort to a greater extent when the
project is either started, is starting to show good results, or has been completed as
the market is able to observe the good effects of the CSR project. Thus, the fact
that we are looking at the CSR announcements themselves in this study; and have
chosen an event window of maximum 181 days; we do not have a sufficient
window to capture the complete effect of the whole CSR process. Thus, this may
be a reason why we observe negative abnormal results from our analyses.

Page 34
Master Thesis GRA 19003 02.09.2013
In summary, there are several potential reasons explaining why our analyses show
that CSR announcements result in negative abnormal returns. Some suggested
explanations may relate to a somewhat general macroeconomic picture, whereas
some explanations are moreover related to the specifications of our methodology.
However, despite our suggested reasons, one can find good arguments supporting
intuitions behind both a positive and a negative relationship between a firm’s CSR
announcements, and its abnormal returns. Thus, the empirical data is what really
matters; and the data for this particular thesis suggest that there is a negative
relationship between a firm’s CSR announcements and its abnormal returns.

5.4 Robustness Tests

After having performed all the tests in our analyses, we wanted to test the
robustness of our results. We performed three different robustness tests to see how
our results and significance levels changed with some variations in the data. In the
first test we eliminated the two CSR announcements with the highest dollar value,
and the two CSR announcements with the lowest dollar value. In the second test
we eliminated the two announcements which got the highest market betas from
our regressions, and the two announcements with the lowest market betas. In the
third test we eliminated the three announcements with the highest standard errors
from our regressions. We alternated the t-statistics test by applying corresponding
degrees of freedom to the amount of remaining announcements in the data set. We
performed the robustness tests on the Carhart Model, since this is the most
extensive model in our analysis, obtaining the highest R2.

The robustness tests caused very few changes in the results of our analyses, and
exhibit 3 is an excerpt from the results that we obtained. It corresponds to table 4
in section 5.2, and thus presents the CAAR analysis of the longer term event
windows. It shows the CAARs, t-statistics and significance levels from the
original Carhart analysis, as well as the corresponding values from the three
robustness tests. As presented in the exhibit, there is a small number of changes.
Some event windows increase their significance, while some decrease their
significance. Overall, the robustness tests show somewhat more significant results
than our original results. However, the changes caused by the robustness tests are
so small such that they cannot change the overall findings and analyses presented

Page 35
Master Thesis GRA 19003 02.09.2013
in the previous subsections. Thus, we conclude that the robustness tests confirm
the overall findings in our thesis.

5.5 Comparing AAR and CAAR before, during and after the 2008 crisis
In order to get an impression of how our overall results are influenced by the 2008
financial crisis, we performed a new analysis where we split all our CSR
announcements into three groups according to times of expansion and
contractions in the American Economy from 2005-2012. Following the business
cycle reference dates from the National Bureau of Economic Research (NBER
2013), we defined the period of January 2005 until November 2007 as a period of
expansion, the period of December 2007 until June 2009 as a period of
contraction, and the period of July 2009 until December 2012 as a period of
expansion. Therefore, the announcements in Period 1, 2 and 3 correspond to the
three periods mentioned above, respectively. Period 1 includes 21 announcements,
Period 2 includes 15 announcements, and Period 3 includes 32 announcements.
Since the periods contain relatively few data points, we must interpret the results
with some caution. We performed this analysis using the Carhart Model, since this
model had the highest R2 in our original tests. Below are the graphs depicting the
CAARs for the three different periods for the -3,3 and the -90,90 intervals:

Graph 3: CAARs -3,3 Days for Different Time Periods

Page 36
Master Thesis GRA 19003 02.09.2013
Graph 4: CAARs -90,90 Days for Different Time Periods

The results that came out from this test were somewhat surprising. As shown,
period 1 of expansion; 2005-2007; has the lowest CAAR, period 2 of financial
crisis; 2007-2009; has the highest CAAR, while period 3 of expansion; 2009-2012
has its CAAR in between of the other two. When performing the significance tests
of the three periods, we found that all three periods are insignificant in the -3,3
interval, period 2 and 3 are insignificant in the -90,90 interval, and period 1 is
significant at the 99% level in the -90,90 interval.

Most of all, it is surprising that the period of financial crisis has the highest
CAAR, and that period 1 of expansion has the lowest CAAR. Although most of
these CAARs were insignificant, we still think it is interesting to provide some
intuition for why we see this particular distribution of the CAARs. After all, this is
the underlying distribution for the all the data points in our overall study. Firstly,
the firms that are divided into the three periods are not the same. Ideally, all three
periods would include the same firms, such that we would be able to compare firm
specific performances across the three periods. However, given that we have 68
CSR announcements from 39 different firms, it would not be possible to create
three periods that contains the same firms. Secondly, although the economy was
in a financial crisis in period 2, there were in fact some firms that performed well
during this time. Thus, it may be that we have caught some of these firms in our
sample. Likewise, some firms did not perform well in the two periods of
expansion, and it may be that some of the firms in our sample were among those.
Graph 5 shows the distribution of the market return from 2005-2012, as provided
by Kenneth French. This distribution does indeed show that each period
Page 37
Master Thesis GRA 19003 02.09.2013
experienced positive and negative returns in the market, which implies that
specific firms may have experienced much larger swings.

Graph 5: Market Return 2005-2012

Another reason why we see the surprising CAARs in this analysis may lie in the
way that the abnormal returns are actually calculated. Abnormal returns are really
“actual returns – estimated returns”. Thus, in the periods of expansion (2005-07
and 2009-12), one expects and estimates returns to be higher, and then it may be
harder for the actual returns to exceed the expectations, and hence there is
negative abnormal returns. Conversely, in the period of crisis, one expects lower
estimated returns; perhaps even negative; and then if a company actually has some
positive returns, it may be “easier” to exceed what was expected. Additionally, the
fact that companies are able to do CSR efforts in times of crisis, may have a
particularly positive effect on a company’s stock returns in period 2. First, this
signals that the company is in a financial position which allows it to set aside
funding for something that does not generate directly related income. Second,
instead of exploiting additional funds as dividends or other capital investments,
the firm is thinking about the greater good and is helping out society in times of
difficulties. These two points may support the notion that CSR creates positive
abnormal returns in times of crisis.

Lastly, and perhaps the most important reason why we see the surprising CAARs,
may lie in the size of our data sample. Overall, we have 68 CSR announcements
in our sample, which is a decent number in order to make analyses of the whole
period of 2005-2012. However, when we split these announcements into three
groups, some of the samples are very small. Period 1, 2 and 3 contains 21, 15 and
32 announcements, respectively. Thus, the periods have fairly small and
asymmetric sample sizes, which furthermore can contribute to obscuring the

Page 38
Master Thesis GRA 19003 02.09.2013
results. Thus, this makes it more difficult both to rely on the results that we get in
each CAAR, as well as the comparisons among each period. Additionally, the
small sample size is probably also contributing to making most of the results from
this analysis insignificant. This helps us realize that the main benefit from doing
this analysis of separating between the three periods relating to the cycles of the
economy, is to understand the reason why the overall CAAR graph from section
5.2 is the way it is. We plotted the overall CAAR from graph 2 in section 5.2 into
a plot together with the CAARs from the three different periods, depicted in the
following graph:

Graph 6: CAARs for all Three Periods, and Overall CAAR

The thick, black line is the original overall CAAR from section 5.2. As shown in
graph 6, it is very similar to the CAAR graph from period 3; the period of 2009-
2012. This is mainly due to the fact that the sample size for this period is much
larger than the sample sizes from the other two periods. It does lie somewhat
above the CAAR for period 2, which is due to the high and positive CAAR from
period 1. Essentially, this tells us that our overall CAAR is heavily influenced by
the CSR announcements that we collected from the period of 2009-2012. Thus,
this contributes to explaining why we observe a negative overall CAAR in our
analyses of the effects of CSR announcements on a company’s stock returns, since
we now know that the CAAR in the period of 2009-2012 was negative.

The findings from this interim analysis may also support findings in previous
literature, where Bird et al. (2007) suggested that the market’s attitude of CSR
Page 39
Master Thesis GRA 19003 02.09.2013
changes over time. The CAARs from our three defined periods vary from each
other; one being very negative, one being relatively negative one being very
positive. Furthermore, this analysis also supports the fact that different previous
studies come to different conclusions. Some find that CSR has a positive effect on
stock returns, some find a negative effect, and some find that there is no effect at
all. Thus, even though the CAARs for the three defined periods generally are not
significant, they may still be related to findings of previous studies.

Page 40
Master Thesis GRA 19003 02.09.2013

6. Conclusion
In this thesis we have studied the relationship between a company’s CSR
activities and their effects on a company’s stock returns. More specifically, we
have researched whether CSR announcements create abnormal returns for a
company, and if such announcements may be considered an anomaly to market
efficiency. We focused on CSR announcements in the American market during
the time period of 2005-2012, and we had a total of 68 announcements in our
analysis. Furthermore, we used both a short term and a longer term event window
of 7 days and 181 days in our analysis. We performed a detailed study using the
Market Model, the Fama & French Model and the Carhart Model, and we focused
on the findings from the Market Model and the Carhart Model as these two
models had much more significant results than the Fama & French Model. Based
on our analyses, we came to the conclusion that CSR announcements have a
negative impact on a company’s abnormal return, as well as on a company’s
cumulative abnormal returns. The CAARs in our analyses continue to drift
negatively, and based on these findings we suggest that CSR announcements
represent an anomaly to market efficiency.

There are several potential explanations for these negative results. For example,
the market’s valuation of CSR changes over time and the period of 2005-2012
seem to be a period which does not value CSR; CSR efforts are expensive; CSR
activities may create poor publicity if a company is not doing CSR only for the
good will but rather for the pure interest of publicity, returns or tax reliefs; it may
be that social and environmental CSR dimensions (which we have focused on in
this thesis) are negatively related to a firm’s stock returns; or it may be that
markets do not value the CSR announcements themselves (which we focus on in
this thesis), but rather the performance and completion of the CSR activity.
Additionally, it may be that the financial crisis from December 2007 until June
2009 had a strong impact on our results. We attempted to test this in section 5.5,
however, we mostly obtained insignificant results in our analyses and hence could
not make a complete interpretation of the analysis with respect to the financial
crisis.

Page 41
Master Thesis GRA 19003 02.09.2013
Previous literature does not provide a completely united view on the effect of CSR
activities on a firm’s stock returns. Some find a positive relationship between
CSR and a firm’s stock returns, whereas others find either a negative relationship
or no relationship at all. Our study is unique in the way that we combine the three
main specifications of our study; ie that we have a time frame of 2005-2012 on the
American market, that we look at CSR announcements, and that we apply the
Market Model, the Fama & French Model and the Carhart Model. We think that
our results and our thesis as a whole add value to the existing base of literature on
the topic of the effect of CSR on a firm’s stock returns.

After having completed the analyses in this thesis, we have also observed some
weaknesses of our study. First and foremost, the strength of the study would have
increased if we had been able to find more announcements that could go into the
study. Although a sample size of 68 CSR announcements is a decent amount, the
trustworthiness of our results would increase with more announcements. Also,
some of our abnormal returns and cumulative abnormal returns are not significant;
and this could potentially change if the sample size increased. Additionally, it is
somewhat puzzling that the Fama & French Model gives such different results
from the Carhart Model, and a larger sample size could potentially have
contributed to changing this.

Furthermore, the criteria that we determined for what announcements to include in

the study have eliminated some announcements from the sample. If we had
increased the range of funding into the project, opened up for more CSR
dimensions to be included, allowed for more than three announcements to be
included per firms, etc., we could have obtained a larger sample size. However,
we found it important to have some clearly defined criteria for our sample, such
that we would have the highest possibility of making valid and market-wide
inferences from our results. But, it seems as though we could have benefited from
having somewhat looser criteria, and instead had a larger sample size.

Additionally, the fact that we are using CSR announcements in our study; and not
another measure such as CSR ratings; may be both a strength and a weakness to
our study. It contributes to making our study unique since hardly any studies have
done this in the past, and it enables us to make an inference of whether CSR

Page 42
Master Thesis GRA 19003 02.09.2013
announcements may be an anomaly to market efficiency. On the contrary, it may
be a weakness because the market might value the start or completion of a project,
and not the announcement about a CSR project itself. Also, we do not know how
much of a “surprise” this announcement is. It may be that such CSR projects; or
potentially CSR budgets; are mentioned in some firms’ annual reports. Thus, if
this is the case, then the information that is released on the CSR announcement
day is not actually “new” information, and may already be incorporated into the
firm’s stock price. Hence, this may indicate that we could have observed stronger
results if we had chosen to look at the effects on firms’ performance in relations to
their CSR ratings in various databases, instead of looking at the effects of the CSR
announcements on firms’ stock returns.

Through this study, we find a negative relationship between a firm’s CSR

announcements and its abnormal returns, and we also suggest that CSR
announcements represent an anomaly to market efficiency. We had expected to
find the opposite; that CSR has a positive impact on a firm’s stock returns. The
world is dependent on massive CSR efforts from companies around the globe, and
it would likely have been more beneficial for society if we had found a positive
relationship. If there is a positive relationship between CSR and a firms’ stock
returns, it may be easier to motivate more firms to perform CSR. Thus, proving a
positive relationship may contribute to increasing the overall CSR efforts that
firms perform, which furthermore could help improving various issues and
challenges all over the world. As mentioned, it may be that we had found different
results in this thesis by using different specifications in our study. In line with this,
we think that an interesting topic for future research would be to analyze the
effects of finished CSR projects on the value of companies. In other words,
instead of looking at the effects of CSR announcements presenting a new CSR
project (like we have done in this thesis), it would be interesting to know how the
value of a company changes when the completion of a large CSR project is
announced. Perhaps the market would value the results of CSR projects
positively, and hence suggest that CSR projects have a positive influence on the
value of a company.

After having performed this study and having seen our results, another interesting
question strikes us. The amount of CSR efforts worldwide; and in the US in

Page 43
Master Thesis GRA 19003 02.09.2013
particular; have grown significantly over the past few decades, and most national
and multinational American companies now have a well-established CSR
program. Therefore, it seems that CSR efforts have become something that now is
expected by the market, rather than something that companies do voluntarily. In
other words, would the negative abnormal results that we found in this study be
even more negative, if the firms did not perform any CSR activities at all? Has
CSR developed into an expectation from the market rather than a motivation from
the company to perform a good will to society? We think that this question also
represents an interesting topic of future research. It would be interesting to see a
research that identifies companies that do not perform CSR activities, and analyze
how these companies perform compared to firms that do perform CSR activities.
Hopefully, for the greater good of our society, this study could prove that firms
that do not perform CSR activities underperform compared to the firms that do
perform CSR.

After having completed this thesis, we have gained useful economic insight and
our awareness of the immense importance and magnitude of firms’ CSR efforts
have increased. Despite the fact that our results suggest that CSR announcements
have a negative impact on companies’ stock returns, we hope that companies will
maintain their CSR investments in the future.

Page 44
Master Thesis GRA 19003 02.09.2013

7. Bibliography

Arya, Bindu and Gaiyan Zhang. 2008. Institutional Reforms and Investor
Reactions to CSR Announcements: Evidence from an Emerging Economy.
USA: University of Missouri- St. Louis. In text (Arya and Zhang 2008).

Arx, Urs von and Andreas Ziegler. 2009. The Effect of Corporate Social
Responsibility on Stock Performance: New Evidence for the USA and
Europe. Switzerland: Zürich. In text: (Arx and Ziegler 2009).

Barber, Brad M. and John D. Lyon. 1997. Detecting long-run abnormal stock
returns: The empirical power and specification of test statistics. Journal of
Financial Economics 43 (3) : 341-372. [Link]/10.1016/s0304-
405x(96)00890-2. In text: (Barber and Lyon 1997).

Bird, Ron, Anthony D. Hall, Fransesco Momentè and Fransesco Reggiani. What
Corporate Social Responsibility Activities are Valued by the Market?
Journal of Business Ethics76 (2) :189-206. DOI 10.10.1007/s10551-006
9268-1. In text:(Bird [Link] (2007)).

Bodie, Zvi, Alex Kane and Alan J. Marcus. 2011. Investments and Portfolio
Management. 9th ed. McGraw-Hill Irwin. In text: (Bodie, Kane and
Marcus2011).

Brammer, Stephen, Chris Brooks and Stephen Pavelin. 2006. Corporate Social
Performance and Stock returns: UK Evidence from Disaggregate
Measures. Financial Management 35 (3): 97- 116. In text: (Brammer,
Brooks and Pavelin (2006)).

Brooks, Chris. 2008. Introductory Econometrics for Finance. 2nd ed. USA,NY:
Cambridge University Press. In text: (Brooks 2008).

Cellier, Alexis and, Pierre Chollet. 2011. The Impact of Corporate Social
Responsibility Rating Announcement on Stock Price: An Event Study on

Page 45
Master Thesis GRA 19003 02.09.2013
European Markets. Paris: Universitè Paris- Est.. In the text: (Cellier and
Chollet 2011).

Chen, Honghui and Xiayang Wang. 2011. Corporate social responsibility and
corporate financial performance in China: An empirical research from
Chinese firms. Corporate Governance 11 (4): 361 – 370. In text: (Chen
and Wang 2011).

Cheung, Yan Leung, Weiqiang Tan, Hee-Joon Ahn and Zheng Zhang. 2009.
Does Corporate Social Responsibility Matter in Asian Emerging Markets?
Journal of Business Ethics 92: 401-413. DOI 10.1007/s10551-009-0164-3.
In text: (Cheung [Link] 2009).

CSRwire. 2013. The Corporate Social Responsibility Newswire. Accessed:

January-March 2013. < [Link] (In text: [Link]).

Dahlsrud, Alexander. 2006. How Corportae Social Responsibility is Defines: an

Analysis of 37 Definitions. Wiley Inter Science DOI: 10.1002/csr.132. In
text: (Dahlsrud 2006).

European Commission 2012. Sustainable and responsible business; Corporate

Social Responsibility (CSR). Accessed December 28, 2012, 12.05 o’clock.
<[Link]
social- responsibility/index_en.htm>. In text: (EU Comission2012).

French, Kenneth R. 2013. Kenneth R. French Website. Accessed:

January-April 2013.
<[Link]
ml>. (In text: Kenneth French website).

Galbreath, Jeremy and Paul Shum. 2012. Do customer satisfaction and reputation
mediate the CSR-FP link? Evidence from Australia. Australian Journal of
Management 2012 37: 211. DOI:10.1177/0312896211432941. In text:
(Galbreath and Shum 2012).

Page 46
Master Thesis GRA 19003 02.09.2013
Investopedia. 2012. Corporate Social Responsibility. Accessed: October 28. 2012,
15.10 o’ clock. <[Link]
[Link]#axzz2GRfHCmn4>. In text: (Investopedia 2012)

Iqbal, Nadeem , Naveed Ahmad, Nauman Ahmad Basheer and Muhammad

Nadeem. 2012. Impact of Corporate Social Responsibility on Financial
Performance of Corporations: Evidence from Pakistan. International
Journal of Learning & Development 2 (6). DOI:
10.5296/ijld.v2i6.2717. In text: (Iqbal [Link] 2012).

Mollet, Janick Christian and AndreasZiegler. 2012. Is Socially Responsible

Investing Really Beneficial? New Empirical Evidence for the US and
European Stock Markets. Switzerland: ETH Zûrich. In text: (Mollet and
Ziegler (2012)).

National Bureau of Economic Research. 2013. US Business Cycle Expansions and

Contractions. Accessed: February 22. 2013, 10.03 o’cock.
<[Link] (In text: NBER 2013)

Newell, Graeme and Chyi Lin Lee. 2012. Influence of the corporate social
responsibility factors and financial factors on REIT performance in
Australia. Journal of Property Investment & Finance 30 (4): 389-403.
DOI:10.1108/14635781211241789. In text: (Newell and Lee 2012).

NIST. 2012. Engineering Statistics Handbook. Accessed: February 15. 2013,

12.15 o’clock.
<[Link]
(In text: Exhibit 1. Source: [Link]).

Wang, Yungchih George. 2011. Corporate Social Responsibility and Stock

Performance- Evidence from Taiwan. Modern economy 2011 (2): 788
799. DOI: 10.4236/me.2011.25087. In text: (Wang 2011).

Yahoo! Finance. 2013. Yahoo! Finance. Accessed: January-April 2013.

< [Link] (In text: [Link]).

Page 47
Master Thesis GRA 19003 02.09.2013

8. Appendix: Exhibits

Exhibit 1; T- statistics

Student T- distribution with ν- degrees of freedom

90 % 95 % 99 %
α (Two- tailed) 0,1 0,05 0,01
ν Degrees of Freedom
66 1,6683 1,9966 2,6524
Source: [Link]

Exhibit 2.1; Descriptives Table; All Companies

Exhibit 2.1
Date of Investment
Firm Founded Sector Firm Industry S&P500
announcement into CSR Project
Abbott Laboratories (ABT) 1888 Healthcare Medical appliances and equipment Yes 19.01.2010 $ 2 500 000
Abbott Laboratories (ABT) 1888 Healthcare Medical appliances and equipment Yes 24.07.2012 $ 1 000 000
Adobe Systems, via Adobe Foundation (ADBE) 1982 Technology Application software Yes 21.06.2006 $ 10 000 000
American Eagle Outfitters (AEO) 1977 Services Apparel stores No 01.04.2005 $ 1 000 000
American International Group, Inc. (AIG) 1919 Financial Property and casualty insurance Yes 31.07.2007 $ 5 000 000
AT&T (T) 1875 Technology Telecom service - domestic Yes 17.09.2008 $ 29 000 000
Avon Products (AVP) 1886 Consumer goods Personal products Yes 05.09.2012 $ 1 000 000
Baxter International Inc. (BAX) 1931 Healthcare Medical instruments and supplies Yes 23.10.2008 $ 5 000 000
Becton, Dickonson and Company (BDX) 1897 Healthcare Medical instruments and supplies Yes 03.01.2005 $ 1 500 000
Becton, Dickonson and Company (BDX) 1897 Healthcare Medical instruments and supplies Yes 27.01.2010 $ 5 000 000
Cisco Systems, Inc. (CSCO) 1984 Technology Networking and communication devices Yes 01.07.2008 $ 45 000 000
Cisco Systems, Inc. (CSCO) 1984 Technology Networking and communication devices Yes 10.06.2010 $ 1 000 000
Coca Cola (KO) 1892 Consumer goods Beverages - softdrinks Yes 16.03.2009 $ 30 000 000
Coca Cola (KO) 1892 Consumer goods Beverages - softdrinks Yes 22.05.2011 $ 6 000 000
CSX Corportaion 1827 Services Railroads Yes 17.10.2011 $ 8 275 000
CVS Caremark (CVS) 1892 Services Drugstores Yes 05.02.2008 $ 5 500 000
CVS Caremark (CVS) 1892 Services Drugstores Yes 14.01.2010 $ 6 300 000
CVS Caremark (CVS) 1892 Services Drugstores Yes 13.09.2011 $ 6 000 000
Darden Restaurants Inc (DRI) 1968 Services Restaurants Yes 30.04.2012 $ 1 700 000
Dell Inc. (DELL) 1984 Technology Personal computers Yes 22.09.2010 $ 10 000 000
Dow Chemical Company (DOW) 1897 Basic Materials Chemicals - major diversified Yes 06.03.2006 $ 2 000 000
Dow Chemical Company (DOW) 1897 Basic Materials Chemicals - major diversified Yes 20.05.2008 $ 2 000 000
Dow Chemical Company (DOW) 1897 Basic Materials Chemicals - major diversified Yes 04.08.2011 $ 2 000 000
ExxonMobil (XOM) 1859 Basic Materials Major integrated oil and gas Yes 02.05.2007 $ 33 000 000
FedEx (FDX) 1998 Services Air delivery and freight services Yes 04.08.2011 $ 5 375 000
GE (GE) 1892 Industrial goods Diversified machinery Yes 24.10.2005 $ 20 000 000
GE (GE) 1892 Industrial goods Diversified machinery Yes 30.06.2008 $ 18 000 000
GE (GE) 1892 Industrial goods Diversified machinery Yes 15.01.2010 $ 2 500 000
Goldman Sachs (GS) 1869 Financial Diversified investments Yes 11.04.2005 $ 1 700 000
Goldman Sachs (GS) 1869 Financial Diversified investments Yes 30.05.2007 $ 2 000 000
IBM 1911 Technology Information Technology Services Yes 24.10.2011 $ 50 000 000
Intel Corporation (INTC) 1968 Technology Semiconductor - broadline Yes 12.05.2006 $ 4 000 000
ITT 1920 Industrial goods Diversified machinery No 07.09.2010 $ 10 500 000
JCPenny (JCP) 1902 Services Department stores Yes 03.11.2008 $ 9 000 000
JCPenny (JCP) 1902 Services Department stores Yes 13.01.2011 $ 6 700 000
Johnson Controls Inc (JCI) 1885 Consumer goods Autoparts Yes 01.09.2005 $ 1 000 000
Kohl's (KSS) 1988 Services Department stores Yes 29.03.2012 $ 2 000 000
Marriott (MAR) 1927 Services Lodging Yes 07.04.2008 $ 2 000 000
MasterCard (MA) 1966 Services Business services Yes 26.04.2012 $ 19 500 000
Men's Wearhouse (MW) 1973 Services Apparel stores No 16.02.2011 $ 1 000 000
Merck Company (MRK) 1891 Healthcare Drug Manufacturers - major Yes 19.11.2007 $ 2 800 000
Merck Company (MRK) 1891 Healthcare Drug Manufacturers - major Yes 19.07.2012 $ 3 000 000
MGM Resorts International (MGM) 2000 Services Resorts and casinos No 14.12.2011 $ 4 600 000
Microsoft (MSFT) 1975 Technology Application software Yes 21.02.2006 $ 41 000 000
Motorola Solutions (MSI) 1928 Technology Communication equipment Yes 10.04.2006 $ 1 000 000
Motorola Solutions (MSI) 1928 Technology Communication equipment Yes 14.11.2007 $ 3 500 000
Motorola Solutions (MSI) 1928 Technology Communication equipment Yes 19.01.2010 $ 7 500 000
Nike (NKE) 1964 Consumer goods Textile - apparel footwear and accesories Yes 27.05.2008 $ 45 000 000
Nike (NKE) 1964 Consumer goods Textile - apparel footwear and accesories Yes 30.11.2010 $ 6 000 000
PepsiCo (Foundation) (PEP) 1965 Consumer goods Beverages - softdrinks Yes 22.01.2008 $ 2 500 000
PepsiCo (Foundation) (PEP) 1965 Consumer goods Beverages - softdrinks Yes 14.01.2010 $ 1 000 000
Philip Morris International (PM) 1847 Consumer goods Cigarettes Yes 15.03.2011 $ 1 200 000
Starbucks Corporation (SBUX) 1985 Services Speciality eateries Yes 15.03.2007 $ 2 000 000
Target Corporation (TGT) 1962 Services Discount, variety stores Yes 19.03.2006 $ 15 600 000
Target Corporation (TGT) 1962 Services Discount, variety stores Yes 11.05.2011 $ 3 500 000
The Hartford financial services group (HIG) 1810 Financial Property and casualty insurance Yes 15.07.2007 $ 1 000 000
Union Pacific (UNP) 1862 Services Railroads Yes 03.03.2011 $ 7 400 000
United Technologies (UTX) 1929 Industrial goods Aerospace/defense products and services Yes 10.07.2008 $ 1 000 000
UPS (UPS) 1907 Services Air delivery and freight services Yes 11.11.2009 $ 1 000 000
UPS (UPS) 1907 Services Air delivery and freight services Yes 01.02.2012 $ 6 000 000
Wal-Mart Stores (WMT) 1962 Services Department stores Yes 10.12.2008 $ 25 000 000
Wal-Mart Stores (WMT) 1962 Services Department stores Yes 23.03.2011 $ 5 000 000
Walt Disney (DIS) 1923 Services Entertainment - diversified Yes 01.09.2005 $ 2 500 000
Walt Disney (DIS) 1923 Services Entertainment - diversified Yes 14.05.2007 $ 12 500 000
Walt Disney (DIS) 1923 Services Entertainment - diversified Yes 14.05.2009 $ 1 000 000
Well's Fargo (WFC) 1852 Financial Money center banks Yes 06.04.2006 $ 5 000 000
Well's Fargo (WFC) 1852 Financial Money center banks Yes 15.02.2008 $ 1 500 000
Western Union (WU) 1851 Financial Credit services Yes 22.12.2011 $ 3 400 000

Page 48
Master Thesis GRA 19003 02.09.2013

Exhibit 2.2; Descriptives Chart; Sector Distribution

Page 49
Master Thesis GRA 19003 02.09.2013

Exhibit 3; Robustness Tests

Exhibit 4
Event Original Values, Test 1: Test 2: Test 3:
Statistic
Window Carhart Model Announcement Value Market Beta Standard Error
-90,90 CAAR -0,1449 -0,1589 -0,1383 -0,1426
T-statistic -2,4120 -2,5227 -2,2068 -2,5171
Significance ** ** ** **
-90,0 CAAR -0,0290 -0,0308 -0,0241 -0,0252
T-statistic -1,5622 -1,576 -1,266 -1,4033
Significance insign insign insign insign
0,90 CAAR -0,0441 -0,0469 -0,0425 -0,0481
T-statistic -1,9583 -1,967 -1,8145 -2,0814
Significance * * * **
-75,75 CAAR -0,0466 -0,0457 -0,0401 -0,0461
T-statistic -2,1019 -1,9495 -1,7897 -2,1387
Significance ** * * **
-75,0 CAAR -0,0209 -0,0213 -0,0167 -0,0172
T-statistic -1,4935 -1,4371 -1,1471 -1,2345
Significance insign insign insign insign
0,75 CAAR -0,0293 -0,0282 -0,0271 -0,0324
T-statistic -1,6444 -1,4945 -1,5102 -1,812
Significance insign insign insign *
-50,50 CAAR -0,0398 -0,041 -0,0377 -0,0406
T-statistic -2,3384 -2,2656 -2,2249 -2,7145
Significance ** ** ** ***
-50,0 CAAR -0,0164 -0,0169 -0,0144 -0,0128
T-statistic -1,5690 -1,5249 -1,306 -1,3962
Significance insign insign insign insign
0,50 CAAR -0,0271 -0,0279 -0,0271 -0,0312
T-statistic -1,9822 -1,9268 -2,0271 -2,3368
Significance * * ** **
-30,30 CAAR -0,0187 -0,0186 -0,0163 -0,0169
T-statistic -1,6782 -1,619 -1,4211 -1,5628
Significance * insign insign insign
-30,0 CAAR -0,0069 -0,0075 -0,004 -0,0025
T-statistic -0,9113 -0,9686 -0,5313 -0,3564
Significance insign insign insign insign
0,30 CAAR -0,0155 -0,0149 -0,0161 -0,0179
T-statistic -1,4690 -1,3485 -1,5372 -1,7252
Significance insign insign insign *
-20,20 CAAR -0,0025 -0,0032 -0,0029 -0,0032
T-statistic -0,2681 -0,3195 -0,2982 -0,3571
Significance insign insign insign insign
-20,0 CAAR 0,0005 0,0006 0,0018 0,0029
T-statistic 0,0843 0,1 0,2813 0,5012
Significance insign insign insign insign
0,20 CAAR -0,0067 -0,0076 -0,0084 -0,0096
T-statistic -0,7678 -0,8265 -0,966 -1,1111
Significance insign insign insign insign
-10,10 CAAR -0,0147 -0,0163 -0,015 -0,0139
T-statistic -2,1666 -2,3155 -2,0895 -2,153
Significance ** ** ** **
-10,0 CAAR -0,0052 -0,0056 -0,004 -0,0047
T-statistic -1,0397 -1,1334 -0,7639 -0,9378
Significance insign insign insign insign
0,10 CAAR -0,0132 -0,0145 -0,0147 -0,0127
T-statistic -2,3999 -2,5233 -2,5553 -2,3102
Significance ** ** ** **

Page 50
Master Thesis GRA 19003 02.09.2013

9. Appendix: Preliminary Master Thesis

Page 51
Master Thesis GRA 19003 02.09.2013

Preliminary Master Thesis

- The Effect of Corporate

Social Responsibility
Announcements on a
Company’s Stock Returns
and the Efficient Market
Hypothesis -

Hand-in date:
15.01.2013

Campus:
BI Oslo

Examination code and name:

GRA 1900 Master Thesis

Programme:
Master of Science in Business & Economics, Major Finance

Page 52
Master Thesis GRA 19003 02.09.2013

Table of Content
ABSTRACT ............................................................................................................................... 54

1. INTRODUCTION ............................................................................................................ 55

1.1 CORPORATE SOCIAL RESPONSIBILITY ............................................................................. 56

1.2 MARKET EFFICIENCY ..................................................................................................... 56
1.3 DISCUSSING THE THESIS ................................................................................................. 57
1.4 THE STRUCTURE OF THIS THESIS ..................................................................................... 59

2. RELATED LITERATURE .............................................................................................. 59

3. METHODOLOGY ........................................................................................................... 62

4. DATA ................................................................................................................................ 67

5. BIBLIOGRAPHY ............................................................................................................ 70

Page 53
Master Thesis GRA 19003 02.09.2013

Abstract
In this thesis we will study the effects of Corporate Social Responsibility (CSR)
announcements on a company’s stock returns. We will focus on announcements
among American corporations from 2005-2012. We will perform an event study
where we use CAPM and Fama and French regressions, and we will look for
abnormal returns on a firm’s stock returns. We will furthermore investigate
whether the potential the excess returns can be considered an anomaly to market
efficiency. In the event study we will look at the effects of different types of CSR
announcements, separating between environmental and social CSR efforts. We
expect to see results indicating that a corporation’s CSR announcements create
abnormal returns, and will be an anomaly to market efficiency. This is essentially
because CSR programs create positive publicity for a company, they signal financial
stability to the investment community, and CSR programs may recruit new and
talented employees which can increase the overall quality of the company.

Page 54
Master Thesis GRA 19003 02.09.2013

1. Introduction
Over the last decade(s), companies all over the world have increased their
Corporate Social Responsibility (CSR) efforts. Many companies take the
responsibility of contributing to creating a better environment, reducing global
warming, improving health- and social situations for poor countries, improving
the working conditions for employees, etc., very seriously. Therefore, these
companies have established comprehensive CSR programs to combat with many
of the serious issues and problems that the world is dealing with; both on a local
and global level. However, these CSR plans do not only help improving the
world; oftentimes they also increase the value of the respective companies. When
a large company has made an announcement about a significant prospective CSR
investment, it has been reported that these announcements have led to increasing
stock returns for the company (Arx and Ziegler, 2009. Cellier and Chollet, 2011).
However, is this really the case? Do CSR announcements actually increase stock
returns of a company, even though these CSR programs oftentimes cost the
business a large amount of money without generating directly related revenues? In
this thesis we will investigate the relationship between companies’ CSR
announcements and the development of the stock returns of these companies.
More specifically, we will examine whether the firms experience abnormal returns
which can be related to the CSR announcement. We will see whether historic data
from 2005-2012 can show that CSR announcements create abnormal returns, and
if they may be considered an anomaly to market efficiency. Thus, the research
question for this thesis is as follows:
“Does a Corporate Social Responsibility announcement among American
multinational companies over the last seven years indicate abnormal returns?
If so, can this be considered an actual anomaly to market efficiency?”

Furthermore, we will also investigate whether there is a difference between the

effects of two dimensions of CSR efforts on the potential excess stock returns.
More specifically, we will look at the environmental and the social CSR
dimensions separately, and their effects on stock’s excess return.

Page 55
Master Thesis GRA 19003 02.09.2013

1.1 Corporate Social Responsibility

There are many different definitions of CSR, many of which include the fact that
CSR is considered to be a firm’s responsibility and commitment to the society and
the firm’s stakeholders. This commitment is also seen as a contribution which
exceeds what is actually required from a firm by different legal aspects and
expectations form regulators (Investopedia 2012). We will use the following
definition in our thesis, as we feel that this statement comprehensively involves
important aspect of corporate social responsibility: “A process to integrate social,
environmental, ethical and human rights concerns into their business operations
and core strategy in close collaboration with their stakeholders” (EU Commission
2012).

CSR can furthermore be divided into five different dimensions (Dahlsrud 2006),
where each one of the dimensions will define CSR regarding different criteria and
thus make distinct areas of CSR. The five dimensions include the environmental,
social, economic, stakeholder and voluntariness dimensions. We will focus on the
former two dimensions in our study; the environmental and the social dimension.
They consider CSR regarding “the natural environment” and “the relationship
between the business and the society”, respectively. The environmental dimension
will i.e be to contribute to a better environment by for example introducing
recyclable products, while the social dimension will be to contribute to a better
society by for example providing school books for children in primary schools.

Later in our thesis we will set more detailed criteria for what types of CSR
announcements we will include in our study.

1.2 Market Efficiency

An efficient market is a market where stock prices reflect all available
information. One can separate between weak, semi-strong and strong form market
efficiency. Weak form means that stock prices reflect all information available by
examining market trading data, a semi-strong form means that prices additionally
reflect all public information about the prospect of a company, while strong form
means that stock prices reflect all information relevant to the company; including
inside information. We will refer to market efficiency as a semi-strong form in

Page 56
Master Thesis GRA 19003 02.09.2013
this thesis. New information is immediately embedded into the prices, and the
only way to obtain higher expected returns is to take on higher risk. Thus, neither
technical nor fundamental analysis should be able to give investors abnormal
returns, as abnormal returns represent returns above what is justified by risk
(Bodie, Kane and Marcus 2011, 371-402).

Anomalies to market efficiency represent patterns of returns that seem to

contradict market efficiency. Easily accessible statistics such as a stock’s P/E
ratio, dividends announcements and book-to-market size sometimes seem to give
abnormal returns above what is justified by risk, and can represent anomalies to
the efficient market hypothesis (Bodie, Kane and Marcus 2011). In this study we
are going to examine whether we may claim that a company’s CSR
announcements provide a company with abnormal returns, and also represent an
anomaly to market efficiency.

1.3 Discussing the Thesis

CSR has become a central theme for most American corporations, and many
executives, investors and researchers question the effects of CSR on a company’s
financial performance. It is important to get an estimate of the financial effects of
CSR, as this likely will have an impact on future CSR efforts by various
corporations. CSR efforts are undoubtedly important around the globe, and
contribute to solve or improve numerous issues that the world is dealing with. If
researchers and experts are able to prove that CSR efforts not only improve the
world, but also have a positive impact on a firm’s financial performance, it is
more likely that firms will continue to maintain and hopefully expand their CSR
programs. Thus, the importance of this research topic cannot be underestimated,
as positive results and valid evidence may contribute to improving the world
through increased CSR efforts.

In this thesis we will proceed with an event study to examine whether CSR may
be an anomaly to market efficiency. Our contribution to existing literature lies in
the way we will combine the various specifications in our study; meaning how we
will combine the time frame of data collection, methods and data. Our main
specifications are as follows: 1) we will research CSR announcements for
American companies from 2005-2012, 2) we will use the CAPM and the Fama

Page 57
Master Thesis GRA 19003 02.09.2013
and French model, and 3) we will separate between environmental and social CSR
efforts. While some of these specifications have been applied in previous studies,
they have not been studied in combination like we will in this thesis. Thus, it will
be interesting to see the results from this study, and how they will coincide with
existing literature on the topic of CSR’s effect on financial performance.

Prior to having done the event study, we do expect that CSR efforts provide a firm
with positive stock excess returns, and we expect to see that CSR announcements
represent an anomaly to market efficiency. There are several reasons why we
expect this. First, it is because firms can create positive publicity and reputation
by contributing to society. This is likely to improve people’s impression of the
company, which furthermore may have a positive influence on the stock return of
the company. Second, performing CSR efforts may have a positive signaling
effect towards the investment community. The fact that a firm is able to invest in
CSR programs signals that the firm is financially stable, which will be positively
valued by the market. Thus, this may increase the stock return of the company.
Additionally, the fact that a company performs CSR may make people wanting to
associate themselves with the company, which furthermore may help firms recruit
new and talented employees. Thus, CSR activities might attract more employees
with higher qualifications, which increases the overall quality of the workforce
and hence the company as well.

Although there are many reasons why CSR investments can be positively related
to stock returns, there are also some reasons why there might be a negative
relationship. First and foremost, CSR efforts are generally expensive. They do not
generate any directly related income, which means that CSR investments may be
seen as pure expenses for a company. Thus, the investment community might see
CSR investments as something that lowers the financial value of a company, and
the stock returns of a company may thereby decrease. Furthermore, CSR efforts
can in some situations create negative publicity for a company if the company is
accused of doing CSR “only for creating positive publicity and higher returns”
rather than for the “good will”. This is likely to have a negative effect on a
company’s stock returns.

Page 58
Master Thesis GRA 19003 02.09.2013
Prior to performing the empirical study, we believe in the former intuition
supporting a positive relationship between CSR announcements and stock returns.
Thus, we expect to see that CSR announcements positively affect stock returns of
a company.

1.4 The Structure of this Thesis

This thesis will have the following structure: in the second section we will
introduce the main findings of related literature on this topic. In the third section
we will present the methodology of our research, and will go into the details of the
event study. In section four will present the data that we use in our research.
Furthermore, in section five we will present and analyse the empirical results of
our study, and we will proceed to conclude the thesis in section six. We will also
include a detailed bibliography and appendix in the end of the thesis.

This preliminary report will include all of the above, except section five and six
with the empirical analysis and the conclusion, as well as the appendix.

2. Related Literature
During the recent decades, the amounts of CSR efforts that companies have
commit to have grown significantly, and discussions about the wider effects of
these social actions have risen. Several claimed that such CSR programs
influenced the value of the companies in a positive manner, and this incentivised
researchers to further examine the relationship between companies’ CSR
programs and the value of the respective companies. Thus, previous research does
exist on the topic of this thesis, and the researches vary in their findings. Some
studies have results indicating that CSR affects returns positively; others claim
that CSR affects returns negatively, while some also claim that CSR does not
really have an effect on financial returns at all. Furthermore, some studies looked
at all CSR dimensions as one whole, while some studies separated between the
different dimensions of CSR. In this literature review we will present several
studies that have performed various studies and present various findings.

Brammer, Brooks and Pavelin (2006) researched the relationship between

corporate social performance and stock returns in the UK for the time period of

Page 59
Master Thesis GRA 19003 02.09.2013
2002-2004. Their main finding was that firms with high scores on CSR; i.e. have
invested quite significantly in CSR; have lower stock returns, while companies
with the lowest possible score on CSR outperformed the market. Thus, this study
found that CSR investment is largely destructive on shareholder value. The study
also differentiated between social, environmental, employment and community
CSR performance. It found that social CSR efforts tend to perform the worst
relative to stock returns, sequentially followed by environment, employment and
community CSR activities.

Bird et al. (2007) researched what CSR activities are valued by the market in the
US, and had a time frame of data collection from 1991-2003. They researched the
five dimensions of community, diversity, employee relations, environment and
product both on the scale of strengths and concerns; implying ten dimensions all
together. Their main finding was that the market seems to value most firms that
satisfied only the minimum requirements of the dimensions of environment and
diversity (mostly as required by law), and that the market is most proactive
towards CSR in the dimension of employee relations. Furthermore, they found
that the market’s attitude towards CSR activities change over time, and that the
activities recently being valued by the market appear to be diversity, employment
and environment. Finally, they also found evidence to suggest that companies
being identified in the market as having a wide spectrum of CSR activities are
being rewarded in the market place (and vice-versa), indicating that there are
reputational benefits (and costs) related to CSR programs.

Arx and Ziegler (2009) measured the effect of corporate social responsibility on
stock performance in the US and in Europe for the time period of 2003-2006.
Their analysis showed that financial markets do value environmental and social
activities of a firm compared with other firms within the same industry.
Furthermore, they found that the positive effects seemed to be more robust in the
US rather than in Europe.

Cellier and Chollet (2011) measure the impact of CSR rating announcements on
stock prices on the European Market from 2004-2009 on short term European
stock returns. Their study showed that CSR really matters for financial markets,
and that different CSR components have different effects on the stock prices. CSR

Page 60
Master Thesis GRA 19003 02.09.2013
announcements regarding human rights seem to have a positive effect,
environment and human resources seem to have a negative effect, while
community involvement has a mixed effect.

Mollet and Ziegler (2012) measured the impact of socially responsible investing
(SRI) on American and European stock markets in the time span of 1998-2009.
They used the four-factor Carhart model, including risk factors for common
market return, size, value and momentum. The researchers found that SRI is
mostly related to large-sized firms. Furthermore, they also found that when all
four risk factors are included, there was no evidence suggesting that SRI was
neither panelized nor rewarded by the stock markets. Thus, this recent study
suggested that CSR did not have an effect on stock returns in Europe and the US.

Research on this topic has also been performed in other parts of the world outside
the US and Europe. Several similar studies have been conducted on the Asian
market, in which the general consensus seems to be that CSR has a positive
impact on stock performance. A general study of the Asian market found
significant evidence that CSR has a positive effect (Cheung [Link] 2009), in
addition to more narrow studies of China and Taiwan (Chen and Wang 2011;
Wang 2011). A study from Pakistan claimed that CSR had no effect (Iqbal [Link]
2012).Studies from Australia also have mixed findings, where it is both suggested
that CSR does not have a significant impact of stock prices (Newell and Lee
2012) and that there is a weak link (Galbreath and Shum 2012). Furthermore, a
study from South Africa also suggested that CSR announcements have a positive
impact on stock performance; particularly announcements of substantive
monetary value (Arya and Zhang 2008).

The list of previous research could be far longer, and as can be observed,
researchers have found different results over the last decade. A general consensus
seems to be that at least some CSR activities are valued positively by some
markets over various time periods. However, as Bird et al. (2007) found, the
market’s attitude towards CSR change over time. Therefore, it will be interesting
to see how the American market has valued CSR announcements regarding social
and environmental efforts in the time span of 2005-2012 in this thesis, applying
the CAPM and the Fama and French model.

Page 61
Master Thesis GRA 19003 02.09.2013

3. Methodology
The purpose of this thesis is to investigate whether a CSR announcement can
create abnormal stock returns, and if this can be considered an anomaly with
respect to market efficiency. In order to examine this we are going to apply an
event study approach, which is a widely used and accepted research methodology
in finance. The actual event in this thesis will be a public CSR announcement
from a company, in which a company announces that it will invest in a CSR
project. The collected data of the announcements that we are going to use in our
study have to meet certain criteria that we will define in section four of this thesis.

First, we are going to study if this new information of a CSR announcement will
trigger an immediate response in the market, which will lead the stock return to
increase and thus create an abnormal or excess return in the short run. Second, we
will see if this possible increase will continue after the actual event, and hence can
be considered an anomaly to market efficiency.

When looking at the CSR announcements we will separate between two types of
CSR actions; being environmental and social CSR efforts. First we will combine
both types of announcements into a CAR study and see if they have an effect on
return together, and if CSR announcements may be an anomaly to market
efficiency. Afterwards we will perform a multiple regression of the two types of
CSR efforts, in order to see the different effects of environmental and social CSR
activities. This may allow us to examine whether one of the two types of
announcements has a stronger effect on excess returns. Furthermore, we also
expect to see different results from the two types of models that we will be using.
The CAPM and the Fama and French model have different factors that they
measure, and this may have an impact on the testing of the impact of CSR
announcements on market efficiency.

We will look at companies which have had CSR announcements during the period
of 2005 and up to 2012, which will be the overall event window. When collecting
data for companies’ stock returns, we will use the returns 3 months prior to and
after the actual event. We will calculate the stocks’ return by using the following
formula, and hence we will get the changes in the stock’s return:

Page 62
Master Thesis GRA 19003 02.09.2013
Stock return changes: (price today – price yesterday) / price yesterday

For all tests that we perform in this study, we will use a 95% significance level,
where a p- value lower than 5% will show that the results are statistically
significant.

The first step in our event study is to make an index or a proxy for what one can
expect the normal stock return in the market (rM) to be if the CSR announcement
events never happens. This proxy can be made in several ways, and we will use
the CAPM to make such a proxy. We are going to make this index by collecting
data from the S&P500 from 2005-2012, since this is our total period of data
collection for the CSR announcements. By collecting for this whole period of
time, we will decrease the chances that the market index will be affected by the
actual CSR events that we are studying.

When performing this event study, we will use both the single index market
model CAPM and the three factor market model of Fama and French. As
mentioned, this will allow us to test for abnormal returns and market efficiency
with somewhat different aspects.

The CAPM takes into consideration that a stock’s expected return is affected by
one market factor and one firm-specific factor:

Eqtn 1: E(ri) = rf + βi[E(rM) – rf]

E(ri): Expected return of stock i
rf: risk free rate
β: stock return’s sensitivity to market return
E(rM): Expected market return
(Bodie, Kane and Marcus 2011, 321)

However, we will use the index model representation of CAPM in our further
analysis:

Page 63
Master Thesis GRA 19003 02.09.2013
Eqtn 2: rt = α + βrMt + εt

rMt: market rate of return in period t

β: the stock’s sensitivity to the market return
εt: part of the security’s return that comes from firm specific events
α: the average rate of return the stock would realize in a period with a zero
market return
(Bodie, Kane and Marcus 2011, 381)

Fama and French also take into consideration that the return can be affected by
one market factor and one firm-specific factor, while also adding two factors for a
firm’s size and book to market value:

Eqtn 3: rit = αi + βiMRMt + βiSMBSMBt + βiHMLHMLt + εit

SMB = Small Minus Big; i.e. the return of a portfolio of small stocks in excess
of the return on a portfolio of large stocks
HML = High Minus Low; i.e the return of a portfolio of stocks with a high book
to- market ratio in excess of the return on a portfolio of stocks with a low book
to –market ratio.
(Bodie, Kane and Marcus 2011, 363)

In both cases we expect the intercept term (α) to be equal to zero, hence, the
hypothesis will be:
H0: α = 0
HA: α ≠ 0

If we reject this hypothesis, it is suggested that the stock has over- or

underperformed compared to the market. Thus, this can be an indicator whether or
not the stock has obtained abnormal returns; perhaps as a result of the CSR
announcement. However, one should be careful in relying too much on the
interpretation of the alpha in an event study like this (Brooks 2008).
Consequentially, we will conduct more in-depth analyses as well, in order to get a
better assessment of the potential abnormal returns resulting from the CSR
announcement.

Page 64
Master Thesis GRA 19003 02.09.2013
The betas in the different models will show how much the return is affected by,
and correlated with, the market factor in both models and the SMB and HML
factors in the Fama and French model. The betas are also an indicator of risk, as
the assumption of the market efficiency hypothesis is that higher risk gives higher
expected return, as one should be rewarded for taking upon more risk (Bodie,
Kane and Marcus 2011).

The residuals of the regression models will indicate and capture the unexpected;
thus, being an indicator for how much the stocks’ return may be affected by the
CSR event we are looking into. Hence, we will have to investigate the residuals
further. We will do this by calculating the stocks’ abnormal, unexpected return by
taking the actual return minus the estimated normal return from the market proxy:

Eqtn 4: CAPM: εt = rt – (α + βrMt)

Eqtn 5: FF: εt = rt – (α + βrMt + βSMBSMBt + βHMLHMLt)
(Bodie, Kane and Marcus 2011, 321)

We also make another hypothesis test to indicate whether CSR announcements

have an impact on abnormal return. In both equations 4 and 5 above, we expect
the epsilon to be equal to zero. Thus, we have the following hypothesis:
H0: ε = 0
HA: ε ≠ 0

If ε is significantly different from zero, we will reject the null hypothesis. We will
measure this by looking at the p-value from our initial regression with the CAPM
and the Fama and French model. Rejecting the null hypothesis will suggest that
CSR announcements represent excess return.

We calculate the abnormal return for each of the companies included in the study.
We gather stock prices from each firm starting three months ahead and ending
three months after the announcement. We will use stock prices for three days prior
to and after the announcement to look for the immediate effects of the
announcements. Furthermore, we will use the stock prices from the six-month
interval to look for abnormal returns in the longer run and potential anomalies to
market efficiency. However, this will not be sufficient in order to explain whether

Page 65
Master Thesis GRA 19003 02.09.2013
the CSR announcement can be considered an anomaly to market efficiency. We
also have to consider the fact that there could be a leakage of information prior to
the event that could affect the stock’s performance and cause abnormal return.
The next step in this study will therefore be to make an indicator that will account
for this. We will do this by summing up the excess returns of all of the
companies’ stocks, and will get what is called the cumulative abnormal return
(CAR). Making a graph with these results will show us if there will be continuous
fluctuations of the stocks’ return after the actual event or not (Bodie, Kane and
Marcus 2011, 382-393). We have the following hypothesis:
H0: CAR remains stable
HA: CAR continues to drift

If the CAR graph continues to drift after the event; implying that we reject the null
hypothesis; then the CSR announcements can be considered an anomaly to market
efficiency with respect to the given market models.

After having investigated whether the CSR announcements can be considered an

anomaly to market efficiency, we will continue our study by doing an analysis of
the different types of CSR announcements’ effect on excess return. We will
perform a regression analysis where the excess return from equation 3 and 4 is the
dependent variable, and the two different dimensions of CSR; environmental and
social; are the independent variables. Then we will see whether these CSR
announcements are statistically significant and can be explanatory variables to the
potential excess return, or if there could be other factors that can explain the
abnormal return. Furthermore, we will also be able to see if one of the CSR-
announcement variables have more impact on excess return than the other, and
hence, may be financially more important for a company to consider in its CSR
strategy. We will look at the excess return of the two market models, CAPM and
Fama and French, separately to see if it will make a difference in our results.
Thus, we will use the following multiple regression lines:

Eqtn 6: CAPM: RExcessCAPM = α + βrexENV + βrexSOC + ε

Eqtn 7: FF: RExcessFF = α + βrexENV + βrexSOC + ε

Page 66
Master Thesis GRA 19003 02.09.2013
The RExcess is the dependent variable, and the input will be all of the excess returns
we have found in our data sample. The independent variables r exENV and rexSOC will
be all of the stock returns of the companies which have made an environmental or
a social CSR announcement, respectively. Hence, we will distinguish between the
two different types of CSR announcements. The betas of these variables will
indicate how much each of them contributes to explain the potential excess return.
We will also check if these results are statistically significant.

4. Data
Data of proper quality is a key ingredient for most empirical researches, and is
also significantly important for this thesis. As described in the previous section,
we will perform an event study in order to investigate the relationship between
companies’ CSR announcements and their returns. In this section we will describe
the data that we will use in the study.

There exist numerous variations of CSR announcements. CSR actions differ in

size, field, type and budget, and the companies also differ in size, industry,
nationality, etc. Thus, it is necessary to define certain criteria for the CSR
announcements that we will include in our study. These criteria concern the CSR
action itself, as well as the actual company. We have set the following criteria for
this thesis:

a) The CSR announcements have to be of a certain size for us to include them in

our study, and we have set this limit to $1,000,000-$50,000,000. This range
will allow us to include a wide range of announcements, while simultaneously
eliminating the smallest and largest announcements such that we increase the
chances of avoiding big outliers in our data sample. Furthermore, the CSR
announcements have to be of an environmental or social character, and the
CSR efforts can be domestic and/or international.

b) We have also set certain criteria with regards to what firms we will include in
our study. First and foremost, we will look at American companies. The
companies must be at least ten years old, as we do not want to include younger
companies that are in a natural growth phase. If we include growth companies,
it can be more difficult to find the actual effect of the CSR announcement. We
Page 67
Master Thesis GRA 19003 02.09.2013
will include a maximum of three announcements per company with at least a
year in between each of them. Furthermore, the firms do not necessarily have
to be multinational. Many firms will be multinational; however, since we are
looking at the effects of CSR announcements on the American stock market,
we will also include firms that are merely of a domestic character. The firms
have to be listed on an American stock exchange, and we will not include not-
for profit organizations. Lastly, the firms do not have to be within a certain
industry. We will include CSR announcements from firms across a wide
specter of industries, for example banking, retail, technology, food and drink,
sports, etc.

The quality of the data is vital for an empirical study like this, and finding
sufficient data is oftentimes a challenge. There exist several databases which
contain substantial information about companies’ CSR programs and
announcements. For this thesis we plan use the webpage “[Link]” as
the main database for gathering information about companies’ CSR
announcements. This webpage has gathered public releases regarding CSR efforts
from more than 2,800 companies worldwide, dating back to approximately year
2000. Thus, we will research thousands of press releases from various American
firms regarding CSR, and we will match the different CSR announcements with
the criteria that we have set for our study. We plan to use an absolute minimum of
fifty announcements for our study, such that we will be able to make valid
inferences.

We will use data from CSR announcements that have taken place between 2005
and 2012. When looking at companies’ press releases for their CSR
announcements, we are able to find the exact dates of the announcements. Thus,
to find data on company returns, we gather stock prices from each firm starting
three months ahead and ending three months after the announcement. After
having found the specific dates of the announcements, we retrieve stock prices
from “[Link]”. Additionally, we will also retrieve S&P 500
prices from this database in order to make the initial market proxy.

When finding the book/market value and the market capitalization of the firms in
order to apply the Fama and French model, we retrieve information from the

Page 68
Master Thesis GRA 19003 02.09.2013
annual report/quarterly report closest in time of the announcement. For example,
for an announcement dated February 3rd 2008 we will refer to the last quarter of
2007 to retrieve the book/market value and the market capitalization of this
corporation. These annual/quarterly reports have also been found at
“[Link]”.

Page 69
Master Thesis GRA 19003 02.09.2013

5. Bibliography

Arya, Bindu and Gaiyan Zhang. 2008. Institutional Reforms and Investor
Reactions to CSR Announcements: Evidence from an Emerging Economy.
USA: University of Missouri- St. Louis. In text (Arya and Zhang 2008).

Arx, Urs von and Andreas Ziegler. 2009. The Effect of Corporate Social
Responsibility on Stock Performance: New Evidence for the USA and
Europe. Switzerland: Zürich. In text: (Arx and Ziegler 2009).

Bird, Ron, Anthony D. Hall, Fransesco Momentè and Fransesco Reggiani. What
Corporate Social Responsibility Activities are Valued by the Market?
Journal of Business Ethics76 (2) :189-206. DOI 10.10.1007/s10551-006
9268-1. In text:(Bird [Link] (2007)).

Bodie, Zvi, Alex Kane and Alan J. Marcus. 2011. Investments and Portfolio
Management. 9th ed. McGraw-Hill Irwin. In text: (Bodie, Kane and
Marcus2011).

Brammer, Stephen, Chris Brooks and Stephen Pavelin. 2006. Corporate Social
Performance and Stock returns: UK Evidence from Disaggregate
Measures. Financial Management 35 (3): 97- 116. In text: (Brammer,
Brooks and Pavelin (2006)).

Brooks, Chris. 2008. Introductory Econometrics for Finance. 2nd ed. USA,NY:
Cambridge University Press. In text: (Brooks 2008)

Cellier, Alexis and, Pierre Chollet. 2011. The Impact of Corporate Social
Responsibility Rating Announcement on Stock Price: An Event Study on
European Markets. Paris: Universitè Paris- Est.. In the text: (Cellier and
Chollet 2011).

Page 70
Master Thesis GRA 19003 02.09.2013
Cheung, Yan Leung, Weiqiang Tan, Hee-Joon Ahn and Zheng Zhang. 2009.
Does Corporate Social Responsibility Matter in Asian Emerging Markets?
Journal of Business Ethics 92: 401-413. DOI 10.1007/s10551-009-0164-3.
In text: (Cheung [Link] 2009).

Dahlsrud, Alexander. 2006. How Corportae Social Responsibility is Defines: an

Analysis of 37 Definitions. Wiley Inter Science DOI: 10.1002/csr.132. In
text: (Dahlsrud 2006).

European Commission 2012. Sustainable and responsible business; Corporate

Social Responsibility (CSR). Accessed December 28, 2012, 12.05 o’clock.
<[Link]
social- responsibility/index_en.htm>.In text: (EU Comission2012).

Galbreath, Jeremy and Paul Shum. 2012. Do customer satisfaction and reputation
mediate the CSR-FP link? Evidence from Australia. Australian Journal of
Management 2012 37: 211. DOI:10.1177/0312896211432941. In text:
(Galbreath and Shum 2012).

Honghui Chen, Xiayang Wang. 2011. Corporate social responsibility and

corporate financial performance in China: An empirical research from
Chinese firms. Corporate Governance 11 (4): 361 – 370. In text: (Chen
and Wang 2011).

Investopedia. 2012. Corporate Social Responsibility. Accessed: October 28, 2012,

15.10 o’ clock. <[Link]
[Link]#axzz2GRfHCmn4> In text: (Investopedia 2012)

Iqbal, Nadeem , Naveed Ahmad, Nauman Ahmad Basheer and Muhammad

Nadeem. 2012. Impact of Corporate Social Responsibility on Financial
Performance of Corporations: Evidence from Pakistan. International
Journal of Learning & Development 2 (6). DOI:
10.5296/ijld.v2i6.2717. In text: (Iqbal [Link] 2012).

Page 71
Master Thesis GRA 19003 02.09.2013
Mollet, Janick Christian and AndreasZiegler. 2012. Is Socially Responsible
Investing Really Beneficial? New Empirical Evidence for the US and
European Stock Markets. Switzerland: ETH Zûrich. In text: (Mollet and
Ziegler (2012)).

Newell, Graeme and Chyi Lin Lee. 2012. Influence of the corporate social
responsibility factors and financial factors on REIT performance in
Australia. Journal of Property Investment & Finance 30 (4): 389-403.
DOI:10.1108/14635781211241789. In text: (Newell and Lee 2012).

Wang, Yungchih George. 2011. Corporate Social Responsibility and Stock

Performance- Evidence from Taiwan. Modern economy 2011 (2): 788
799. DOI: 10.4236/me.2011.25087. In text: (Wang 2011).

Page 72