Does the stock market value the inclusion in a sustainability stock index? An
event study on the Ethibel sustainability indices.
Author: Sebastiaan Schmidt
1793403
Supervisor
: prof. dr. L.J.R. (Bert) Scholtens (Rug)
20-06-2014
Abstract:
1. Introduction
In the last two decade a growing number of so called sustainability indices has emerged. Two of these that have attracted a growing share of attention are the Vigeo owned Ethibel Excellence indices. These indices are generally composed of the top performing corporations on corporate social responsibility (CSR) related issues. Though the rating criteria differ between the sustainability indices, generally firms part of the indices are considered to have an above average Corporate Social Performance (CSP). Similar to some of the recently published studies (Oberndorfer 2013, Bechetti et al. 2012, Consolandi et al. 2009), this paper will use the appearance of a corporation in the sustainability index as a proxy for a firms’ above average engagement in CSR. Using event study methodology the effect of this above average CSP of a firm on the stock prices of a firm will be measured. As the stock prices are argued to represent the value of a firm (Bechetti et al. 2012), the performance of a firm on the stock market should be a fair representation of the financial performance of a firm. This paper is an attempt to empirically; using event study methodology, clarify the relationship between CSP (or CSR) and CFP.
Nonetheless, recent reports by Eurosif and USSif show that there is a growing importance of Social Responsible Investment (SRI) in the investment world. . As a recent research paper published by Eurosif (2012) states; the size of socially responsibly invested funds has grown from €6 billion in 2006 to €48 billion in 2011, which represents a 700% growth in size. Similarly the USSif report (2012) estimates the size of SRI funds at $3.74 trillion in assets, or around 11.3 % of the total assets managed in the United States. Given this substantial growth in SRI, one might expect that a firm with a high CSP would be rewarded on the capital market. Recent studies (Doh et al 2009, Bechetti, etc alle event studies uit tabel) have tried to address this question on additions and deletion of firms in several different so called sustainability indices such as amongst others the Calvart Index, Dow Jones Sustainability Index and the Domini index. These indices are composed by different organizations following an assessment of the CSP of different firms. The leaders on CSP, in for example in each industry, are then grouped together to form the sustainability stock index. Since these indices are composed following an assessment of a firm’s CSP, the reasoning is that being added into such an index can be seen as a proxy for a firm engaging in CSR. At the same time fluctuations in stock prices can be seen as a proxy for the financial performance of a firm. Even though it is only a partial performance measure, since it only reflects the value of the firm. Nonetheless, stock market performance is argued to be; “an ex ante measure of performance that has been found to correlate well with ex post performance, demonstrating predictive validity.“ (Haleblian, Kim, & Rajagopalan, 2006; Kale, Dyer, & Singh, 2002) Consequently, studies using event study methodology to assess whether the announcement of an inclusion or deletion of a firm from the index leads to abnormal returns on the stock market, can be seen as a valid method to assess the stock market reaction (CFP) to a firm’s CSP. However, as with other studies on the relationship between CSP and CFP, these different studies provide mixed results. On top of that the samples used are relatively outdated with the most recent one ending in 2010.
are well known sustainability indices, similar to the earlier mentioned indices, which are composed by the Ethibel Forum grouping leaders on CSP in different industries. Similar to the earlier performed studies (for an overview see Table 1) this study will use the addition/deletion from the index as a proxy for a firm’s CSP, and the abnormal returns as a proxy for a firm’s CFP to answer the main research question:
Does a firm’s engagement in corporate social responsibility lead to a better or worse corporate financial performance?
Subsequently, following upon the recent reports of Eurosif (2012) and USSif (2011) indicating an significant growth of importance of SRI, this study will try to assess whether this presumed growth is visible as a trend within the dataset over the year from 2005 to 2013. The follow up question would be:
Is there a trend visible over time in the effect of a firm’s engagement in corporate social responsibility on the corporate financial performance?
The remainder of this paper will be structured as follows; section 2 will provide a in depth literature study used to compose the hypotheses to be used in the empirical analysis, section 3 will explain the event study methodology, section 4 will elaborate on the dataset used, section 5 will provide the results, section 6 will provide a discussion on the results, and section 7 will provide a conclusion.
2. Literature review.
First of all, this section will provide an overview of the research and the different theoretical concepts concerning the relationship between CSP and CFP. Secondly, the reasoning behind choosing event study analysis to test this relationship will be further explained. Thirdly, an overview of the recent event study researches on the CSP-CFP relationship and their results is provided. Following up on that, the recent developments in the field of SRI are analyzed. To conclude this section the different hypotheses to be tested in this paper are listed.
voluntary actions to improve social or ecological conditions” (Siegel & McWilliams 2001) Event though this definition is at the moment the most widely accepted the question arises what are indeed “voluntary actions”. Where some argue that this is what a firm does beyond the legal requirements, other argue that firms are also influenced by isomorphic pressures to engage in CSR related actions, and thus does actions are not “voluntary”. (Cañon-de-Francia, 2009; Oberndorfer, 2013) This lack of a concise definition of what constitutes CSR and the measurability of it makes research on the subject difficult. Nonetheless, there seems to be a consensus that actions by a firm to improve social or ecological conditions constitute CSR. This is the definition this paper will adhere to.
As for CFP, there is a discussion what constitutes actual financial performance. On the one hand you could say financial performance is the realized growth or realized returns, absolute or relative to competitors. The difficulty with these definitions of financial performance is the measurability, and the difficulty to attribute them to one specific factor such as a firms CSP. Another possible definition of CFP, which will be used in this study, is the firms performance on the stock market. The reasoning behind this that stock market prices should reflect the fundamental value of the firm, i.e. the discounted sum of dividends accruing to shareholders. (Bechetti et al, 2012) An condition for this definition to hold is that stock markets are efficient, meaning that they reflect accurately the fundamental value of the firm. This is supported by (Haleblian, Kim, & Rajagopalan, 2006; Kale, Dyer, & Singh, 2002) who find that; “it is an ex ante measure of performance that has been found to correlate well with ex post performance, demonstrating predictive validity.“ On top of that, stock market performance has been found to be a relatively unbiased predictor of corporate performance, and to be invariant to differences in accounting policies across nations and adopted by firms. (Cording, Christmann, & King, 2008) Though this assumption is open to discussion (Orlitzky, 2013), this paper assumes stock market performance as a proxy for financial performance.
shareholder value. When a corporation invest in environmental or social goals, the corporation is essentially investing corporate financial funds in goals that serve other purposes than maximizing shareholder value. As Friedman (1970) and later Cañon-de Francia (2009) and Telle (2006) argue the cost of investing in such environmental and social goals outweigh the financial benefits. Thus a corporations effort to enhance CSP is detrimental for a corporations CFP, and leads to a reduced shareholder value. Consequently Friedman (1970) predicts that there is no room for CSR in the management of corporations.
On the other hand, those who predict a positive relationship build on the stakeholder theory (Freeman 1984, Donaldson and Preston 1995, Jones 1995). Stakeholder theory builds on the idea that a corporation relies on several different stakeholders for its continuation of business. This means that management should attempt to satisfy different stakeholder groups such as, the government, the society, NGO’s, employees, competitors and clients. The argument goes that, when one of these stakeholder groups is not satisfied this could to a conflict between the corporation and the stakeholder group, e.g. a NGO attacking the corporation for violating human rights, which could lead to significant cost for the corporation.(McWilliams et al 2006) Voluntary investment by the corporation to satisfy the different stakeholder groups would reduce the likeliness of such a conflict and ultimately lead to enhanced financial performance. Adequately satisfying the different stakeholder groups would not only be essential to reduce the likeliness of costly conflict, but would also strengthen the competitive position of ultimately the financial performance of the firm (Derwal et. all 2005, Barnett et al. 2006, Waddock 1997,). The idea that investment in CSR may lead to a competitive advantage is supported by Porter and van der Linde (1995). In their article they provide an example where investment in pollution reduction not only satisfies societal goals, but also leads to innovation in energy efficiency and thus a strengthening of the competitive position of the corporation. Ultimately, the stakeholder argument goes against the traditional or shareholder theory view that suggests a trade-off between CSP and CFP. Rather, it is argued that satisfying different stakeholder groups by increasing CSP ultimately leads to enhanced CFP.
Event study methodology is used to measure the effect of an economic event, such as for example an announcement for a merger, on the value of a firm. The value of the firm is measured using financial market data on stock prices. As the argument goes the stock price accurately reflects the actual value of the firm. (MacKinlay, 1997) The use of event study methodology is not a novelty in CSR research ( Siegel 1999, McWilliams & Siegel 1997), for example Posnikoff (1997), and Wright & Ferris (1997), already used event study methodology to measure the impact of voluntary divestment from South-Africa during the apartheid had on the value of the firms involved. However, using the addition or deletion from a sustainability index as an event is a relatively new approach. Sustainability Indices, such as the Domini Index and the Dow Jones Sustainability Index, are composed by rating firms on a range of criteria, and consequently grouping the top performers into the index. Though each index uses different selection criteria and different methodologies to acquire these ratings, overall they are intended to reflect a firms performance in environmental and social fields. Consequently, as the argument goes, being part of such a sustainability index reflects that the firm has a relatively high CSP. Being added into such an index can thus be seen as a proxy for a firm investing in their CSP (relatively to other firms), and the announcement of this addition into the index (or the deletion out of) can be used as the event studied.
Table 1 shows an overview of the most important event studies on sustainability indices performed in recent years. First of all, and most importantly to note is that they show mixed results. Where Oberndorfer (2013) and Chueng & Roca find negative effects of an addition of a firm to the index, which is in line with the shareholder theory argument. Others, such as Chatterji and Mitchel (2012) find a positive effect on stock price following the announcement of an addition into the index, in line with the stakeholder theory argument. At the same time there are those (Consolandi et al 2009, Doh et al 2010, Bechetti et al 2012) that find no significant effect for additions into a sustainability index, suggesting that there is no relationship between the two. It is important to understand the possible reasons why these seemingly similar studies provide such different results.
one changes the geographic location the stakeholders involved change as well. As the stakeholders change, their preferences change as well. And since the stakeholders argument reasons that a firm’s success is dependent on its ability to satisfy its different stakeholder. Logically, a change in stakeholder preferences changes the requirements for a firm to satisfy them
Secondly, there is the obvious difference in indices used in the research. As shown in table 1 there are six different indices used in the research performed. Since these six different indices all use different rating methodologies and selection criteria, it is possible that being added into the DJSI World index reflects a different CSP as being added to the Calvart index. Therefore, being added into one index might provide different results than being added in a other one.
Third of all, the time periods used for the samples differ greatly between the different studies. Overall the sample periods range from the earliest staring in 1990 (Bechetti et al, 2012) to the latest ending in 2010 (Cheung & Roca, 2013). Most of them however are using a sample within the time range between 1999 and 2008. As event study reflects a stock price reaction at the exact time the event takes place, they are actually a representation of the investors preferences at that moment in time. Meanwhile, investors are influenced by study outcomes such as those by Derwal et al (2005) and Orlitzky et al (2003) changing their perception of the effect of CSP on CFP. Consequently, as time goes by the environment changes and consequently the investors’ preferences change as well. A prime example of this is the growing interest in an size of Social Responsible Investments. (Eurosif 2012, Ussif 2012)
signing the principles asset managers commit to investing responsibly. Responsible investment is then defined as; “an approach to investment that explicitly acknowledges the relevance to the investor of environmental, social and governance (ESG) factors, and the long-term health and stability of the market as a whole. It recognizes that the generation of long-term sustainable returns is dependent on stable, well-functioning and well governed social, environmental and economic system.” (PRI 2014) These three different databases on SRI growth seem to indicate a trend in investor preferences. The question which comes to mind is whether this is also visible on the stock market, i.e. an appreciation of stocks of firms with a relatively high CSP. In the context of event study research a different reaction should be visible within the sample as you progress through time. An addition into a sustainability index, which is effectively an endorsement of a firm’s investment in its CSP, should have a different effect on stock prices in the earlier years of the sample as compared to the later years in the sample. (the same goes for deletions out of the index) As visible in table 1 Becchetti et al (2012) already experimented with different sub-samples to account for differences over time. However, their sample only ranged to 2004, and as the Eurosif (2012) data shows the real growth in importance of SRI started mainly as of 2009. As the samples of the earlier studies performed ranged to approximately 2008 they could not have been able to capture the effect of this trend. As this study uses a database ranging from 2005 to 2013 there is a possibility to analyze whether this trend is also reflected in stock price reaction following the announcement of addition or deletion from a sustainability index.
This paper will continue to test a number of hypotheses. First of all, similar to the other event studies performed this study will perform a test on the full sample of additions into, and the deletions out of the Ethibel Indices. The expectations as derived from the review of similar studies performed earlier are mixed.(Table 1) Taking that into account the following hypotheses will be tested will be tested on the full sample ranging from 2005-2013.
For the announcements of an addition of a firm into the index:
H1a: Addition to the index will have a negative effect on the abnormal returns of the individual
firms. (Oberndorfer 2012, Chueng & Roca2013, Cañon-de Francia 2009)
H1b: Addition to the index will have no significant effect on the abnormal returns of individual
H1c: Addition to the index will have a positive effect on the abnormal returns of individual firms.
(Chatterji and Mitchel 2013)
As none of the previous studies showed a positive effect for a deletion out of the index, the hypotheses for the announcements of a deletion of a firm out of the index are:
H2a: Deletion out of the indexes will have a negative impact on the abnormal returns of the
individual firms. (Becchetti et al 2012, Doh et al 2010, Chueng & Roca 2013, Consolandi et al 2009)
H2b: Deletion out of the indexes will have a no significant effect on the abnormal returns of the individual firms.
Secondly, following the analyses of the developments of the size and importance of SRI over the years, expectations are that this should be visible within the dataset. This gives us the following hypotheses to be tested.
For the announcements of an addition of a firm into the index:
H3a: Over time we will see a change in the effect of an addition into the index on the abnormal
returns of individual firms.
H3b: Over time the effect of an addition into the index on the abnormal returns of the individual
firm will become increasingly positive.
For the announcements of a deletion of a firm out of the index:
H4a: Over time we will see a change in the effect of a deletion out of the index on the abnormal
returns of individual firms.
H4b: Over time the effect of a deletion out of the index on the abnormal return of an individual
firm will become increasingly negative.
TABLE I
Schematic overview off previous event studies on Sustainability indices Authors Year Published Index or Program Name(s) Sample Time Period Depend ent Variable s Event Window(s) Sample Size(s) Independent Variables Results Oberndo rfer et al. 2013 DJSI WORLD, DJSI STOXX German listed firms 1999-2002 CAAR (3-factor) 5-day event window 23-27 Announcement day (AD) of inclusion, day of change of index Negative return following announcement Chueng, Roca
2013 DJSI WORLD Asian listed firms, 2002-2010 CAAR, 1-day to 60-day event windows, 178 Announcement day (AD) of inclusion/deletio n, day of change of index Negative return following AD (both inclusion & deletion),
Chueng 2011 DJSI WORLD US listed firms 2002-2008 CAAR, 1-day to 60-day event windows, 177 Announcement day (AD) of inclusion/deletio n, day of change of index No significant results for CAR.
Consolan di et al.
Authors Year Published Index or Program Name(s) Sample Time Period Depend ent Variable s Event Window(s) Sample Size(s) Independent Variables Results Hawn, Chatterji, Mitchell
2014 DJSI World Globally listed firms (58% U.S. firms) 1999-2007 CAAR (1-factor) 1 to 5-day event windows 418 Announcement day (AD) of inclusion/deletio n, day of change of index,
3. Methodology
First of all, a decision has to be made concerning the choice of methodology to test the aforementioned hypotheses. In this study, similarly to the studies shown in Table 1, the event study methodology will be used. First of all, this methodology is chosen because one of the goals of this paper is to compare the results found with those found in the previous studies. (Table 1) Choosing a similar methodology enhances comparability of the results. Secondly, another goal of this paper is to assess the relationship between CSP and CFP. As explained before, the stock market performance is used a proxy for CFP, and the announcement of a deletion or addition from the EEE indices as a proxy for CSP. Therefore event study methodology, utilizing abnormal stock market returns as a dependent variable, and an event providing new information to the stock market as an independent variable, provides an ideal tool to measure the relationship between the two. (Brown, Warner, 1988)
MacKinlay et al. (1997) an Siegel and McWilliams (1997,1999) gave a detailed description of the methodology for the standard event study method, involving five steps: (1) event definition, including definition of the event and estimation windows; (2) selection criteria for including firms in the study; (3) estimation of normal performance within the estimation window and predicting normal returns during the event window in the absence of the event; (4) calculation of the Abnormal Returns (AR) and cumulative ARs (CAR) within the event window; and (5) testing whether the AR and cumulative AR (CAR) for all firms treated as a group are statistically different from zero. When following these five steps it is important to note that the three assumptions of event study research are adhered to. The first (assumption 1) is the market efficiency implies that stock prices incorporate all relevant information which is available to market traders. If this is true, then any financially relevant information that is newly revealed to investors will be quickly (instantaneously) incorporated into stock prices. Therefore, an event is anything that results in new relevant information. (McWilliams & Siegel, 1997) The second
(assumption 2) assumption relates to the first in that it is assumed that the event is unanticipated.
Event study would become complicated because it is then uncertain at which time the information reached the market. (McWilliams & Siegel, 1997) The third (assumption 3) and last assumption is that there are no confounding events within the event window. Confounding events are events that are not researched but can possibly influence stock prices and thus the abnormal returns in the event window. Examples of confounding events are earnings announcements, merger & acquisition announcements, dividend related announcements, and new product announcements. As McWilliams & Siegel (1997) emphasize this is possibly the most important assumption for event study research, and thus has to be considered appropriately. The event that will be researched is the official announcement of an addition or deletion of a firm from the ESI Excellence Europe and the ESI Excellence Global. These announcements are made public semi-annually through an official press release by Forum Ethibel and Vigeo rating S.A. The date of this review is defined in a rulebook published by Vigeo and is strictly adhered to and are as follows:
(From June 2002-2009)Annual review; Third Friday of September announcement of changes in index.
(from 2010 onwards) Annual review; First Friday of September announcement of changes in index.
(from June 2002 onwards) Semi-annual review; First Friday of March announcement of changes in index.
they might not be sure what the effect is on the value of the firm. Whether their reaction thus accurately reflects the effect of the announcement on the CFP is debatable. Nonetheless, we assume that market efficiency holds in this case (assumption 1).
Furthermore, to comply with the second assumption, the announcement of the change in the EEE indices have to be considered as new and unanticipated information. As the constituents of the indices are made public on an exact agreed upon date, without any prior announcements, the announcement can be considered as containing new and unanticipated information. Thus fulfilling the second assumption. (assumption 2)
When extending the event window, the chance of confounding effects influencing the results increases. When taking a smaller event window, such as the 3- to 5-day event window used, this chance is limited. When then controlling for events such as; earnings announcement, merger & acquisition announcements, and or other extraordinary events possibly influencing stock price movement. Then with relative certainty one can assume that the stock price movement can be attributed to the actual event researched. (Assumption 3). Since the ESI Excellence Europe and the ESI Excellence Global are publically well known indices, there is no reason to infer that longer event windows are necessary to account for delay in information reaching the market. For this research the dates mentioned above will be defined as the event date (0), and a short 3-day event window (-1, +1) as proposed by McWilliams & Siegel (1999) will be used. Next to this main event window, similar to the study by Oberndorfer (2013), an extra day prior and after are added to the event window to create a 5-day event window (-2, +2).
An estimation window will be used to estimate the parameters used to calculate normal returns in the event window. In previous event study research, estimation windows ranging from a 90-, to a 200-day window prior to the event are used. In this paper however, an estimation window of 250 to- 3 days prior to the event is used. Different to other studies (Oberndorfer 2013, Siegel & McWilliams 1999), the estimation window we use ends 3 days prior to the actual event rather than 10 to 25 days prior to the event. The reasoning behind this is that it would not make sense to exclude the most recent events from your estimation, this because the more recent the event the likelier it is to influence the stock market returns. Thus excluding the two or so weeks prior to the event would potentially give biased results.
The reason 2005 is chosen is because this is the year that the index became property of Vigeo SA. This sample will then be screened for confounding events in the event window possibly influencing stock prices. Confounding events that are checked for are earnings announcement, merger & acquisition announcements, and or other extraordinary events possibly influencing stock price movement. The databases used are the Financial Times, and Wall street Journal press releases, as well as the company’s own official press releases. All announcements with confounding events will be excluded from the sample leaving a final sample that will be tested.
To calculate abnormal returns the simple Capital Assets Pricing Model first introduced by Fama and Macbeth (1967) is used. In its simple form:
With E( ) = 0 and Var( ) =
Where Rit and Rmt are the return on security i and the return on the market portfolio respectively. = the systematic risk of stock i, and the error term. Since abnormal returns are defined as the difference between the normal (expected) return and the actual return, abnormal returns (AR) then are essentially the . Or to write it differently:
As for example suggested by Siegel et al. (1999) in their paper to account for differences in volatility of stocks caused by differences in the β the next step is to compute the standardized abnormal return (SAR). This is done by dividing the AR by its standard deviation. The standard deviation used is the standard deviation of the over the estimation window (-250, -3).
Where is the residual variance from the market model as computed for firm i, Rm is the mean return on the market portfolio calculated during the estimation period, and T is the number of days in the estimation period.
The standardized abnormal returns can then be accumulated over the number of days of the event window to get the cumulative abnormal returns (CAR) over the event window (k).
A standard assumption is that is independent and identically distributed across firm. With this assumption, we can make all identically distributed by dividing them by their standard deviation which is equal to (Siegel et al 1999 )
Thus the average standardized cumulative abnormal return for n firms across the event window can be computed by;
In the last step then the test statistic which is used is the z-value, when significant this means that the value of the ACAR is different from zero. The z-value is computed by
The first series of tests will be performed on the entire sample on multiple event windows around the announcement date of additions and/or deletions of a firm from the index. The results will be tested as mentioned before using a t-test to assess whether the abnormal and/or cumulative abnormal returns differ significantly from zero. In order to reject the null hypothesis a 95 % significance level will be adhered to, which is an arbitrary level. Given that previous researchers have found such mixed results, this level seems more appropriate than the less strict 90 % significance level. On the other hand, given the relative small sample (as compared to other empirical studies), the chance of a Type II error occurring is relatively high, choosing the very strict 99 % level would only increase this chance. Therefore the 95% significance level is chosen. To account for non-normality in the sample observations two different non-parametric tests, the rank (normal approximation) and the Wilcoxon signed rank test. They will be performed on the sample to test whether the abnormal and/or cumulative abnormal returns differ significantly from zero. The results of these tests are used to assess hypotheses H1 & H2. Secondly, to test hypotheses H3 & H4; whether we see a trend emerging over time the sample is divided into multiple different sub-samples over time. These sub-sample time periods are either decided upon following a breakpoint analysis, or by clustering of years together. Where the clustering of years could be done by taking the first and last three years of the sample. A graphical data-analysis is performed to look whether a trend is visible within the data available. Following up on this the various sub-samples created will be utilized to perform an equality of means test to assess whether there are differences in time periods. To account for non-normality in sub-samples a non-parametric test on equality of medians will also be performed. The tests used will be the Wilcoxon/ Mann Whitney test and the Kruskal-Wallis test. The Wilcoxon/ Mann Whitney is chosen because of it is still a useful tool to test a difference in sample medians even if the assumption of equal distribution between samples is violated. The Kruskal-Wallis test is performed because of its high statistical power when the samples have equal variances.
4. Data
following an assessment of the firm’s performance by the Ethibel Forum. They base their assessment following a company report provided by Vigeo S.A. These company reports rate the firms on six different categories, namely; Human Resources, Environment, Corporate Governance, Business Behavior, Community Involvement, and Human Rights. The Ethibel Excellence Europe index constitutes of European firms from 18 different countries, whereas the Ethibel Excellence Global constitutes of firms from 21 different countries around the world (of which 18 European firms similar to those in the EEE). An important difference between the two indices is that the EEE constitutes of the leaders in each industry totaling up to 200 firms in the index, whereas the EEG has no given number of constituents and only includes global leaders on CSR. The decision on the constituents is made in closed quarters by the members of Forum Ethibel a couple of days prior to the official announcement day. Prior to the official announcement day the results of this assessment are not allowed to be made available. The official announcement of the change of constituents of the indices is made public through an official press release by Vigeo and the Ethibel forum, on top of that a notification of the changes in the indices is emailed to the mayor fund managers and other subscribers of the newsletters related to the indices. The date of this official public announcement is used as the event day within this study.
5. Results.
This part is structured as following; First of all, Hypotheses H1 & H2 will be tested using the full sample over 2008 till 2013. This will be done by testing the equality to zero tests on the observed CAR’s. Secondly, a break point analysis is performed in order to justify a possible break point within the data. Following up on that, the sub-samples derived from the break point analysis will be used to test Hypothesis H3 & H4. This will be done by performing an equality of means test between the derived sub samples.
H1 & H2
of the t-tests and the non-parametric tests are presented in Table 2. For the additions we see a highly significant negative sign on day 0, both for the parametric and non-parametric tests. However, if we look at the 3 day event window the sign is negative but with a probability which is only significant at a 10% significance level for the parametric t-tests, both non-parametric tests show no significant results. These results can be interpreted as being somewhat similar to those found by most earlier studies, (Table 2) who found mostly insignificant results for the abnormal returns following an announcement of a firm’s inclusion into a sustainability index. On the other hand the negative sign and the near significance can be seen as supporting Oberndorfer (2012) and Chueng, Roca (2013) findings of a negative stock market reaction following the announcement of a firm’s inclusion into a sustainability index. These results lead to the conclusion that over the time period between 2005-2013 there is a negative effect of addition into the Ethibel sustainability indices over the 3 day event window. However, these results are only significant at a p-value of 0.1, implying that the strength of these results is only weak. Given that the chosen significance level was at 95% this value is not statistically significant. This means that for this sample we reject H1a given the 3-day event window used. This consequently means that H1b holds for the database used. As an important note it has to be mentioned though; when solely observing the SAR on day-0 the results are a highly significant with a negative sign for the SAR on day-0.
The results for the deletions out of the index are presented in Table 2. Similarly to the results for additions the results show a statistical negative sign for the 3 day event window, and again only at a p-value of 0.1. However, there is no significant effect on any of the 3 separate days (-1, 0, +1). Given the 95% significance level chosen we cannot conclude that there is any significant negative effect of a deletion out of the index. These results are contrary to those found by earlier studies, which find significant negative results for the abnormal returns following the announcement of a firm’s deletion out of a sustainability index (Becchetti et al 2012, Doh et al 2010, Chueng & Roca 2013, Consolandi et al 2009). Consequently we reject H2a, and we accept H2b as it seems that the announcement of a deletion from the index has no significant impact on the abnormal stock performance of the firm.
First of all graph I provides a graphical illustration of the development of the abnormal returns in the 3 and 4-day event window around the announcement of a firm’s inclusion in either the EEE or EEG. Though there seems to be a rather high fluctuation between the different years, possibly caused by small samples, there is an upward trend visible over the time period 2005 to 2013. Especially as of 2009 their seems to be a gradual upward trend, though there still seems to be a high volatility between years. Even though the high volatility is apparent, this graph can be interpreted as an indication that hypotheses H3a and H3b could be confirmed. Nonetheless, the graph just provides a visual illustration of the database clustered into different years. What is of importance is whether the CAR’s significantly differ from zero, and whether there is a difference found within the sub samples.
Graph I
CAR over the years for additions
Graph shows the yearly developments of the average CAR following the announcement of an addition into the Ethibel Excellence indices. Both for the 3and 4-day event window.
To test whether a trend is visible within the dataset several different sub samples have to be decided upon. One possible technique to test for differences over time is to cluster different years together, for example to cluster the first three years and the last three years into two different groups, and subsequently testing them for differences in mean and/or medians. In the case of this dataset this would result in a sub-sample from 2005-2007 and one from 2011-2013. Another way to decide upon different sub-samples is a break point analysis which essentially looks to divide
the sample based on important break points over time. Such a break point can be derived from the theory, or from a statistical analysis of the data. A possibility in the case of sustainability indices could be a sudden change in criteria or publication method. In the case of the Ethibel indices there does not seem to be such an important breaking point linked to the index specifically. However, when expanding the analysis to encompass the SRI environment as a whole there is an important break point visible. The latest Eurosif report (2012) show an exceptional growth in sustainability themed investments from €25 billion in 2009 to €48 billion in 2011. Whereas between 2007 and 2009 this amount stayed relatively constant at €25 billion. This would indicate that as of 2009 the SRI investment experienced a sudden rise in size. Expanding the analysis even more to encompass the entire business environment inevitably the financial crisis in 2008 cannot be factored out as an influential event in time. To reduce the effect of this event in time two sub-samples can be selected excluding the data from 2008, resulting in a sub-sample from 2005-2007 and on from 2009-2011. Theoretically then, these three sub-samples could be derived. When performing a break point analysis on the addition database using a Bai- Perron test we find a breakpoint exactly September 2009. (See appendix) Consequently, choosing sub-samples before and after 2009 seems to be statistically justified within our sample of additions.
Given the indication of possible break points and the reasoning for different clusters three different sub samples are selected and tested for equality to zero and equality between each other. The results for the individual sub-samples abnormal returns tested for equality to zero are presented in table 3 and 4.
H3: Trend in additions over time
day-1 in the latter two samples. These results are in line with our expectations derived from the graphs and can be interpreted as confirming our H3a, that their seems to be a trend visible over time. To confirm this expectation that there is a difference over time visible for our sub sample the individual samples are tested for equality in means. The results of these tests for our 3-day event window are presented in table 5. As expected from the analysis of the results in table 3 and the results of the equality of means test provide statistically highly significant results confirming a difference in means between the earlier and later sub-samples. These results can be interpreted as an indication that the effect of the announcement of an addition into a sustainability index changes over time. Moreover, they indicate that the announcement had a negative effect on stock prices in the earlier years from 2005-2007, but they changed to having a positive effect in the more recent years. Thus H3b can be assumed to hold given the dataset used.
H4: Trend in deletions over time
Table 4 presents the test results for the cumulative and standardized abnormal returns for the different sub-samples. The test results show a negative sign which is statistically significant for the sub sample 2005-2007. However, as the table shows this is based on 52 observations, which can be considered as being a small sample. For the second sub sample ranging from 2009-2013 no statistical significant deviation from zero can be observed in the dataset. The third sample ranging from 2011-2013 shows a significant positive sign for day-1. Furthermore, a positive median is found to be statistically significant for the 3 and 4-day event windows if the non-parametric rank (normal approx.) test is used.
TABLE 2
Test results for the entire sample 2005-2013
Table shows the test results for the parametric t-test, and the non-parametric tests for the equality to zero test for both the standardized abnormal returns (SAR,) and the cumulative abnormal returns (CAR). Included are both the addition and the deletion announcements from the Ethibel Excellence indices. N=number of observations.
Parametric Non-Parametric
Event
Windows SAR CAR Prob.
TABLE 3
Test results for the individual addition sub-samples 2005-2007, 2009-2013, and 2011-2013
Table shows the test results for the parametric t-test, and the non-parametric tests for the equality to zero test for both the standardized abnormal returns (SAR,) and the cumulative abnormal returns (CAR). Included are the 2005-2007, 2009-2013, and 2011-2013 sub-samples for addition announcements into the Ethibel Excellence indices .N=number of observations.
Parametric Non-Parametric
Event
Windows SAR CAR Prob.
TABLE 4
Test results for the individual deletion sub-samples 2005-2007, 2009-2013, and 2011-2013
Table shows the test results for the parametric t-test, and the non-parametric tests for the equality to zero test for both the standardized abnormal returns (SAR,) and the cumulative abnormal returns (CAR). Included are the 2005-2007, 2009-2013, and 2011-2013 sub-samples for deletion announcements out of the Ethibel Excellence indices .N=number of observations.
Parametric Non-Parametric
Event
Windows SAR CAR Prob.
Table 5
Equality of means test between sub-samples
Table shows the test results for the parametric t-test, and the non-parametric test for the equality of means or medians between the different sub-samples 2005-2007, 2009-2013, and 2011-2013. These tests are performed on the CAR for the 3-day event window (-1, +1) of the different sub-samples. Df = degrees of freedom.
Parametric Non-parametric
First event window
6. Discussion
The question is what these results mean in practice? How can these results be interpreted in relation to the Corporate Social Performance and Corporate Financial Performance relationship? Can any assumptions be made on the evolution of this relationship over time following the results found in this study? And how do the results relate to the results of the earlier event studies performed on this subject?
test results. One possible reason for this could be that the recent significant growth developments mentioned in the EuroSif and USSif studies occurred as of 2009. As most previous study samples only range to 2008 (Table 1) they could not have captured this change in environment in their results. The sample in this study however ranges from 2005 to 2013. Consequently, there is a possibility within the sample to encompass the development in SRI fund growth as of 2009.
First of all, when looking at the results for the (cumulative) abnormal returns for the additions in the three different sub-samples (2005-2007, 2009-2013, 2011-2013) some interesting results are observed. The sub-sample ranging from 2005-2007 shows statistically very significant negative values (Table 3). Following the argument of this paper this would suggest that in the period prior to 2008, a firm its efforts to improve its CSP are penalized by the market and thus lead to a decreased CFP. The two sub-samples ranging from 2009-2013 and 2011-2013 however show statistically significant positive abnormal returns following the announcement. Hence, these results suggest that in the time period as of 2009 the market reaction changed. It appears that a firm its efforts to improve its CSP are now appreciated by the market and hence lead to an enhanced CFP. To test whether this difference can statistically be confirmed an equality of means test between the different sub-samples is performed. (Table 5) These results confirm that there is a statistical highly significant difference in the CAR’s for the event windows of the different sub-samples. Consequently both H3a and H3b are supported by the results; there is a change in the effect over time following an announcement of an addition into the index. Moreover, this effect has changed from negative between 2005-2007 to positive in 2009-2013. This is in line with the expectations derived from the studies by EuroSif and USSif (2012), which documented a significant growth in importance of SRI as of 2009. These findings are relatively novel in the context of event studies performed on sustainability indices; previous event studies have not been able to include the development in SRI as of 2009, since their samples did not range beyond 2008.
Furthermore, to be deleted out of the index, a firm first has to part of the index. This would suggest that somewhere in the past the firm has put effort into its CSP relative to its peers. When it then loses this status this can be perceived by investors as an indication of less than optimal management, or other problems threatening the organization.(Geunster et al., 2006) When analyzing the previous studies (Table 1) we observe that this negative reaction to the deletion out of the index is found the majority of the studies. The test results of the sub sample 2005-2007 show similar results; a statistically significant abnormal return following the announcement of a deletion of the firm out of the index. Following up on this, given the results for the additions, the expectation is that the reaction on the deletion out of the index in the two sub-samples as of 2009 (2009-2013, 2011-2013) would become increasingly negative. However, the results show no significant abnormal returns following the announcements for these two sub samples. Thus even though a change in time is observed in the sample following the equality of means tests (Table 5), supporting H4a, the effect does not become increasingly negative. Consequently we cannot accept H4b, as the results are contrary to our expectations. What might have caused these results remains unknown, as it is beyond the scope of this study. However, a possible reason could be that as SRI has grown in importance through the years as supported by our results for additions, the CSP of firms have received increased attention as well. A great deal of the deletions from the index is caused by extraordinary events harming the CFP of the firm (e.g. the oil spill by BP, Apple’s problems with Chinese factory labor conditions. (Wall street Journal, 2011)). As these events are under greater attention, the stock market might have responded when this information was made public. Consequently, when at the half-yearly review the firm was deleted from the index this did not contain any new information. Another possible explanation could be that the fact that a firm is deleted from the index does not necessarily mean that the company has a poor CSP. It could still be that the firm has relatively high CSP; it’s just not at the top of the industry anymore. Already Doh et al (2009) found in their studies that a high prior CSR reputation (CSP) mitigates the negative effects of news harming their CSP reputation such as a deletion from a sustainability index. As these are only assumptions, we cannot confirm nor reject these arguments. Finding the actual drivers of these results provides an interesting field for future research.
test with the 3-factor model to predict the normal returns. This study does not contain such a robustness check. Another aspect which has to be considered is the actual economic relevance of this study. Even though the results do have a statistical significance as shown in the test results, the economic impact is relatively small. The abnormal returns in the stock price translate to actual fluctuations of the stock price below 0,01%. This is by any means only a marginal fluctuation in the stock price, and thus only has a marginal impact on the firm value. Economically, such a marginal influence on stock prices will probably be not high on the agenda of management or investment managers.
What do these results add to the ongoing discussion on the relationship between CSP and CFP? Overall, the results of this study support the idea that SRI has grown in importance to such an extent that engagement in CSR (CSP) is rewarded by the stock market. This does support the argument of the academics favoring stakeholder theory. Apparently, investment in the improvement of a firm its CSP leads to an increased CFP. What do these results mean for the managers in firms today? Even though there are some limitations. The results of this study provide a compelling argument that efforts to improve the firm its CSP are rewarded by the stock market. Consequently, following the argument of this study, efforts to be included in a sustainability index are rewarded with an improved CFP. It is thus beneficial for a manager today to pursue a strategy aimed at increasing the firms CSP, as this is leads to a better CFP. In other words, managers are encouraged to engage in environmental programs, social development programs and other voluntary commitments to increase the CSP of the firm.
7. Conclusion
abnormal returns over the entire sample ranging from 2005-2013 show no significant deviation from zero. These results imply that a firm’s commitment to improve its CSP has no negative, nor positive effect on the firm’s CFP.
8. Acknowledgements
I would like to thank prof. Dr. L.J.R. Scholtens first and foremost for his assistance, guidance, and useful comments which made this thesis possible.
Furthermore, I would like to thank Ir. C.H. Riphagen for reviewing my writing.
On top of that I’d like to thank T. van Esch, M. Wever, and J.W. Stoffers for their moral support during the process of writing this thesis.
9. References
Barnett, M. L., Jermier, J. M., & Lafferty, B. A. (2006). Corporate reputation: The definitional landscape. Corporate Reputation Review, 9(1), 26-38.
Becchetti, L., Ciciretti, R., Hasan, I., & Kobeissi, N. (2012). Corporate social responsibility and shareholder's value. Journal of Business Research, 65(11), 1628-1635.
Becchetti, L., Di Giacomo, S., & Pinnacchio, D. (2008). Corporate social responsibility and corporate performance: evidence from a panel of US listed companies. Applied Economics,
40(5), 541-567.
Brown, S. J., & Warner, J. B. (1985). Using daily stock returns: The case of event studies.
Journal of financial economics, 14(1), 3-31.
Cañón-de-Francia, J., & Garcés-Ayerbe, C. (2009). ISO 14001 environmental certification: a sign valued by the market?. Environmental and Resource Economics, 44(2), 245-262.
Cheung, A. W. K. (2011). Do stock investors value corporate sustainability? Evidence from an event study. Journal of Business Ethics, 99(2), 145-165.
Cheung, A. W. K., & Roca, E. (2013). The effect on price, liquidity and risk when stocks are added to and deleted from a sustainability index: Evidence from the Asia Pacific context.
Journal of Asian Economics, 24, 51-65.
Clacher, I., & Hagendorff, J. (2012). Do announcements about corporate social responsibility create or destroy shareholder wealth? Evidence from the UK. Journal of business ethics,
Consolandi, C., Jaiswal-Dale, A., Poggiani, E., & Vercelli, A. (2009). Global standards and ethical stock indexes: The case of the Dow Jones Sustainability Stoxx Index. Journal of
Business Ethics, 87(1), 185-197.
Cording, M., Christmann, P., & King, D. R. (2008). Reducing Causal Ambiguity In Acquisition Integration: Intermediate Goals As Mediators OfIntegration Decisions and Acquisition Performance. Academy of Management Journal, 51(4), 744-767.
Cording, M., Christmann, P., & King, D. R. 2008. Reducing causal ambiguity in acquisition integration: Intermediate goals as mediators between integration decisions and acquisition performance. Academy of Management Journal, 51(4): 744-767.
Derwall, J., Guenster, N., Bauer, R., & Koedijk, K. (2005). The eco-efficiency premium puzzle. Financial Analysts Journal, 51-63.
Doh, J. P., Howton, S. D., Howton, S. W., & Siegel, D. S. (2010). Does the market respond to an endorsement of social responsibility? The role of institutions, information, and
legitimacy. Journal of Management, 36(6), 1461-1485.
Donaldson, T., & Preston, L. E. (1995). The stakeholder theory of the corporation: Concepts, evidence, and implications. Academy of management Review, 20(1), 65-91.
Eurosif. 2012. ‘European SRI study 2012’. Available at
http://www.eurosif.org/research/eurosif-sri-study (accessed 5december 2013).
Fama, E. F., & MacBeth, J. D. (1973). Risk, return, and equilibrium: Empirical tests. The
Journal of Political Economy, 607-636.
Fisher-Vanden, K., & Thorburn, K. S. (2011). Voluntary corporate environmental initiatives and shareholder wealth. Journal of Environmental Economics and Management, 62(3), 430-445.
Guenster, N., Bauer, R., Derwall, J., & Koedijk, K. (2011). The Economic Value of Corporate Eco‐Efficiency. European Financial Management, 17(4), 679-704.
Haleblian, ]., Kim, )., & Rajagopalan, N. 2006. The influence of acquisition experience and performance on acquisition behavior: Evidence from the US commercial banking industry. Academy of Management journal, 49(2): 357-370.
Hawn, O., & Chatterji, A. (2014). How Firm Performance Moderates the Effect of Changes in Status on Investor Perceptions: Additions and Deletions by the Dow Jones Sustainability Index. Available at SSRN 2418300.
Jones, T. M. (1995). Instrumental stakeholder theory: A synthesis of ethics and economics.
Academy of management review, 20(2), 404-437.
Kale, P., Dyer, J. H., & Singh, H. 2002. Alliance capability, stock market response, and long-term alliance success: The role of the alliance function. Strategic Management journal, 23(8): 747-767.
Lyon, T., Lu, Y., Shi, X., & Yin, Q. (2013). How do investors respond to Green Company Awards in China?. Ecological Economics, 94, 1-8.
M. Friedman, The Social Responsibility of Business is to Increase its Profits, The New York Times Magazine, 13 September 1970.
MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of economic
literature, 13-39.
Martin Curran, M., & Moran, D. (2007). Impact of the FTSE4Good Index on firm price: An event study. Journal of environmental management, 82(4), 529-537.
McWilliams, A., & Siegel, D. (1997). Event studies in management research: Theoretical and empirical issues. Academy of management journal, 40(3), 626-657.
McWilliams, A., Siegel, D. S., & Wright, P. M. (2006). Corporate social responsibility: Strategic implications*. Journal of management studies, 43(1), 1-18.
McWilliams, A., Siegel, D., & Teoh, S. H. (1999). Issues in the use of the event study methodology: a critical analysis of corporate social responsibility studies. Organizational
Research Methods, 2(4), 340-365.
Oberndorfer, U., Schmidt, P., Wagner, M., & Ziegler, A. (2013). Does the stock market value the inclusion in a sustainability stock index? An event study analysis for German firms.
Journal of Environmental Economics and Management, 66(3), 497-509.
Table 6
Descriptive statistics for sub-samples.
