The impact of Corporate Social Responsibility on stock returns around mergers & acquisitions for US firms Thesis MSc Finance

Thesis MSc Finance

The impact of Corporate Social Responsibility on stock returns

around mergers & acquisitions for US firms

Name: Thys Jacobs

Student number: S2569094

Study Program: MSc Finance

Supervisor: A. Dalò

Keywords: Mergers & Acquisitions, Corporate Social Responsibility, abnormal

returns

Date: 09-01-2019

ABSTRACT

Table of contents

1. Introduction ... 3

2. Literature review ... 5

2.1 CSR and firm value ... 5

2.2 Ways to value CSR ... 6

2.3 CSR and stock performance ... 8

2.4 Factors influencing abnormal returns in M&A’s ... 9

3. Methodology ...10

3.1 Research method ...10

3.2 Control variables ...13

4. Data ...14

4.1 CSR data collection ...14

4.2 M&A data collection...15

4.3 Sample ...16

5. Results ...16

5.1 Summary statistics ...16

5.2 Regression analysis ...21

5.2.1 Regression first hypothesis ...22

5.2.2 Regression second hypothesis ...25

5.3 Robustness tests ...28

6. Conclusion ...30

7. Discussion ...31

8. References ...33

Appendices ...37

Appendix A: Pearson correlation matrices and VIF ...37

Appendix B: Jarque-Bera tests ...39

Appendix C: Regressions subsamples ...40

Appendix D: Regression analysis different windows ...44

Appendix E: Fama-French regression ...50

1. Introduction

Over the last years there has been an increase in Mergers & Acquisitions (M&A’s), according to the Institute of Mergers, Acquisitions and Alliances (IMAA). The IMAA (2018) reports that the amount of deals increased from 38,641 in 2013 to 52,745 in 2017 worldwide (IMAA, 2018). The highest value of the deals was in North America, while most deals were made in Europe (IMAA, 2018). When looking at the deal values, the IMAA (2018) reports that the total deal value increased since 2013, except for 2016 where the deal value decreased. This is confirmed by the statistics of Statista, which showed that the total deal value increased from 2.53 trillion US dollars to 3.66 trillion US dollars in 2017 (Statista, 2018). Considering these statistics, it is clear that M&A’s are an important aspect for businesses and it may be a part of the business strategy for various reasons, such as diversification or in order to grow. However, forecasts are unclear about the future of M&A’s. The IMAA (2018) forecasts that there will be a decrease in the number of deals in 2018 to 45,000 worldwide. On the other hand, Deloitte asked 1,016 executives at corporations and private equity firms about their thoughts of the volume and value in M&A’s in 2018, and more executives believe that both the volume size and value of M&A’s will increase in 2018 than executives that think the volume and size will either stay the same or decrease (Deloitte, 2017). Giving the attention received by M&A’s and since the IMAA (2018), Deloitte (2017) and Statista (2018) all expect an increase in the total deal value of M&A’s, it is interesting to study different aspects that may play a role around M&A’s.

With the large amount of M&A’s and the importance of CSR, it is interesting to study whether these concepts impact each other. As mentioned before, companies with a high CSR performance produce better results than companies without a high CSR performance. Keeping this in mind, there is a chance that CSR positively impacts the stock price around M&A’s. Nevertheless, there is also evidence that CSR has a negative impact on the abnormal returns around M&A’s (Lagas, 2013). Both the CSR of the acquirer and target might impact the abnormal returns of one of the companies. A report by PwC, on behalf of PRI (Principles of Responsible Investments), assessed whether or not acquirers take CSR in consideration when targeting a company (PwC & PRI, 2012). The report stated that poor CSR performance can have a negative impact on the deal value. However, more than 50% of the respondents mentioned that good CSR performance can also lead to a higher deal value (PwC & PRI, 2012). Therefore, acquirers might be willing to pay a premium for the CSR performance when it comes to M&A’s, since these targets have a chance to perform better as a result of the CSR performances (compared to companies with a low CSR performance). This can result in high abnormal returns around and after the announcement date. The respondents of the report of PwC and PRI (2012) also stated that they thought that the impact of CSR on M&A activities would increase: 75% of the respondents predicted that there would be a large increase three years from 2012 (so up to and including 2015). Firms may have several reasons to engage in CSR activities. The most important reasons to engage in CSR are the following: instructions from the Board of Directors or top management, to attract employees, reputation, moral and for economic purposes (Österman, 2014). Another paper supports the study by Österman (2014) and also states that CSR investments are not only driven by profit motives (Banerjee, Homroy, & Slechten, 2015).

Based on the possible relationship between CSR and abnormal returns around M&A’s, the research question is: If there is a relationship between CSR and M&A’s, what components of

CSR have an impact on abnormal returns in M&A’s for US firms?

Commission is, for example, encouraging firms to undertake CSR activities (European Commission, 2011). Studies showed that US firms are discussing their CSR activities more in public than European firms, and that they are presenting it as an extension of their core values (Maignan & Ralston, 2002, p. 507). Another study also reported differences between the amount of CSR activities between European and American firms and also found different motivations to engage in CSR activities between them (Forte, 2013). This might induce a difference in the way shareholders value CSR and, therefore, it is interesting to investigate if CSR creates value for shareholders around M&A’s for US firms. The results of this study will be compared to the recent results for European firms. In this study, firm value will be defined as the market capitalization, which is a multiplication of the number of outstanding shares and the current stock price (Adenugba, Ige, & Kesinro, 2016). This study focusses on the abnormal returns, which is measured using stock prices.

The remainder of this paper will be as follows. First an overview of the literature about the different topics is given in section 2, including the hypotheses tested in this study. Section 3 explains the methodology used in this study, followed by the discussion of the data(sources) in section 4. Results can be found in section 5, with a conclusion in section 6. Section 7 is a discussion of this study and the results found. In section 8 one can find the references and the paper concludes with several appendices.

2. Literature review

a low customer awareness, the relation is negative or insignificant (Servaes & Tamayo, 2013). Not only high CSR can result in a higher value, but low CSR can also result in lower firm value and customer loyalty (Sawhny, 2008).

Besides the conducted studies about the CSR and firm value, there are also some studies that investigated the impact of CSR on M&A deals. Most studies found a positive relation between the value bidders gave to the target and the CSR performance (Ackerman, 2016; Deng, Kang, & Low, 2013; Gomes & Marsat, 2017), but not a lot of research has been done on this topic and there are various methods to conduct such a study. For example, what countries are the focus of the study, and which CSR measurements are being used. In addition, it is important that the firms have to be transparent about their CSR actions. If they do not share information about their CSR quality, it is difficult for the public to accurately perceive it (Graafland, Eijffinger, & Smid, 2004). That is one of the reasons why firms publish (part of) their CSR measures as a part of their annual reports. By doing so, they want to inform their stakeholders and customers about their CSR. Firms with a high CSR score are taking stakeholders interests’ more in consideration than firms with a low CSR score when making investments (Deng et al., 2013). This may result in the situation where stakeholders accept favorable contracts when a firm has a high CSR score1_{. Stakeholders play a role in the outcome of the acquisition process,}

because they have to reach an agreement. One could argue that bidding firms are able to complete deals faster and probably more profitable when stakeholders are willing to accept a favorable contract (from the bidder’s perspective). This indicates that firms with a high CSR score make socially responsible investments, where stakeholders are able to make a good deal and that it is perceived as such by their stake- and shareholders. This may result in higher stock returns compared to a situation where the M&A process is taking longer, because stakeholders do not reach an agreement. Considering this, M&A’s are important events to study the relationship between CSR and stock returns.

2.2 Ways to value CSR

Because of the increasing focus on CSR, the number of databases that measure CSR also increased. There are some well-known databases with CSR scores for a lot of companies. Bouten, Cho, Michelon & Roberts (2016) reported that there are currently three major suppliers of CSR ratings. They found that studies with a topic that involves CSR and which are published in one of the major journals of accounting, finance and management, mostly use one of the following CSR ratings suppliers: Thomson Reuters (ASSET4 database), MSCI (ESG Intangible Value Assessment, formerly known as KLD) and Sustainalytics (ESG indicator). The

1 _{When performing a merger or acquisition, stakeholders have to give their approval. For example employees}

conclusion of their paper is that there are differences in the way these CSR raters are constructing their ratings. This might be due to the fact that different CSR rating providers have different theorizations of CSR and, therefore, measure CSR performance differently (Bouten, Cho, Michelon, & Roberts, 2016, p. 29). Another result is that the measurement process is highly subjective, which implies that even two analysts from the same rating agencies might score the CSR performance of a company differently (Bouten et al., 2016, p. 29). Both results indicate that it can be difficult to completely rely on the scores given.

& Wolniak (2016) found that the overall quality is generally low. In addition, they found that the legislation of including CSR data increases the quality of the CSR reports. To improve the quality of the CSR reports, the Global Reporting Initiative (GRI) introduced guidelines that firms can follow when reporting their CSR activities.

2.3 CSR and stock performance

It might seem obvious to think that CSR will influence stock performance in a positive way, because when a report with a high CSR outcome is published, stakeholders will be pleased, but when the CSR outcome is low they might be disappointed and have an incentive to sell the stocks, leading to lower stock prices. However, former studies observed contradicting results about the relationship between CSR and stock prices.

There are some studies that indeed found a positive relationship between the CSR performance of a firm and their stock performance and/or annual returns (Dornean & Oanea, 2017; Huang & Zong, 2016; Mwamburi, 2013). Nevertheless, other studies found evidence of a negative relationship between the two (Brammer, Brooks, & Pavelin, 2016; Zhang, 2017). The differences between those studies may be partly explained by the different regions they focused on. Based on the data used in these studies, one cannot explain the differences due to different time-periods. Another study found a positive relationship for the USA, but neither positive nor negative outcomes for Europe (Von Arx & Ziegler, 2013). Looking at the industry-level, Von Arx & Ziegler (2013) again found neither positive nor negative results for both regions. This could be explained by the fact that some studies only focused on environmental and social activities, and that different databases for the CSR scores were used. Some studies used the KLD database (Zhang, 2017), whereas another study used data from the Swiss Bank (Von Arx & Ziegler, 2013), or annual reports from the firms (Dornean & Oanea, 2017). Huang & Zong (2016) used a list from the Corporate Responsibility Magazine to obtain the CSR scores and Brammer et al. (2016) used The Ethical Investment Research Service (EIRIS). These differences in CSR ratings may influence the differences in results.

Based on the previous literature and keeping the increased interest in CSR and M&A’s in mind, the following hypotheses will be tested:

H1: The CSR performance of the acquirer will have a positive impact on the cumulative

abnormal returns of the acquirer just before and right after the announcement date.

score. This is based on the idea that firms with a high CSR score are operating in the interests of their stakeholders more than low-scoring CSR firms (Deng et al., 2013). Besides the hypothesis that focuses on the CSR performance of the acquirer, the following hypothesis will focus on the CSR performance of the target and its impact on the abnormal returns of the acquirer.

H2: The CSR performance of the target will have a positive impact on the cumulative

abnormal returns of the acquirer just before and right after the announcement date.

The idea behind this hypothesis is that when a firm acquirers a firm with a high CSR score, it can be perceived as a socially responsible investment and, therefore, leads to an increase in the valuation by shareholders.

2.4 Factors influencing abnormal returns in M&A’s

There are several factors that can influence the abnormal returns of a firm. To make sure this study only measures the influence of CSR on abnormal returns around M&A’s, these factors will be included in the regression model as control variables. This section will cover the different control variables, based on existing literature. This section is mainly based on the paper of Masulis, Wang & Xie (2007), as most of previous similar studies did.

For deal characteristics, the first control variable is the method of payment. Acquisitions can be paid in various ways, namely, cash, stock or partially cash/stock. Studies showed that acquisitions that are (partially) stock paid, result in negative abnormal returns (Masulis et al., 2007). Since it affects the abnormal returns, it has to be one of the control variables. Another control variable is whether or not the deal is between two firms in the high-tech sector. Masulis et al. (2007) expects that when firms are both in a high-tech sector, the acquirer is experiencing negative returns due to the difficulty of integration in those sectors. The last control variable is the relative deal size. Masulis et al. (2007) state that relative deal size has an effect on abnormal returns, however the evidence on whether it is a positive or negative effect is ambiguous. In section 3.2 there will be a more comprehensive explanation of how the variables are used and measured in this study.

3. Methodology

Because previous studies showed that it is unclear when CSR improvements result in a higher firm value, this study will use subsamples for all CSR components based on their scores, besides the regressions for the whole sample. The components that will be considered are Environmental, Social and Governance (ESG). Using a univariate approach, the sample will be ranked for each component from high to low. The sample will then consist of six subsamples, as shown in Table 1 below. With those subsamples a clear overview of the impact on CAR for each component is given. The median of each component will be the critical value: if a firm has a score above the median it is categorized in the ‘high’ subsample, if the score is below the median the firm is categorized in the ‘low’ subsample. For the market returns, this study uses the Fama-French North American three factors data, which is freely available.

Table 1. Overview subsamples

This table provides an overview of the different subsamples for the ESG pillars. Based on the medians of the three components each firm will be divided into one of the groups. If the score is below the median the firm will be in the ‘low’ subsample, otherwise in the ‘high’ subsample.

Subsample Description

High environment Firms with a high environmental score Low environment Firms with a low environmental score High social Firms with a high social score

There are several ways to calculate the expected returns in an event study. The expected returns are needed to calculate the abnormal returns. For example, one can choose the market-model, Capital Asset Pricing Model (CAPM) or one of the Fama-French models (Moeller et al., 2004). The difference between the market-model and the CAPM is that the market-model takes the intercept as a constant, where the CAPM uses the risk-free rate as intercept. The Fama-French models are an extension of the CAPM by using more risk factors. The Fama-French three factor model for example combines the market excess return factor with two other factors: SMB (small minus big, for the size) and HML (high minus low, for book-to-market-equity) (Sitthipongpanich, 2011). This study uses the CAPM to calculate the expected returns and the Fama-French three factor model as a robustness test. The estimation window will be 200 days, following Masulis et al. (2007). The event window will be 7 days (-3, +3), with the event day (t=0) being the announcement date. The reason behind this is that the market and shareholders have time to respond to the acquisition announcements. Besides the (-3, +3) event window, results for other event windows are also presented (Lagas, 2013; Masulis et al., 2007; Westerbeek, 2018). In an event study, the abnormal returns need to be calculated. The formula for the abnormal returns is as follows:

𝐴𝑅_$,& = 𝑅_$,&− 𝑅)_$,& (1)

Where 𝐴𝑅_$,& are the abnormal returns for firm i at time t, 𝑅_$,& are the actual returns for firm i at time t and 𝑅)$,& are the expected returns. To calculate the abnormal returns, you first need to calculate the actual returns. The CAPM gives the following formula for expected returns:

𝑅)$,&− 𝑅*,& = 𝛼,$ + 𝛽/$(𝑅1,&− 𝑅*,&) (2) In formula (2) 𝑅_1,& is the return of the market portfolio and 𝑅_*& is the risk-free rate. The terms

i and t are again firm i and time t. To calculate the returns this study uses the following formula:

𝑅_$,& =(34,5634,578)9:4

34,578 (3)

Where 𝑃_$,& is the share price at time t and 𝑃_$,&6< is the share price one day before 𝑃_&, with i indicating the firm. The D stands for the dividend of firm i. Following MacKinlay (1997), the formula to calculate the cumulative abnormal returns is as follows:

𝐶𝐴𝑅_$(&<,&>)= Σ_&<&>_𝐴𝑅

𝐶𝐴𝑅_$(&<,&>) are the cumulative abnormal returns for firm i between t1 and t2. In this study multiple time ranges are studied, with (-3,+3) to start with2_{. To test whether or not there is a}

normal distribution the chi-square test will be used. In case the returns are indeed normally distributed, the parametric t-test will be used to test the significance. If the chi-square test indicates a non-normal distribution I will use a non-parametric test, besides the parametric t-test. This study uses the chi-square test over the Jarque-Bera test, because the Jarque-Bera test is more accurate for large samples with thousands of observations, where the chi-square adjusts for small samples. The chi-square is based on the skewness and kurtosis and will be conducted through Stata. However, the Jarque-Bera test will also be performed in order to compare the results with the chi-square test. Afterwards, regression analyses will be performed in order to estimate the relationship between the variables using a linear regression model (OLS). For each ESG component I will run a regression analysis to test the impact of each component individually. Before running the analyses, the assumptions of the OLS will be checked. The assumptions (Gauss-Markov theorem) are: unbiased, minimum variance, consistent (as N increases, the estimators reach the true parameters) and normal distribution. This means that OLS estimators are BLUE; best linear unbiased estimators.

Because the abnormal returns can be influenced by more factors than only the CSR performance, this study will also use a list of control variables, as stated in section 2.4. Aim of this list is to reduce the omitting variable bias in the estimation of the relation between CSR performance and abnormal returns.

The regression equations will look as follows: 𝐶𝐴𝑅 @@@@@@(𝑡 − 3, 𝑡 + 3)𝑎𝑐𝑞$ = 𝛼$ + 𝛽<𝐶𝑆𝑅𝑎𝑐𝑞$,&6<+ 𝛽>𝑇𝑜𝑏𝑖𝑛L𝑠𝑞 + 𝛽N𝐹𝐶𝐹 + 𝛽P𝑃𝑀𝑚𝑒𝑡ℎ𝑜𝑑 + 𝛽V𝐻𝑇 + 𝛽X𝐷𝑆 + 𝛽Z𝐼𝐷 + 𝛽\𝑌𝐹 + 𝜀$,& (5) 𝐶𝐴𝑅 @@@@@@(𝑡 − 3, 𝑡 + 3)𝑎𝑐𝑞_$ = 𝛼_$ + 𝛽_<𝐶𝑆𝑅𝑡𝑎𝑟_$,&6<+ 𝛽_>𝑇𝑜𝑏𝑖𝑛L_{𝑠𝑞 + 𝛽} N𝐹𝐶𝐹 + 𝛽P𝑃𝑀𝑚𝑒𝑡ℎ𝑜𝑑 + 𝛽V𝐻𝑇 + 𝛽X𝐷𝑆 + 𝛽Z𝐼𝐷 + 𝛽\𝑌𝐹 + 𝜀$,& (6) Equation (5) is to test the first hypothesis, whereas equation (6) is to test the second hypothesis. 𝐶𝐴𝑅

@@@@@@(𝑡 − 3, 𝑡 + 3)𝑎𝑐𝑞$ are the cumulative abnormal returns for acquirer i in the 7-day event window (the other event windows are also tested), 𝐶𝑆𝑅𝑎𝑐𝑞_$,&6< is the CSR performance of the acquirer one year before the announcement date. The other parts are the control variables, where FCF is free cash flow, PMmethod is the dummy variable for the payment method, HT is the dummy variable for high-tech firms, DS is the deal size control variable, ID is to check for

industry-effects and YF is for fixed year-effects. In equation (6), 𝐶𝑆𝑅𝑡𝑎𝑟_$,&6< is the CSR performance of the target one year before the announcement date. Furthermore is 𝛼_$ the intercept and 𝜀$,& the error term. The other terms of equation (6) are exactly the same as in equation (5).

To address the issue of possible endogeneity, this study followed Deng et al. (2013). In their study they add two instrumental variables in order to perform a two-stage least squares (2SLS) regression for possible endogeneity of the CSR scores. The instrumental variables for the CSR scores are religion rank and a dummy variable for blue states. Since this study also focuses on US firms, those measures are appropriate to use in this study. Religion rank is the number of religious adherents divided by the total population for each state of America. Each state receives a score from 1 to 51, where a score of one means low religiosity in the corresponding state and a high score means higher religiosity. In order to give every observation a score this study looked at the state in which the headquarters is located. Previous research showed that religiosity has a positive correlation with CSR, suggesting a positive relationship between a firm’s CSR and religiosity (Angelidis & Ibrahim, 2004). The blue state dummy equals one if a headquarters is in a Democratic (blue) state and zero if not. According to Rubin (2008), firms with a high CSR have their headquarters in Democratic states. Therefore, one may expect that CSR score has a positive relationship with this dummy variable.

3.2 Control variables

Following Masulis et al. (2007) there are two categories to consider when studying acquirer abnormal returns, namely bidder characteristics and deal characteristics. As stated in section 2.4, this study controls for Tobin’s q, leverage, free cash flow and firm size. For deal characteristics the following factors are considered: method of payment (full cash or [partially] stock), whether or not both firms operate in a high-tech sector and relative deal size.

As stated before, Tobin’s q is measured as the ratio of the market value of assets divided by the book value of assets. The formula is as follows:

𝑇𝑜𝑏𝑖𝑛L_{𝑠 𝑞 =}`ab

cab (7)

In formula (7) 𝑀𝑉_e is the market value of the assets and 𝐵𝑉_e the book value of the assets. How leverage is defined has been explained in sector 2.4. The data for the FCF is gathered through Eikon, where the formula is as follows:

Where FCF is free cash flow to equity holders, EBITDA is earnings before interest, taxes, depreciation and amortization and CAPEX are capital expenditures. In this study, following Masulis et al. (2007), FCF is scaled by book value of total assets. Firm size is measured as the natural logarithm of the book value of the assets (Masulis et al., 2007).

For the deal characteristics some dummy variables are included. First of all, a dummy will be introduced for the payment method, which equals one if an acquisition is fully paid with stock and zero otherwise. The second deal characteristic control variable is whether or not the deal is between two firms operating in high-tech sectors. The dummy for this control variable equals one if the deal is between two high-tech firms and zero otherwise. To determine whether or not a company operates in a high-tech sector this study used the two-digit SIC codes (Loughran & Ritter, 2004). The last control variable is relative deal size, which is calculated as deal size divided by the book value of assets (Westerbeek, 2018). Since this study focuses on the cumulative abnormal returns of the acquirers, the control variables are based on the financials of the acquirer and no target-specific control variables are used. Lastly, to control for year effects, time-fixed effects are applied in the regression model and to control for industry effects the two-digit SIC-codes are used.

4. Data

This study uses the Thomson Reuters ASSET4 database for the CSR data. This database contains information of more than 7,000 companies. They also distinguish the different ESG components. The CSR score is structured as follows: 178 comparable measures are used, distributed over ten themes, belonging to one of the three ESG components. The themes are distributed as follows: three for the environmental score, three for governance and four for social. In Table 2 there is an overview of the different themes. Raters, selected by Thomson Reuters, give ratings on a scale of 1-100. Because scores are standardized, they are comparable across different firms. The ASSET4 database gives both aggregate scores as well as scores for the separate components, which will both be studied (Reuters, 2018). They operate from different locations across the world and get data from publicly available sources. The database is characterized by the fact that they use different weights for each theme3 _{(van den Heuvel,}

2012). The ASSET4 database uses the following weights for the components: 34% for environmental, 35.5% for social and 30.5% for governance. These weights are based on the number of indicators that are in each component compared with the total indicators (Thomson

Reuters, 2018). The data of the ASSET4 database is available through Eikon, a dataset from Thomson Reuters.

Table 2. ESG themes

Table 2 provides an overview of the different themes that are included to calculate the ESG scores for the ASSET4 database of Thomson Reuters.

4.2 M&A data collection

The Zephyr database is used for the data regarding the acquisition deals. Zephyr is from the Bureau van Dijk and is the most comprehensive database with data about transactions. The database is very up-to-date, because it is refreshed every hour. It includes a lot of information about the deals, for example the deal value. Given the increased interest in M&A’s and CSR the last few years, this study focuses on deals in the period 2013-2017 involving listed American acquirers. The sample will only contain full acquisitions (100%) and therefore the acquirer is not allowed to have a stake in the target before the deal took place. Following Deng et al (2013), the deal needs to have a minimum deal value of ten million dollars. The rationale behind this is that very small acquisitions will not affect the results, preventing distorted results. To summarize the requirements above, the deals included in this study needed to meet the following requirements: announcement date between 1 January 2013 and 31 December 2017, both acquirer and target are located in the US, percentage of initial stake is a maximum of 0%, percentage of final stake is 100%, acquirers need to be listed, minimum deal value is ten million dollars and the current deal status is announced or completed.

To gather the stock data, Thomson Reuters Eikon is used. Eikon is a financial analysis tool from Thomson Reuters which contains financial information, for example stock prices, for a lot of firms worldwide. It has time-series for more than 2 million financial instruments and securities. The data for the stock returns is also gathered through Eikon. Firms for which the required stock data was not available through Eikon were gathered through Yahoo Finance. When the data was not available through Yahoo Finance, the firm was excluded from the sample due to lack of data. The data for the control variables is gathered through Eikon and Zephyr. The market return data and the data for the Fama-French model is gathered through the

Component Themes

website of K. French, using the North American 3-factor data. (French, n.d.). The market return data of Fama-French is based on all firms listed on the NYSE, AMEX, and NASDAQ (French, n.d.).

4.3 Sample

For the first hypothesis, this study started with a sample size of 1326 firms. Due to lack of available data of either stock prices, CSR scores or other variables, the final sample consisted of 465 firms. This is after the firms with several acquisitions in the period were deleted, so that windows do not overlap for more than one acquisition. For the second hypothesis, the initial sample consisted of 670 firms. However, for a lot of targets the CSR scores were not available. After gathering the CSR scores, the sample only consisted of 154 firms. For 115 firms, the stock prices were available, so the final sample for the second hypothesis consisted of 115 firms. For the remainder of this paper, when speaking of first sample it means the sample that measures the impact of acquirer CSR on acquirer CAR and second sample refers to the sample that measures the impact of target CSR on acquirer CAR.

5. Results

In this section the empirical results of this study are described. The first part starts with some summary statistics of both samples for the first and second hypothesis. Subsequently, the statistical significance of the CAR will be discussed. Finally, the results of the regression analyses are shown and interpreted. This section ends with a description of robustness tests.

5.1 Summary statistics

The first sample, consisting of 465 observations, included the dependent variable CAR (-3,+3), the independent variables ESG, environmental, social & governance and the mentioned control variables: FCF, firm size, leverage, Tobin’s q, a dummy variable if the acquisition is fully stock-paid, a dummy variable if both acquirer & target operate in a high-tech sector and the relative deal size. In Table 3 the summary statistics including the chi-square test for normality are given for the first sample.

and, therefore, this study will use a non-parametric test besides the parametric t-test. The mean of the CAR is 1.04%, which is higher than the CAR of Masulis et al. (2007). They found a CAR with a mean of 0.22%, but they used a different window (-2,+2) which can explain the difference. The sign of their CAR is positive, which is in correspondence with this study. In Table 4 the outcome of both the parametric t-test and the non-parametric Wilcoxon signed-rank test is given in order to test the null hypothesis that CAR = 0. Both outcomes in Table 4 show that the CAR (-3,+3) is statistically significant and that both the mean and median are not equal to zero.

Table 3. Summary statistics first sample

This table provides an overview of the summary statistics of the first sample, consisting of 465 observations. The Table includes mean, median, standard deviation, minimum value, maximum value, skewness and kurtosis. The p-value for the joint chi-square is also included.

Variable N Mean Median Std. Dev.

Table 4. Test statistics

In this table the test statistics for both the parametric t-test and the non-parametric Wilcoxon test are given for the first sample, consisting of 465 observations, as well as the corresponding p-values. The null hypothesis states that the CAR is equal to zero. Aim of these tests is to test the significance.

CAR (-3,+3) Student t-test Wilcoxon t-test

t-statistic 3.095 3.332

p-value 0.002 0.001

The Pearson correlation matrix for the sample of the first hypothesis can be found in Table A.1 in appendix A. The bold numbers may indicate multicollinearity (a variable is made bold if the correlation coefficient is above 0.6). As the Table shows, the only variables that may have multicollinearity are the variables related to the CSR score. The explanation for this is that the three components (environmental, social, governance) together determine the total ESG score. As a result, they are related to each other. To make sure the other variables do not suffer from multicollinearity, Table A.2 in appendix A presents the variance inflation factors (VIF) when the variables are used in one regression model, with a threshold of 10 (O’Brien, 2007). The Table shows that all of the variables are far below the threshold. However, to ensure that the results of the study are not influenced by multicollinearity, separate regressions for all of the ESG components will be performed. Another outcome of the correlation matrix is the fact that the CAR correlates negatively with all of the ESG components, except the total ESG score. This may induce a negative relationship between the CAR and the ESG pillars.

hypothesis cannot be rejected. This would mean that there are no significant differences in acquirer CAR as a result of high or low CSR scores, except for the governance score.

Table 5. Univariate analysis

This table shows some statistics for the subsamples of the first sample, as described in section 3.1. The mean is the mean of the CAR(-3,+3) for the corresponding subsample and N is the number of the observations in the subsample. The test statistics are used to analyze the differences in CAR between the subsamples.

Student t-test Wilcoxon t-test Subsample CAR

Mean

N t-statistic p-value t-statistic p-value ESG High ESG 1.04 192 _{-0.021 0.983 0.407 0.684} Low ESG 1.03 273 Environmental High Env 0.81 202 _{0.161 0.872 1.276 0.202} Low Env 0.92 263 Social High Soc 1.06 217 _{-0.064 0.949 0.670 0.503} Low Soc 1.02 248 Governance High gov 0.51 245 _{1.672 0.095 2.109 0.035} Low gov 1.63 220

Table 6. Summary statistics second sample

This table provides an overview of the summary statistics of the second sample, consisting of 115 observations. The table includes mean, median, standard deviation, minimum value, maximum value, skewness and kurtosis. The p-value for the joint chi-square is also included.

Table 7. Test statistics

In this table the test statistics for both the parametric t-test and the non-parametric Wilcoxon test are given for the second sample, consisting of 115 observations, as well as the corresponding p-values. The null hypothesis states that the CAR is equal to zero. Aim of these tests is to test the significance.

CAR (-3,+3) Student t-test Wilcoxon t-test

t-statistic -2.905 -3.321

p-value 0.004 0.001

Variable N Mean Median Std. Dev.

The results of the univariate analysis for the subsamples of the second sample, as described in Table 1, are shown in Table 8. Since the second sample consisted of 115 observations, the subsamples are smaller than the subsamples of Table 5. The Table shows that the mean CAR’s are negative for all the subsamples, what can be confirmed by the mean of the total sample in Table 6. Further, one can see that for two of the four pillars (social and governance) firms with a lower CSR score have better returns, although still negative. The same tendency has been seen in Table 6. There is only one significant subsample, namely the environmental pillar for the 90% confidence level. This means that there is no significant difference between the CAR for firms with either a high or low CSR score for all the other subsamples.

Table 8. Univariate analysis

This table shows some statistics for the subsamples of the second sample, as described in section 3.1. The mean is the mean of the CAR (-3,+3) for the corresponding subsample and N is the number of the observations in the subsample. The test statistics are used to analyze the differences in CAR between the subsamples.

Student t-test Wilcoxon t-test Subsample CAR

Mean

N t-statistic p-value t-statistic p-value ESG High ESG -1.76 53 -0.137 0.891 -0.314 0.753 Low ESG -1.93 62 Environmental High Env -0.61 44 -1.549 0.124 -1.675 0.094 Low Env -2.63 71 Social High Soc -2.34 50 _{-0.863 0.390 -0.547 0.584} Low Soc -1.22 65 Governance High gov -2.19 63 _{0.576 0.566 0.303 0.762} Low gov -1.45 52 5.2 Regression analysis

robustness tests, discussed later, the CAR’s are calculated with the Fama-French three-factor model to test the robustness of the regressions. Another robustness test applied in this study is the use of more event windows, amongst the (-3,+3) event window.

5.2.1 Regression first hypothesis

The correlation matrix (see appendix A) shows that all of the CSR components are positively related to the CAR, except the social component. The social component has a small negative relation with the CAR. To see the direct impact of the CSR score of acquirers on the CAR of acquirers around acquisitions, the regression stated in section 3.1 is performed. The outcome can be found in the Table 9, which can be found below. Table 5, with the analysis of the subsamples, indicated that there may be a negative (or at least weaker) relation between CSR score and CAR’s for the environmental and governance pillar, since those pillars show lower CAR’s in the subsample with high-scoring firms than in the subsample with low-scoring firms. The coefficients that can be found in Table 9 confirmed this negative relation for the governance pillar. This is not in line with the first hypothesis, which stated that there is a positive relationship between acquirer’s CSR score and CAR.

Table 9. Multiple regression: explaining acquirer CAR (-3,+3) with acquirer CSR score

This table represents the OLS regression results using the CAR based on the CAPM with the event window (-3,+3) as dependent variable for the whole sample. Each regression uses one pillar of the ASSET4 CSR scoring system. The top numbers show the coefficient of the variable and the bottom numbers show the t-statistic. As stated before, the regression uses heteroscedasticity-consistent standard errors. *, **, *** shows significance for respectively 10%, 5% and 1% levels.

Variables (1) (2) (3) (4) Independent variables ESG 0.011 0.472 Env 0.016 0.918 Soc 0.020 0.970 Gov -0.014 -0.709 Control variables FCF 4.001 0.700 4.160 0.733 3.972 0.701 4.051 0.710 Firm size -0.710 -2.904 *** -0.824 -2.632 *** -0.839 -2.887 *** -0.650 -2.724 *** Leverage -0.007 -0.029 0.010 0.040 0.017 0.071 -0.011 -0.045 Tobin’s q -0.793 -2.528 ** -0.823 -2.633 *** -0.818 -2.609 *** -0.795 -2.505 ** Dummy if stock-paid -0.624 -0.420 -0.559 -0.377 -0.574 -0.388 -0.757 -0.508 Dummy if both high-tech -1.268 -1.434 -1.294 -1.465 -1.272 -1.444 -1.248 -1.392 Deal size 2.638 1.413 2.652 1.443 2.653 1.439 2.844 1.506 Time fixed

year-effects Yes Yes Yes Yes

Industry-effects

Yes Yes Yes Yes

R2

0.162 0.163 0.163 0.162

N

When interpreting the control variables, the free cash flow shows a positive coefficient for all models. This is not in line with the free cash flow theory, which is confirmed by a recent study (Chepkwony, 2014). They also found a positive relation between stock returns and free cash flow. The positive coefficient makes sense from an economic point of view, since free cash flow reflects the money that is left from the company’s operating activities. With regards to the firm size, one could argue that the firm size will have a positive impact on stock returns: a big firm will have better stock returns. However, the regression output shows a significant negative effect on the CAR for all models, which is in line with Masulis et al. (2007). An explanation can be that large firms gather a lot of attention and, therefore, negative publicity is more widespread, which can lead to negative stock returns. On the other hand, following that thought would also apply to positive publicity, which refutes that theory. Moeller et al. (2004) did not find a clear explanation for the negative significant effect of firm size on CAR’s, but they did find that large firms pay a higher premium, which may lead to overpaying and therefore negative stock returns. Leverage shows a small negative coefficient for two pillars, indicating a small impact on stock returns around acquisitions. This is not in line with previous studies that find a positive (significant) impact (Deng et al., 2013; Masulis et al., 2007). For the environmental & social pillar regression however, Table 9 also shows positive coefficients. Masulis et al. (2007) stated that leverage has some power in preventing managers to make risky investments. Nevertheless, this study shows almost no impact, indicating that the influence of leverage is negligible. Tobin’s q has a significant negative coefficient for all models. Previous studies found different results when examining the effect of Tobin’s q (Deng et al., 2013; Lang et al., 1989; Masulis et al., 2007; Moeller et al., 2004), but Table 9 shows similar results as Masulis et al. (2007) and Moeller et al. (2004). An explanation of the negative coefficient can be that with an acquisition, cash flows out of the firm, resulting in a lower Tobin’s q and therefore resulting in a negative effect on the CAR.

reaction on the market. Moeller et al. (2004) confirmed this, however for a subsample of large acquirers they found opposite results. As one can see, the findings from previous studies about the relationship between relative deal size and acquirer stock returns are quite different. When following Moeller et al. (2004), the effect of deal size decreases with the size of the acquirer, meaning that there is a negative relationship. For the sample in this study however, Table 9 shows that the effect of relative deal size on stock returns is highly positive for all models. The regression output for firms with a high score on the CSR components can be found in Table C.1 in appendix C and Table C.2 in appendix C shows the regression output for firms with a low score on the CSR components, in line with the subsamples stated in section 3.1. The differences between these two are in line with what may be expected as a result of Table 5, except for the social pillar. For the other pillars the regression outputs show a positive relationship between CSR & CAR for low-scoring CSR firms and a negative relationship for high-scoring CSR firms, opposite of what one may expect based on the growing attention on CSR. The conclusion based on the subsamples is that firms with a high CSR score receive lower stock returns for every improvement on their CSR score, where for firms with a low CSR score the opposite is true.

When comparing the subsamples with the whole sample, the most eye-catching differences are the (sometimes) negative free cash flow coefficients for the high-scoring CSR firms and the positive coefficients on the high-tech dummy variable for high-scoring CSR firms. The signs of the free cash flow in the subsamples could indicate that for low-scoring CSR firms, shareholders are glad to see that a firm has a lot of cash after their activities, where for high-scoring CSR firms the opposite is true. A reasoning can be that shareholders of high-high-scoring CSR firms prefer firms to invest more in their activities instead of keeping a lot of their money. The significant negative sign of the high-tech dummy for low CSR-scoring firms can be explained by the fact that there are more high-tech acquisitions in the low CSR-scoring firms subsample than in the other subsample and, therefore, this is in line with previous studies, as mentioned earlier (Deng et al., 2013; Masulis et al., 2007; Westerbeek, 2018). The signs of the other variables are mostly in line with the whole sample, except for the stock-paid dummy for high-scoring firms. When examining this, it is striking that when the sign of the dummy variable is positive, the free cash flow sign is negative (as stated above), and there may be a relationship between these two.

5.2.2 Regression second hypothesis

correlation matrix of the first sample is that all the CSR components have an opposite sign in this correlation matrix, indicating a positive relationship between CAR and the ESG pillars. Furthermore, there is a relatively high correlation between the CSR components. An explanation for this is already given in section 5.2.1, because the same situation occurred in the correlation matrix of the first sample.

The signs of the coefficients of the CSR components are exactly the same as in the regression models of first hypothesis. This indicates that besides the fact that acquirers’ CSR4 _{has a}

positive effect on stock returns, also targets’ CSR has a positive effect on stock returns. However, as in the first sample, there is also no significant pillar across the CSR components for this sample. The control variables are showing different signs. The FCF for example has a high negative coefficient, indicating that FCF of the acquirer has a negative relationship with the CAR of the acquirer. This is in line with the free cash flow theory, but previous studies found different results of the sign of the FCF coefficient (Chepkwony, 2014; Deng et al., 2013; Oroud, Islam, & Salha Tunku Ahmad, 2017; Westerbeek, 2018). Compared to the mentioned studies, the coefficient of the FCF is high, indicating a strong negative effect on the CAR. The explanation behind this is, based on the free cash flow theory, that firms with a high cash flow have more resources to involve in empire building (Jensen, 1986). Because the FCF is close to zero, since it is scaled by the book value of assets, the high coefficient is explainable. When the FCF variable increases with 1.0, the CAR decreases with approximately 24. However, FCF has to increase a lot (or assets has to decrease) to let the variable increase by 1. The positive effect of firm size can be a result of the fact that in general large firms have more money, so more power to buy large and well-performing firms, resulting in positive stock returns. On the other hand, when a firm has more money to perform acquisitions, it is possible that managers are more inclined to pay more and do acquisitions that turn out to be negative. The negative coefficient of leverage is not in line with Masulis et al. (2012), but a possible explanation can be that when a firm has large debts and is also buying another firm, shareholders are losing their faith and start selling shares.

The sign of the coefficient of Tobin’s q is not in line with the first sample, but again significant. As stated before, previous studies found mixed results of the relationship between Tobin’s q and stock returns. The stock-paid dummy shows a positive coefficient, which is not in line with previous studies (Fuller, Netter, & Stegemoller, 2002; Masulis et al., 2007). The high-tech dummy also shows a positive coefficient for all models, opposite of what may be expected following the theory stated in section 5.2.1. However, both dummies are not significant in all of the models, so the direct influence on stock returns is unclear. The last variable, deal size, has a negative coefficient for all models and is significant for the 90% confidence level. The

4 _{Besides the governance pillar, but the all-encompassing ESG measures all three components and is positive}

negative coefficient is not what may be expected based on previous studies (Masulis et al., 2007; Westerbeek, 2018), but Moeller et al. (2004) and Deng et al. (2013) also found significant negative effects of deal size on CAR. Moeller et al. (2004) found that relative deal size has a negative effect for large firms, since an increase in bidder size reduces the deal size variable (because deal value is divided by the book value of assets).

Table 10. Multiple regression: explaining acquirer CAR (-3,+3) with target CSR score

This table represents the OLS regression results using the CAR (of the acquirer) based on the CAPM with the event window (-3,+3) as dependent variable for the whole sample. Each regression uses one pillar of the ASSET4 CSR scoring system for the target firm. The top numbers show the coefficient of the variable and the bottom numbers show the t-statistic. As stated before, the regression uses heteroscedasticity-consistent standard errors. *, **, *** shows significance for respectively 10%, 5% and 1% levels.

Variables (1) (2) (3) (4) Independent variables ESG 0.022 0.320 Env 0.070 1.293 Soc 0.021 0.372 Gov -0.018 -0.436 Control variables FCF -24.508 -1.250 -24.594 -1.251 -23.889 -1.229 -24.620 -1.234 Firm size 0.958 1.067 0.552 0.511 0.929 0.985 1.061 1.179 Leverage -0.765 -0.327 -0.531 -0.243 -0.808 -0.354 -0.822 -0.361 Tobin’s q 0.876 2.721 *** 0.878 2.829 *** 0.867 2.602 ** 0.881 2.629 ** Dummy if stock-paid 2.004 0.552 1.280 0.319 2.149 0.570 2.129 0.558 Dummy if both high-tech 0.122 0.037 0.449 0.135 0.074 0.023 0.006 0.002 Deal size -0.265 -1.879 * -0.277 -2.180 ** -0.250 -1.940 * -0.230 -1.732 * Time fixed

year-effects Yes Yes Yes Yes

Industry-effects Yes Yes Yes Yes

R2 _{0.478 0.494 0.478 0.478}

Similar to the first sample, the regression results for the subsamples are reported in the appendices. In Table C.3 of appendix C the regression output for firms with a high CSR score can be found, whereas in Table C.4 of appendix C the regression output for the subsample of firms with a low CSR score can be found. When comparing those two tables, one can see that for all pillars the subsample with high-scoring firms shows a positive relationship between CAR and CSR, whereas the subsample with low-scoring firms shows a negative relationship between CAR and CSR, except for the governance pillar. This may indicate that shareholders of high-scoring firms attach more value to CSR than those of low-high-scoring firms and that a high CSR score of the target is value-increasing for acquirers. When looking at the control variables, the FCF is the most eye-catching. Again, as for the whole sample, the FCF coefficient is high (negative) for some models.

5.3 Robustness tests

To check if the results shown in the previous section are robust, several tests have been done. The first robustness test is the use of different event windows. Differences can show new insights if and when the relationship between CSR performance and CAR is the strongest. Another robustness test that has been performed has to do with the method of calculating the CAR’s. The initial method used is the CAPM method. For validation, this study also used the Fama-French three-factor model as robustness test to calculate the CAR’s.

For the second sample the same event windows are tested in order to see whether or not there are differences between the event windows. Those results can be found in Tables D.4, D.5 and D.6 of appendix D. The results of the (-1,+1) window show the same signs of the CSR components as the (-3,+3) window, but in the (-1,+1) window the environmental pillar is significant for the 10% significance level. The same applies to the (-1,+3) & (-5,+5) windows, which can be found in Table D.5 and Table D.6 respectively, except for the fact that the environmental pillar is not significant in the (-5,+5) window. Based on these results, one can conclude that the second sample has robust results for different event windows. The different event windows also show that the magnitude of the CSR coefficients decreases as the event window increases, in line with the results of the first sample for some ESG pillars.

The regression outputs using the Fama-French three-factor model, instead of the CAPM, can be found in appendix E. For both samples the Fama-French three-factor model outputs are almost the same as the outputs based on the CAPM. The signs of the CSR scores are the same and the magnitudes of the coefficients are also close. In addition, the same variables are significant for the Fama-French three-factor model output as for the CAPM. Therefore, it can be concluded that the results are robust for using a different model for calculating the CAR instead of the CAPM.

6. Conclusion

The aim of this study is to investigate the relationship between CSR and stock returns around M&A’s for US firms. In order to do this, this study used the ESG pillars to measure CSR performance and cumulative abnormal returns based on the CAPM as the dependent variable. First, the study focused on acquirer CSR score and acquirer CAR. Secondly, the target CSR score and acquirer CAR were studied. Data for the acquisitions was gathered through the Zephyr database, CSR scores from the ASSET4 database of Thomson Reuters and the data for stock returns with Thomson Reuters’ Eikon.

The first hypothesis stated that the acquirer CSR score has a positive impact on the acquirer CAR. Because not all pillars have a positive relationship and since none of the pillars is significant, the first hypothesis is rejected. Another reason for rejecting the first hypothesis is the fact that the coefficients are indicating a small relationship. The positive relationship of the total CSR score is in line with Deng et al. (2013) and Gomes & Marsat (2017). When comparing their results with this study (their studies did not focus on all ESG pillars, only on total CSR score), these studies show very large coefficients and both studies found significant results. However, Westerbeek (2018) found a negative relationship in his study with a European sample for all pillars, except for the environmental pillar, and none of the pillars were significant. This would indicate that shareholders of American firms are valuing CSR higher than shareholders of European firms.

The second hypothesis stated that target CSR score has a positive impact on the acquirer CAR. As with the results of the first sample, none of the independent variables is significant. There has not been done a lot of research into the relationship between target CSR and acquirer CAR around M&A’s, but Smeets (2017) did not find a positive effect of target CSR on acquirer CAR for a European sample. That would indicate that in America CSR matters more to shareholders than in European countries, since this study did find positive effects of CSR on CAR. Another reason may be that in Europe CSR is somehow value-reducing for shareholders. However, none of the models show significant CSR scores.

In conclusion, there is still ambiguity around the topic of CSR and the relation with stock returns. Previous studies found different results, but this might be due to the country or region the sample is based on. Since this study did not find significant results, except for the environmental pillar of the second sample when other event windows were used, it is difficult to draw conclusions on whether or not there is a relationship between CSR and CAR. One conclusion that can be drawn is that the target CSR has a (slightly) greater impact on acquirer CAR than acquirer CSR on acquirer CAR.

7. Discussion

In order to draw more reliable conclusions, for further research the sample size needs to be increased. There are different ways of doing so, for example by dropping some of the requirements that M&A’s had to meet, which are stated in section 4.2. Another way to increase the sample size is to look at more or other regions. However, a drawback when doing so is that it is hard to draw conclusions for one country or region and we already saw some differences between Europe and the US.

Another suggestion for further research is to study whether or not there are differences between the past few years and the past, for example between 1990-2000. Back then, there was less attention for CSR and so it would be interesting to see if the relationship between CSR and CAR was also different. As said before, the differences between different CSR databases are also a topic that can be investigated further. In addition, it could be interesting to study the long-term relationship between CSR and CAR. This study only looked at a few days before and after the acquisition announcement, with a maximum of five days before and five days after. It is possible that the effects of CSR will become clear in the stock returns later after the announcement. Hence, it is interesting to study the long-term effects of CSR on CAR around M&A’s.

8. References

Ackerman, M. (2016). The impact of Corporate Social Responsibility and mergers and acquisitions on a multinational corporation. A case study in the elevator industry.

Adenugba, A. A., Ige, A. A., & Kesinro, O. R. (2016). Financial leverage and firm’s value: A study of selected firms in Nigeria. European Journal of Research and Reflection in

Management Sciences, 4(1), 14–32.

Angelidis, J., & Ibrahim, N. (2004). An Explanatory Study of the Impact of Degree of Religiousness upon an Individual’s Corporate Social Responsiveness Orientation. Journal of

Business Ethics, 51(2), 119–128.

Banerjee, S., Homroy, S., & Slechten, A. (2015). Why do some firms invest more in CSR? The Effect of Corporate Holding Stucture on CSR.

Bendell, J. (2010). ESG analysis in deep water. Journal of Corporate Citizenship, 40, 6–25. Bouten, L., Cho, C. H., Michelon, G., & Roberts, R. W. (2014). Does the choice of CSR performance rating provider matter? (p. 55). Presented at the International Congress on Social and Environmental Accounting Research, University of St. Andrews.

Brammer, S., Brooks, C., & Pavelin, S. (2016). Corporate social performance and stock returns: UK evidence from disaggregate measures. Financial Management, 35(3), 97–116. Chatterji, A., Durand, R., Levine, D., & Touboul, S. (2016). Do Ratings of Firms Converge? Implications for Managers, Investors and Strategy Researchers. Strategic Management

Journal, 37(8), 1597–1614.

Chatterji, A. K., Levine, D. I., & Toffel, M. W. (2009). How Well Do Social Ratings Actually Measure Corporate Social Responsibility? Journal of Economics & Management Strategy,

18(1), 125–169.

Chen, R. C. Y., & Lee, C.-H. (2017). The influence of CSR on firm value: an application of panel smooth transition regression on Taiwan. Applied Economics, 49(34), 3422–3434. Chepkwony, C. C. (2014). The relationship between free cash flows and stock returns of firms

listed at the Nairobi Securities Exchange. University of Nairobi.

Deloitte. (2017). The state of the deal.

Deng, X., Kang, J., & Low, B. S. (2013). Corporate Social Responsibility and Stakeholder Value Maximization : Evidence from Mergers. Journal of Financial Economics, 110(1), 87– 109.

Dornean, A., & Oanea, D. C. (2017). CSR and Company’s Stock Price. A Comparative Evidence from Bucharest Stock Exchange. EuroEconomica, 36(1), 175–184.

European Commission. (2011). A renewed EU strategy 2011-14 for Corporate Social

Forte, A. (2013). Corporate Social Responsibility In The United States And Europe: How Important Is It? The Future Of Corporate Social Responsibility. International Business &

Economics Research Journal, 12(7), 815–824.

French, K. R. (n.d.). Data library. Retrieved October 13, 2018, from

http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html#BookEquity Fuller, K., Netter, J., & Stegemoller, M. (2002). What Do Returns to Acquiring Firms Tell Us? Evidence from Firms That Make Many Acquisitions. Journal of Finance, 57(4), 1763– 1793.

Gomes, M., & Marsat, S. (2017a). Does CSR impact premiums in M&A transactions?

Finance Research Letters, (November), 1–10.

Graafland, J. J., Eijffinger, S. C. W., & Smid, J. H. (2004). Benchmarking of Corporate Social Responsibility: Methodological Problems and Robustness. Journal of Business Ethics,

53(1/2), 137–152.

Huang, T.-M., & Zong, S. (2016). Stock performance of socially responsible companies.

Nang Yan Business Journal, 5(1), 1–12.

IMAA. (2018). M&A Statistics. Retrieved from https://imaa-institute.org/mergers-and-acquisitions-statistics/

Jensen, M. C. (1986). Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers.

American Economic Review, 76(2), 323–329.

Kaya, I. (2016). The Mandatory Social and Environmental Reporting: Evidence from France.

Procedia - Social and Behavioral Sciences, 229, 206–213.

Lagas, F. (2013). The effect of corporate social performance on shareholder wealth in

Mergers & Acquisitions. Tilburg university.

Lang, L. H. P., & Litzenberger, R. H. (1989). Dividend announcements: Cash flow signalling vs. free cash flow hypothesis? Journal of Financial Economics, 24(1), 181–191.

Lang, L. H. P., Stulz, R. M., & Walkling, R. A. (1989). Managerial performance, Tobin’s q, and the gains from succesful tender offers. Journal of Financial Economics, 24(1), 137–154. Loughran, T., & Ritter, J. (2004). Why Has IPO Underpricing Changed Over Time?

Financial Management, 33(3), 5–37.

MacKinlay, A. C. (1997). Event studies in Economics and Finance. Journal of Economic

Literature, XXXV(March), 13–39.

Maignan, I., & Ralston, D. A. (2002). Corporate Social Responsibility in Europe and the U.S.: Insights from Businesses’ Self-Presentations. Journal of International Business Studies,

Masulis, R. W., Wang, C., & Xie, F. (2007). Corporate Governance and Acquirer Returns.

Journal of Finance, 62(4), 1851–1889.

Mio, C., & Venturelli, A. (2013). Non-financial Information About Sustainable Development and Environmental Policy in the Annual Reports of Listed Companies: Evidence from Italy and the UK. Corporate Social Responsibility and Environmental Management, 20(6), 340– 358.

Moeller, S. B., Schlingemann, F. P., & Stulz, R. M. (2004). Firm size and the gains from acquisitions. Journal of Financial Economics, 73(2), 201–228.

Mwamburi, S. P. (2013). The effect of corporate social responsibility announcement on stock

returns at the Nairobi securities exchange. University of Nairobi, Nairobi.

O’Brien, R. M. (2007). A caution regarding rules of thumb for variance inflation factors.

Quality & Quantity, 41(5), 673–690.

Oroud, Y., Islam, A., & Salha Tunku Ahmad, T. (2017). The Effect of Cash Flows on the Share Price on Amman Stock Exchange. American Based Research Journal, 6(8), 22–28. Österman, C. (2014). Why companies engage in CSR. Lund University, Lund.

PwC, & PRI. (2012). The Integration of Environmental, Social and Governance Issues in

Mergers and Acquisitions Transactions (p. 27). Retrieved from

https://www.pwc.com/gx/en/sustainability/publications/assets/pwc-the-integration-of-environmental-social-and-governance-issues-in-mergers-and-acquisitions-transactions.pdf Reuters, T. (2018). ESG Research. Retrieved August 10, 2018, from

https://financial.thomsonreuters.com/en/products/data-analytics/company-data/esg-research-data.html

Sawhny, V. (2008). Analyzing Corporate Social Responsibility Measurement Parameters. Institute for Public Relations. Retrieved from www.instituteforpr.org

Servaes, H., & Tamayo, A. (2013). The Impact of Corporate Social Responsibility on Firm Value: The Role of Customer Awareness. Management Science, 59(5), 1045–1061. Sitthipongpanich, T. (2011). Understanding the Event Study. Journal of Business

Administration, 59–68.

Statista. (2018). Value of mergers and acquisitions (M&A) worldwide from 2013 to 2018. Retrieved November 3, 2018, from https://www.statista.com/statistics/267369/volume-of-mergers-and-acquisitions-worldwide/

Tennant, F. (2015). The importance of corporate social responsibility. Financier Worldwide. Retrieved from https://www.financierworldwide.com/the-importance-of-corporate-social-responsibility/#.XAumfC1x90t

van den Heuvel, R. (2012). How Robust are CSR Benchmarks? Comparing ASSET4 with

Sustainalytics.

Von Arx, U., & Ziegler, A. (2013). The effect of corporate social responsibility on stock performance: new evidence for the USA and Europe. Quantitative Finance, 14(6), 977–991. Westerbeek, R. (2018). Corporate Social Responsibility and Value Creation : Evidence from

Mergers & Acquisitions. University of Groningen.

Zhang, Y. (2017). The Impact of Corporate Social Responsibility on Stock Returns: Evidence

Appendices

Table A.1. Correlation matrix

This table shows the Pearson correlation matrix of the first sample, consisting of 465 observations. Values of 6.000 or above are marked bold.

Variables CAR

(-3,+3)

ESG Env Soc Gov FCF Firm

size Lever age Tobin ’s q Stock -paid High-tech Deal size CAR (-3,+3) 1.000 ESG 0.004 1.000 Env -0.043 0.609 1.000 Soc -0.019 0.610 0.733 1.000 Gov -0.040 0.568 0.332 0.351 1.000 FCF 0.034 0.025 0.034 0.098 0.037 1.000 Firmsize -0.105 0.151 0.483 0.520 0.207 0.066 1.000 Leverage -0.013 0.005 -0.014 -0.032 0.004 -0.219 0.268 1.000 Tobin’s q -0.058 -0.003 -0.025 -0.040 -0.040 0.161 -0.342 -0.340 1.000 Stock-paid dummy -0.035 -0.080 -0.126 -0.134 -0.082 -0.221 -0.044 0.180 -0.052 1.000 High-tech dummy -0.088 0.033 0.024 0.017 0.015 0.109 -0.126 -0.208 0.329 -0.010 1.000 Deal size 0.077 0.002 -0.087 -0.097 0.028 -0.006 -0.250 -0.089 0.445 0.171 0.000 1.000

Table A.3. Correlation matrix

This table shows the Pearson correlation matrix of the second sample, consisting of 115 observations. Values of 6.000 or above are marked bold.

Variables CAR

(-3,+3)

ESG Env Soc Gov FCF Firm

size Lever age Tobi n’s q Stock -paid High-tech Deal size CAR (-3,+3) 1.000 ESG -0.020 1.000 Env 0.176 0.569 1.000 Soc 0.058 0.584 0.601 1.000 Gov 0.012 0.666 0.358 0.391 1.000 FCF 0.086 -0.071 0.016 0.058 -0.053 1.000 Firmsize 0.049 0.145 0.284 0.240 0.100 0.001 1.000 Leverage -0.239 -0.039 -0.008 -0.031 0.027 -0.164 0.106 1.000 Tobin’s q 0.182 0.018 -0.027 -0.007 0.069 0.250 -0.287 -0.296 1.000 Stock-paid dummy -0.049 0.058 -0.110 -0.080 0.008 -0.183 -0.316 -0.124 0.137 1.000 High-tech dummy 0.137 -0.011 0.078 0.099 -0.008 0.231 0.265 -0.131 0.015 -0.228 1.000 Deal size -0.052 0.162 0.077 0.032 0.137 0.000 -0.163 -0.070 0.137 0.012 -0.066 1.000

Appendix B: Jarque-Bera tests

Table B.1. Jarque-Bera first sample

Model N p-value JB

ESG 465 0.000

Environmental 465 0.000 Social 465 0.000 Governance 465 0.000

Table B.2. Jarque-Bera second sample

Model N p-value JB

ESG 115 0.008