Does the target’s CSR performance influence target and acquirer gains? Evidence from mergers and acquisitions

(1)

Thesis MSc Finance

Does the target’s CSR performance influence target

and acquirer gains? Evidence from mergers and

acquisitions

ABSTRACT

This paper uses the theory of mergers and acquisitions (M&As) to shed a new light on whether CSR performance enhances firm value. I investigate whether the target firm’s CSR performance influences the target and acquirer wealth gains from the M&A announcement using a sample of deals with public, European targets for the period 2004 to 2016. Target gains are measured by the acquisition premium and the target abnormal returns. Acquirer gains are measured by the acquirer abnormal returns. Using the ESG rating from ASSET4 as a measure of CSR performance, this paper finds that the target’s CSR performance has no influence on the acquisition premium. However, the environmental and social score do positively influence the target gains through the target abnormal returns. The target’s CSR performance has no influence on the acquirer abnormal returns, suggesting that the acquirer is not rewarded for investing in high CSR firms.

Author: Bouke Smeets

Student number: s2369648

Supervisor: N. Heida

Date: June 6th 2017

Word count: 14.502

Keywords: acquisition premium, abnormal returns, CSR, M&A

(2)

1. Introduction

During the last two decades, the subject of corporate social responsibility (CSR) has attracted the attention of the business press, resulting in a significant amount of diversified literature surrounding the issue (Malik, 2014). The objective of this study is to measure whether CSR activities create value in the context of mergers and acquisitions (M&As). Specifically, I examine whether target firms’ CSR performance influences the acquisition premium paid by the acquirers for the target company and whether it influences the target and acquirer short-term abnormal returns.

Studies on the relationship between CSR performance and financial performance find varying results (Marin et al., 2012). Dam and Scholtens (2015) and Klassen and McLaughlin (1996) both find a positive relationship between CSR ratings and financial performance. However, Dam and Scholtens (2015) also find that the relationship is conditional on which financial performance measure is considered. In contrast, Revelli and Vivianni (2015) find no real effect of CSR on financial performance and state that the level of performance depends on the methodological choices made by researchers. According to Gregory et al. (2014), CSR has a positive effect on short- and long-term growth rates in firm’s abnormal earnings. Moreover, according to Malik (2014) CSR enhances financial performance through multiple other channels such as capital market benefits, lowering of risks, increased operating efficiency, product market gains, enhanced employee productivity and improved earnings quality. In view of these contradictory results, the question whether CSR creates value for shareholders remains largely open.

This question is especially interesting since the importance of CSR investment to firm’s operations is growing. Firms invest in CSR as a part of their strategy or as a result of pressure from activist shareholders. Moreover, an increasing amount of firms publish annual CSR reports, providing information about their CSR activities and achievements (Deng et al, 2013). This implies the need for firms to become more transparent about their responsibilities.

(5)

transaction approval process is highly influenced by the target’s and acquirer’s shareholders, M&As serve as important events to examine the impact of CSR on shareholders gains. This study examines the short-term rather than the long-term target and acquirer wealth gains as these are more likely to solely capture the effect of the M&A announcement.

By investigating both the gains to the target and the acquirer this study provides a complete overview of the total gains from the targets CSR performance in the context of M&A. Also, examining the target and acquirer gains is interesting since the effects of a takeover on the target firm and the acquiring firm often differ (Campa and Hernando, 2004). I measure the gains to the target’s shareholders by the acquisition premium and the target abnormal returns.

Since the deal value is not always known on the announcement date1, both indicators could

have different effects. Moreover, the target abnormal returns can be seen as an explorative addition to the premium. If CSR is value enhancing, the CSR performance of the target should positively influence the target gains through expected synergies from the deal. Furthermore, acquirer abnormal returns are used to measure the gains to the shareholders of the acquirer. Acquiring a high CSR performing target could be a signal from the acquirer about the willingness to learn from the CSR activities of the target firm, thus adding value to the acquirer shareholders and increasing the abnormal returns to the acquirer (Aktas et al., 2011). The measure of the company’s CSR performance is the ESG rating, obtained from the Thomson Reuters ASSET4 database. The ESG rating is a reflection of a firm’s performance based on its environmental, corporate governance and social performance.

Using a sample of 133 completed M&A deals with public, European targets for the period 2004 to 2016, I measure the effect of the combined ESG rating, the environmental, the corporate governance and the social rating on target and acquirer gains. I find that none of these target’s CSR performance measures influences the target gains through the acquisition premium. However, the environmental and social score do influence the target gains through the target abnormal returns from the acquisition announcement, whereas the combined ESG score and the corporate governance score have no influence. The target’s CSR performance has no influence on the acquirer abnormal returns, suggesting that the acquirer is not rewarded for investing in CSR.

This study contributes to the current literature in multiple ways. Firstly, previous literature discusses the effect of CSR on M&A gains by using a sample of U.S. or worldwide deals. As

(6)

Duuren et al. (2016) find that there is a substantial difference in the ways in which U.S. and European managers view ESG, this paper adds new insight by focusing on a sample of European target deals. Secondly, I discuss the effect of the combined CSR rating as well as three separate CSR ratings on the target and acquirer gains, whereas previous research often discusses only one rating (i.e. often only the combined CSR rating). Furthermore, to my knowledge, this is the first research concerning CSR that extensively examines both acquisition premium and the target and acquirer abnormal returns.

(7)

2. CSR in the context of M&A

This section provides the theoretical background of this study. Firstly, I discuss the M&A motives and gains in more detail. I continue by elaborating how CSR performance relates to M&A transactions. Finally, the hypotheses of this study are presented.

2.1 Mergers and Acquisitions

2.1.1 M&A motives

In the literature, it is argued that takeovers are driven by a complex pattern of motives (Ravenscraft and Scherer, 1987). Among all theories explaining firms’ M&A motives, value maximization is shown to be the most dominant theory (Lubatkin, 1987; Bradley et al., 1988; Berkovitch and Narayanan, 1993). Research that focuses on value maximization identifies various sources of synergies that firms gain through M&As. The synergy motive assumes that acquirers aim to maximize value and only engage in M&A activities if it results in an overall financial gain (Berkovitch and Narayanan, 1993). The synergy motive entails that M&A increases value through the development of economies of scale- or scope, cost reduction, or by an increase in market (bargaining-) power (Jensen and Ruback, 1983). The average synergistic gain found by Bradley et al. (1988), using a sample of 236 tender offers in the U.S. from 1963 to 1984, represents a 7.4% increase in the combined wealth of the shareholders of the target and acquiring firms, implying takeovers are value-increasing.

Although synergy is the main reason for firms’ M&A decisions, takeovers can be explained by other motives as well. Literature suggests that certain M&A transactions can be motivated by the self-interest of the acquirer’s management. This agency motive potentially destroys value because the acquiring firm’s management identify target firms that are most suited to increase its own welfare at the expense of its shareholders. Another major reason for takeovers is managerial hubris, where acquisitions are motivated by managers’ mistakes and synergistic gains are highly overestimated (Berkovitch and Narayanan, 1993).

2.1.2 Target vs. acquirer gains

(8)

and 41.5% respectively. This implies that the target firm gains from the acquisition through the acquisition premium.

Campa and Hernando (2004) summarise the findings of 13 studies concerning abnormal returns to target firms shareholders and reveal the target abnormal returns are on average significantly positive, despite variations in the observation period, the measure of returns, the industry and the time period. By examining a sample of 262 European deals over a small time window from 1998 to 2004, Campa and Hernando (2004) find that target firm shareholders receive on average a statistically significant cumulative abnormal return of 3.93% centred on the announcement date for a three-day event window. Jensen and Ruback (1983) review much of the scientific literature on the market for corporate control and also find, based on 13 studies, that target firms shareholders benefit from M&A announcements. I conclude target returns are on average significantly positive.

Remarkably, the evidence on returns to acquirer firms’ shareholders is less conclusive. Based on previous literature (i.e. 25 studies), Campa and Hernando (2004) argue that the evidence on abnormal returns to acquirer shareholders is evenly distributed between studies that report negative abnormal returns and those that report zero and slightly positive abnormal returns. Based on their own sample of European deals, Campa and Hernando (2004) find no significant returns to the acquirer. In line, Jensen and Ruback (1983) suggest that abnormal returns to successful bidding firms in M&As are zero on average and abnormal returns to unsuccessful bidders are generally negative.

2.1.3 The effect of target, acquirer and deal characteristics on M&A gains

Prior literature identifies several economic factors that could influence those target and acquirer gains from an M&A transaction (Malik 2014). I define those factors in three groups: target, acquirer and deal characteristics.

Target characteristics

(9)

The target firms profitability also influences M&A gains. Profitable targets are attractive, and so the acquirers are willing to pay more for them (Shawver, 2002). However, the managerial hubris hypothesis indicates that inefficiently managed targets are expected to earn higher returns after a merger (Palepu 1986).

Furthermore, higher growth targets have more capacity to accomplish larger synergistic benefits. Palepu (1986) argues that firms whose market values are low compared to their book values are more likely to be acquisition targets. To capture the target’s growth opportunities, the acquirer may pay a higher premium for a target with a high growth rate (Lamaanen, 2007).

Another target-specific characteristic that affects M&As is the target’s leverage ratio. Prior research finds that value-maximizing bidders are willing to pay more for high-leveraged targets because management, who values control, uses leverage to increase its ownership and its power in the target firms during takeover contests (Raad, 2012). Besides that, Yen and Andree (2016) find that the abnormal returns of the target are negatively correlated with the extent of leverage.

Acquirer characteristics

The size of the acquirer influences M&A gains in several ways as well. Large acquirers are more likely to make overpayments since the absolute value paid constitutes a smaller portion of their value, hence have a smaller impact on their share price (Simonyan, 2014). Also, Alexandridis et al. (2013) find that empire building and managerial hubris are more common for larger firms, as a result larger acquirers can pay a higher premium. On the other hand, larger firms have greater bargaining power, which reduces the paid premium (Niinivaara, 2010). In line with Alexandridis et al. (2013), Moeller et al. (2004) also find that managerial hubris plays a larger role in the takeover decisions of large firms. They imply that, as a result, large firms enter acquisitions with negative synergy gains which leads to lower acquirer announcement returns.

(10)

the growth of the bidder on the acquisition premium. As high growth acquirers have better capabilities to accomplish larger synergistic benefits, acquirer growth has a positive effect on the acquirer’s abnormal returns (Powel, 2004).

Furthermore, Harford et al. (2005) find a significant positive relationship between acquirer leverage and acquirer abnormal returns following a M&A transaction, suggesting leverage improves managerial decision-making.

Deal characteristics

In addition to the target and acquirer specific characteristics, several deal specific characteristics, such as takeover hostility, influence the target and acquirer gains from an M&A transaction as well. According to Schwert (2000), when a takeover is hostile, the acquirer has to pay more to make the target’s shareholders accept the offer. Furthermore, Schwert (2000) finds evidence for reductions in the bidders stock price due to hostile takeovers, resulting in lower acquirer returns. However, Schwert (2000) finds no effect of the takeover hostility on the target abnormal returns.

The industry relatedness of the target and acquirer is well-documented in M&A literature. Industry related M&As are able to increase shareholder welfare through the exploitation of strategic synergies (Doukas et al., 2002). Moreover, Morck et al. (1990) find that diversifying M&As are value destroying. When the target and the acquirer share a related industry, the acquirer may not need to invest additional effort to learn about the target industry (Reuer and Koza, 2000), which increases the acquisition premium.

Furthermore, the method of payment has an effect on M&A gains. Myers and Majluf (1984) argue that acquirers financing a M&A transaction with equity signal that the market overvalues its assets, which they call the ‘signaling hypothesis’. Also, a positive association between cash-payment and premium is expected due to tax-implications (Hansen, 1987).

(11)

Lastly, it is documented that bidding competition significantly affects M&A gains. A target’s bargaining power increases in the presence of competing bidders (Bradley et al., 1988), which positively affects the acquisition premium. In line with this, Moeller et al. (2004) argue that competition for a target increases the return to the target and decreases the return to the acquirer.

2.2 The link between M&A and CSR

2.2.1 Acquiring firm’s CSR and M&A value creation

Although M&A transactions have widely been examined and literature on CSR is emerging, only a few studies attempt to bridge the research literature that addresses these important business trends. Deng et al. (2013) study a large sample of 1556 U.S. deals from 1992 to 2007 and find that high CSR rated acquirers, where CSR is measured by the KLD rating2, realize higher announcement returns and higher long-term stock returns during the post-merger period. They also find that deals with high CSR acquirers take less time to finish and are more likely to be completed. This is in line with Rani et al. (2014), who examine 155 worldwide deals during 2003-2008 and reveal a positive relationship between the corporate governance score, which explains a substantial part of the total CSR rating, and the performance of the acquiring firms in the short-run as well as the long run. On the other hand, Meckl and Theuerkorn (2015) investigate whether CSR engagement of the acquiring company enhances the success rate, measured by long- and short-run abnormal returns, using 113 deals from mainly Germany and the U.S. They demonstrate that there is no significant effect, suggesting that taking social responsibility is not recognized on the capital market. Hence, literature regarding the effect of the acquirer firm’s CSR performance on a M&A transaction produces mixed results.

2.2.2 Target firm’s CSR and M&A value creation

Malik (2014) examines the correlation between the targets’ CSR quality, measured by the

KLD rating2, and the acquisition premium on a sample of 419 U.S. M&A deals. This study

finds a positive relationship between the targets’ CSR quality and the acquisition premium. In addition, Berchicci et al. (2012) find that the target firms’ degree of environmental capabilities affects the acquisition choice of manufacturing firms. Their research is conducted

(12)

on a U.S. sample of 2.485 deals between 1991 and 2005 and shows that the effect is significantly positive only when the acquiring firm has inferior environmental capabilities and the target firm is physically proximate. Since environmental governance explains a substantial part of CSR ratings, this could imply that CSR performance influences the acquisition choice as well. Furthermore, Choi et al. (2012), examine a sample of 215 cash-only acquisitions announced by U.S. firms for 1995-2013 in the context of asymmetric information and CSR. They find support for the theory that acquirers rely on signals associated with CSR, measured by the KLD rating3, in order to deliver messages about the overall quality of a target, which, in turn, affects the acquisition premium. Acquirers dealing with high information asymmetry rely more on externally available information, such as CSR ratings, thus the effect of CSR ratings on the acquisition premium becomes more prominent.

Berchicci et al. (2012), Choi et al. (2012) and Malik (2014) all study a sample of U.S. deals. Aktas et al. (2011) and Harper (2012) conduct research on a sample of worldwide deals and find different results.

Aktas et al. (2011) examine the effect of the target’s CSR performance on target and acquirer abnormal returns using a worldwide sample of 106 deals over a ten-year period from 1997 to 2007. In their research, the Intangible Value Assessment (IVA) is used as a proxy for CSR performance4. Aktas et al. (2011) argue that acquiring a high CSR firm can be value enhancing when the CSR score is seen as a signal that the acquiring firm is willing to learn from the target how to enhance or maintain social responsibility. They find that the target firms’ ability to cope with social and environmental risks has a positive correlation with acquirer announcement abnormal returns. However, they find that the abnormal returns of the target are not dependent on the target’s own CSR rating. Contrary to Aktas et al. (2011), Harper (2012) finds no significant effect from the acquisition of CSR performance on the acquirer gains (i.e. stock returns) around the announcement date. Therefore, Harper (2012) concludes investors in the market do not recognize CSR investments. However, his research on 17.541 worldwide firms over a period from 1993 and 2009, also shows a strong positive relationship between the KLD rating of the target firm, as a measure of CSR engagement, and the likelihood of being a M&A target in the market, possibly due to the value-enhancing capabilities of CSR.

3_{See footnote 2.}

(13)

Overall, the findings of prior literature are not unanimous. In the U.S., there is mostly a positive relationship between the target’s CSR performance and M&A gains (Berchicci et al., 2012; Choi et al., 2012; Malik, 2014). Worldwide, most studies find a positive or insignificant relationship (Aktas et al., 2011; Harper, 2012).

2.3. Hypotheses development

Since the primary motive for a merger or acquisition is to maximize value5, a firm with potential value-enhancing capabilities has a higher likelihood of becoming a M&A target (Harper, 2012; Malik, 2014). The acquirer sees the acquisition as an opportunity to maximize value through gaining the capabilities of the target firm (Berchicci et al., 2012). If CSR is a value-enhancing capability, intuitively, firms with a high CSR rating are more attractive to acquirers. Accordingly, Malik (2014) finds a positive relationship between the target’s CSR rating and the acquisition premium. Laamanen (2007) finds that acquirers of 458 U.S. technology-based firms pay for various value-creating resource combinations that can be obtained by combining the target and acquirer firm’s resources. In this study, R&D investments are identified as potentially synergistic resources. Moreover, Laamanen (2007) finds that because acquirers of those technology-based firms are able to detect signals about a targets future prospects through its R&D performance, those targets receive higher premiums. CSR can be valued as a similar synergistic resource. Therefore, I expect that firms with a higher CSR rating receive a higher acquisition premium compared to firms with a low CSR rating. This results in the following hypothesis:

H1: Target companies with higher CSR ratings receive higher acquisition premiums. If the targets CSR rating positively influences the acquisition premium through its synergistic capabilities, we could expect that the targets abnormal return is positively influenced as well. However, Aktas et al. (2011) find that the abnormal returns of the target do not depend on the targets own CSR rating. On the other hand, Aktas et al. (2011) do suggest that a higher CSR rating of the target firm leads to higher expected synergies from the deal. According to Jensen and Ruback (1983), targets of successful M&As earn significantly positive abnormal returns through the announcement of the deal and through the completion of the deal. Since the CSR rating of the target firm is positively correlated with the likelihood of being a M&A target and the success of the deal (Harper, 2012), this could result in higher abnormal returns for the target. This leads to the following hypothesis:

(14)

H2: The targets CSR rating has a positive effect on the target short-term abnormal returns.

Besides literature that suggests that the target gains (through the premium and abnormal returns) from having a high CSR score in the context of M&A, there are studies which suggest that the acquirer gains (through abnormal returns) from acquiring a high CSR target as well. Acquiring a high CSR target could be a signal from the acquirer about the willingness to learn from the CSR activities of the target firm, thus adding value to the acquirer shareholders (Aktas et al., 2011). Moreover, Aktas et al. (2011) find that if buying a target with a high CSR rating is a value creating decision, acquirer abnormal returns should have a positive relationship with the level of CSR of the target. According to Betton and Eckbo (2000) the most important determinants of the acquirer’s abnormal returns around M&A announcements are the payment means, the target’s status, the relative size of the deal and the uncertainty about the target’s valuation. One of the main determinants of the acquirer’s abnormal returns, i.e. the target’s status, could be closely related to the target firm’s CSR rating. This is in line with Aksak et al. (2016), who argue that CSR is one of the key tools to communicate industry norms and values, and thus increase the firm’s reputation. Furthermore, CSR activities may improve a firm’s access to financing sources because the reputation effect enhances a firms standing with financial market participants and governments (McGuire et al., 1988). This results from the fact that high CSR firms have a stronger reputation for keeping their commitments associated with the implicit contracts (Deng et al., 2013). This results in the following hypothesis:

(15)

3. Methodology

This section starts by describing the used methods to answer the research question. Figure 1, at the end of section 3.1, shows a graphic representation of how this thesis is structured. Finally, the section describes all the variables used in the study and how they are derived.

3.1 Research method

3.1.1 Preliminary analysis

As shown in figure 1, as a first step, univariate methods of analysis are used to compare the differences in means of the acquisition premiums, target abnormal returns and acquirer abnormal returns for subsamples of deals with high and low target ESG scores. Following Aktas et al. (2011), Deng at al. (2013) and Meckl and Theuenkorn (2015), the data sample is divided into two sub-samples, namely target firms with a high and low CSR rating. Aktas et al. (2011) rank the sample firms into two groups according to their IVA ratings, while Deng et al. (2013) divide the high and low CSR acquirers in their sample according to the sample median of their KLD rating. I follow Deng et al. (2013) as the CSR rating used in this study is also scaled rather than grouped6.

Using the Jarque-Bera test, I find out whether the dependent variables follow a normal distribution. If the Jarque-Bera test indicates a non-normal distribution of the variables, I use, additionally to the parametrical significance test, a non-parametrical significance tests to investigate whether the two sub-samples differ from each other significantly (Meckl and Theuenkorn, 2015). The parametric t-test relies on the assumption of normal distribution7 and non-parametric t-tests do not rely on the assumption of normal distribution. Following Meckl and Theuenkorn (2015), the Mann-Whitney test is used as non-parametric test in order to test whether the differences between the means of the high and low ESG score samples are significant.

3.1.2 Cross-sectional regression analysis

In order to examine the effect of the ESG scores on the premium this study performs a regression analysis (Cheng and Chan, 1995). Since not all acquirers in the sample are listed, the acquirer specific control variables are not available for all deals. Also, the acquirer and

6_{The CSR measure used in this study is the ESG rating obtained from the ASSET4 Database of Thomson} Reuters, which is scored on a scale from zero to 100. The IVA rating, used by Aktas et al. (2011), is expressed on a seven-point scale (‘AAA–CCC’).

(16)

target specific variables are likely to be correlated. Therefore, I run two different regressions regarding the premium for each separate ESG score. One regression including target specific control variables and one regression including acquirer specific control variables. The following equations give a representation of the specified regression model:

𝑃𝑅𝐸𝑀𝑖 = 𝛼0+ 𝛽1(𝐸𝑆𝐺𝑠𝑐𝑜𝑟𝑒𝑠)𝑖𝑡+ 𝛽2(𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠𝑡𝑎𝑟𝑔𝑒𝑡)_𝑖𝑡+ 𝜀𝑖𝑡 (1)

𝑃𝑅𝐸𝑀_𝑖 = 𝛼₀+ 𝛽₁(𝐸𝑆𝐺_{𝑠𝑐𝑜𝑟𝑒𝑠})_𝑖𝑡+ 𝛽₂(𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠_{𝑎𝑐𝑞𝑢𝑖𝑟𝑒𝑟})_𝑖𝑡 + 𝜀_𝑖𝑡 (2) In addition to the above, this study examines the effect of the ESG scores on the short-run abnormal returns of the target and the short-run abnormal returns of the acquirer through a regression analysis (Cheng and Chan, 1995). Due to the fact that not all acquirers in the sample are listed, the abnormal returns could only be calculated for the listed acquirers. I run a regression for the target abnormal returns on the ESG scores including target specific control variables and a regression for the acquirer abnormal returns on the ESG scores including acquirer specific control variables. The following equations give a representation of the specified regression models:

𝐶𝐴𝑅_𝑡𝑎𝑟𝑔𝑒𝑡𝑖 = 𝛼0+ 𝛽1(𝐸𝑆𝐺𝑠𝑐𝑜𝑟𝑒𝑠)𝑖𝑡+ 𝛽2(𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑠𝑡𝑎𝑟𝑔𝑒𝑡)_𝑖𝑡+ 𝜀𝑖𝑡 (3)

(17)

Figure 1. Graphic overview of the methodology

3.2 Variables

3.2.1 Dependent variables

Acquisition premium

The acquisition premium paid by the acquirer for the target firm measures, together with the target abnormal returns, the gains to the target from the acquisition. The acquisition premium is calculated as the percentage difference between the deal value per share to the target’s market value per share (i.e. Beckman and Haunschild (2002), as in equation (5).

𝑃𝑅𝐸𝑀_𝑖 =𝑂𝑖− 𝑃𝑖,𝑡−30

𝑃𝑖,𝑡−30 (5)

Where 𝑂𝑖 is the deal value per share, which is calculated as the deal value divided by the

amount of shares outstanding on the effective date. Pi,t-30 is the market value per share. The

(18)

is calculated as the market value four weeks prior to the announcement date divided by the amount of shares outstanding four weeks prior to the announcement date.

Target and acquirer CAR

To obtain the cumulative abnormal returns (CARs) to the target and to the acquirer, I use the standard event study methodology, which is described by Brown and Warner (1985) and MacKinlay (1997). In order to perform an event study, the event window and estimation window are determined. The event window is determined as one day prior to the event date, the day of the event, and the day after the event, which is in line with previous research like that of Aktas et al. (2011). Following MacKinlay (1997), the estimation period is day -121 through -2 for the 3 day CAR, where day 0 is the event date. The event date is referred to as the announcement date of the M&A transaction (Aktas et al, 2011). The abnormal returns of the target (ARi,t), are calculated as shown in formula (6).

𝐴𝑅𝑖,𝑡 = 𝑅𝑖,𝑡– 𝐸(𝑅𝑖,𝑡) (6)

Where Ri,t is the firms daily return and E(Ri,t) is the expected return in absence of the event.

All daily return data are adjusted for dividends and capital appreciations by taking Datastream’s Total Return Index (RIi,t). To calculate actual returns (Ri,t), the return index

(RIi,t) used by taking the log-returns per day, as in formula (7).

𝑅_𝑖,𝑡 = 𝑙𝑛 𝑅𝐼𝑖,𝑡

𝑅𝐼𝑖,𝑡−1 (7)

Then, I construct the market model to estimate the expected returns in the absence of the deal event. The market model is defined as follows:

𝐸(𝑅_𝑖,𝑡) = 𝛼_𝑖+ 𝛽_𝑖𝑅_𝑚𝑡+ 𝜀_𝑖,𝑡 (8)

Where Rmt is the market return on day t. MacKinlay (1997) uses the return index of the S&P

(19)

As MacKinlay (1997) explains, the abnormal return observations must be accumulated in order to draw overall inferences for an event. To compute the cumulative abnormal return (CAR), I determine, for each firm, the AR’s across the 3 days around the announcement day (from day -1 to day +1, where day 0 is the announcement day) (Aktas et al., 2011).

𝐶𝐴𝑅𝑡 = ∑𝑘𝑖=1𝐴𝑅𝑡 (9)

To assess whether the CAR is statistically significant, a Student t-test is applied. Also, literature discusses the process of handling non-normal distribution of the abnormal returns. When a non-normal distribution of the target and acquirer CAR is found, additionally, I test the statistical significance of the CAR’s by means of a non-parametric test. This study employs the Cowan (1992) rank-test. The Cowan (1992) test is an extension of the Corrado (1989) test for dealing with non-normal distributions of event windows longer than a day.

3.2.2 Independent variables

The independent variable in this study is the CSR performance of the target firm. To measure CSR performance, the environmental, social and governance (ESG) scores from the ASSET4 Database of Thomson Reuters are used. ASSET4 is recognized globally as a premier source of ESG data. The ESG score is an equal-weighted rating of a company’s financial and extra-financial performance based on the ASSET4 environmental, social and corporate governance ratings. The ratings are derived by company comparisons for a total of 226 Key Performance Indicators (KPI’s), which originate from 500 separate data points. This data is collected by research analysts who are trained to collect ESG data from multiple sources, including company reports, company filings, company websites, NGO websites, CSR reports and established and reputable media outlets.

(20)

3.2.3 Control variables

As discussed in section 2.1.3, prior literature identifies several economic factors that could influence the size of the premium offered in a deal as well as the target and acquirer abnormal returns (Malik 2014). I attempt to control for those factors. Following i.a. Cheng et al. (2016), the size, profitability, growth and leverage variables are included for both the target and the acquirer, but not specified in the same model. Table 1 provides an overview of all the control variables included, how they are measured and their expected signs. Their expected signs are based on the literature discussed in section 2.1.3.

Table 1. Description of the control variables

This table summarizes the control variables used in the main analysis and the robustness checks. The left column shows the variables, the middle column describes the variables and the right column represents the expected signs. The target and acquirer specific control variables are acquired from Datastream. The dummy variables are obtained from the Thomson One Database. The ROA ratio and the DTE ratio are available on yearly basis. Therefore, I use values at the end of the fiscal year prior to the acquisition announcement (Beckman and Haunschild, 2002). For the market value and the price-to-book ratio the values four weeks prior to the announcement are used.

Variable Description Expected sign

Premium CAR

(T) (A) (T) (A)

Target/Acquirer Size The size of the company is measured by the company's market value. +/- +/- +/- - Target/Acquirer Profitability The profitability of the company is defined by the return-on-assets (ROA)

ratio, net income minus bottom line plus interest expense corrected for tax divided by the average of last year's and currents total assets.

+ + +

-Target/Acquirer Growth The growth of the company is defined by the price-to-book ratio, which is the share price divided by the book value per share.

+ + + +

Target/Acquirer Leverage The leverage ratio of the company is defined by the debt-to-equity ratio, which is total debt divided by common equity.

+ +/- - +

Hostility Dummy Dummy that measures whether the acquisition was hostile. One if the takeover was hostile and zero otherwise.

+ + -

Industry Dummy Dummy that measures the industry relatedness. One if the target and acquiring firm are in related industries and zero otherwise. Industries are defined by two-digit SIC codes.

+ + +

Payment Dummy This dummy captures the pay-method. Dummy of one if the deal is completely cash financed and zero otherwise.

+ + +

Majority Dummy Following Beckman and Haunschild (2002) and Aktas et al. (2011), I do not further restrict the sample size by removing deals for which the bidder holds more than 50% of the target before the announcement. The majority dummy controls for this and measures whether the acquisition consisted of a majority interest purchase. This dummy takes the value of one if the percentage of ownership held by the acquiring firm in the target firm prior to the deal was less than 50 percent and zero otherwise.

+ - +

Competition Dummy Dummy for the number of bidders on the target. This dummy is set to one if there was more than one bidder and zero otherwise.

+ + -

Regional Dummies Regional dummies are included in order to control for geographic effects. I split the target companies in four region; Western, Southern, Northern and Eastern Europe.

(21)

4. Data

This section presents the sources and types of data this study uses. It starts by explaining how the data is collected. Next, I discuss the sample characteristics and present various descriptive statistics on the dataset.

4.1 Data collection

M&A data is acquired from the Thomson ONE, formally known as Securities Data Company (SDC), Mergers and Acquisitions database. This study focuses on deals during the period 2004-2016 with a value above one million euro. Completed deals with publicly traded, European target companies and more than 50% ownership for the acquirer after the deal are selected. The initial sample consists of 3.217 deals.

The ESG ratings are collected from the ASSET4 Database of Thomson Reuters. All known ESG ratings are matched with the listed target companies of the selected 3.217 deals for the year of the announcement of the deal. Due to the unavailability of ESG information 3.077 deals have to be excluded, which restricts the sample size to 140 deals. Subsequently, information needed to calculate the acquisition premium and abnormal returns is collected from DataStream. Seven more deals are excluded due to unavailability of information required for the premium and three deals due to unavailability of information required for the target CAR, this data seems to be missing completely at random. This results in a final sample of 133 deals for the acquisition premium with target specific control variables and 130 deals for the target CAR.

Only 62 acquirers from the full sample are listed and acquirer specific data can only be found for those listed acquirers. Consequently, I end up with a sample of 62 for the acquisition premium with acquirer control variables and a sample of 54 for the acquirer abnormal returns. Figure 1 in section 3.1 gives an overview of the sample sizes for each regression. Unfortunately, the small sample may reduce the explanatory power of the regression model.

(22)

4.2 Sample description

4.2.1 Sample construction and descriptive statistics

Table 2 gives an overview of the sample distribution of this study. The left column represents the geographical distribution of the firms in the sample. The sample consists of 133 European targets listed in 20 different countries. With 26.3%, targets from the U.K. comprise more than

one fourth of the sample. When comparing this distribution with the primary sample8, which

is unrestricted by ESG data unavailability, a comparable geographic distribution is shown. Therefore, I conclude this distribution is representative for all M&A transactions within Europe with a listed and European target. A possible reason for the large amount of U.K. targets could be that the U.K. contains relatively more listed companies than other European countries.

Table 2. Sample characteristics

This table shows the data sample composition. The column 'Country' indicates the different countries for the target and acquirer. The column 'Industry' represents the distribution by targets' industry and acquirers' industry, based on the company SIC codes. The column 'Year’ reports the distribution by the effective year. The column ‘No.’ represents the number of companies per item and the column ‘%’ shows the percentage of the total sample of the respective item.

Target Acquirer Target Acquirer

Country No. % No. % Industry No. % No. % Year No. %

Austria 4 3.01% 3 2.26% Construction 4 3.01% 2 1.50% 2004 1 0.75%

Belgium 1 0.75% 2 1.50% Finance, insurance, real estate 32 24.06% 63 47.37% 2005 8 6.02%

Czech Republic 1 0.75% 0 0.00% Manufacturing 44 33.08% 34 25.56% 2006 18 13.53%

Finland 1 0.75% 2 1.50% Mining 7 5.26% 8 6.02% 2007 14 10.53%

France 13 9.77% 16 12.03% Services 13 9.77% 5 3.76% 2008 7 5.26%

Germany 15 11.28% 13 9.77% Transport, communication 24 18.05% 13 9.77% 2009 4 3.01%

Greece 10 7.52% 7 5.26% Wholesale, retail trade 9 6.77% 6 4.51% 2010 8 6.02%

Ireland-Rep 5 3.76% 6 4.51% Public administration 2 1.50% 2011 18 13.53%

Italy 10 7.52% 8 6.02% 2012 4 3.01% Jersey 1 0.75% 1 0.75% 2013 11 8.27% Luxembourg 3 2.26% 3 2.26% 2014 11 8.27% Netherlands 6 4.51% 10 7.52% 2015 15 11.28% Norway 1 0.75% 1 0.75% 2016 14 10.53% Poland 3 2.26% 1 0.75% Portugal 1 0.75% 1 0.75% Russian Fed 4 3.01% 2 1.50% Spain 7 5.26% 9 6.77% Sweden 2 1.50% 2 1.50% Switzerland 10 7.52% 7 5.26% United Kingdom 35 26.32% 22 16.54% United States 6 4.51% Brazil 1 0.75% British Virgin 1 0.75% Canada 1 0.75% Cayman Islands 1 0.75% Israel 1 0.75% Japan 3 2.26% Kazakhstan 1 0.75% New Zealand 1 0.75% Singapore 1 0.75%

Total 133 100% 133 100% Total 133 100% 133 100% Total 133 100%

The middle column of table 2 shows the industry distribution of the 133 targets in the sample. The companies are classified in the ten main industries based on their two-digit Standard Industrial Classification (SIC) codes. The majority of the targets are concentrated in the

(23)

manufacturing industry with 33.1%. Manufacturing is followed by the finance, insurance and real estate industry with 24.1%. Almost half of the acquirers in the sample are active in the finance industry, followed by manufacturing with 25.6%. When comparing this with the unrestricted sample, this distribution seems representative for all M&A transactions within Europe with a listed and European target.

The right column lists the yearly distribution of the 133 deals in the sample. The percentage of deals in 2004 and 2005 is small. Since the percentage of deals in 2004 and 2005 in the primary sample is much larger7, this is probably due to unavailability of the target firms’ ESG data in the ASSET4 database. This is confirmed by the fact that the number of companies in the ASSET4 ESG database multiplied 4.5 times from 2002 to 20149. The distribution shows a drop after 2008, which is also shown in the unrestricted sample. This drop is likely to be the result of the global financial crisis, which caused a downfall in the number of M&A deals in the consecutive years.

Table 3 shows the descriptive statistics of the sample. It is remarkable how especially market value variables show large a standard deviation in their data. However, this appears logical as the market value of the acquirer range from €142 million to €168,785 million.

Table 3. Summary statistics

This table shows the summary statistics for the sample used in this study. The full sample contains 133 completed deals with a European target, during the period 2004-2016 with a value above one million euro and an available target ESG-rating. The selected deals have 50% of ownership for the acquirer after the deal. The sample contains 62 listed acquirers and 133 listed targets.

Variable No. Mean Median

Std.

Deviation Max. Min. Skew. Kurt.

Dependent Variables Acquisition Premium 133 0.36 0.17 1.06 8.27 -0.99 4.53 27.76 CAR (-1,1) Target 130 0.07 0.04 0.13 0.53 -0.42 0.35 2.52 CAR (-1,1) Acquirer 54 0.03 0.01 0.10 0.60 -0.11 3.97 21.74 Independent Variables ESG score 133 55.80 61.76 30.53 96.73 4.00 -0.26 -1.41 Environmental score 133 59.86 69.14 30.61 97.47 9.13 -0.34 -1.51 Social score 133 61.03 63.74 29.39 97.48 4.45 -0.39 -1.17

Corporate Governance score 133 48.14 46.96 29.88 95.39 2.68 0.03 -1.49

Control Variables

Target Market Value (€ mln) 133 7,484.22 3,129.80 13,098.08 86,777.69 17.99 3.68 15.76 Target Profitability 133 6.24 4.57 19.95 217.76 -28.12 9.02 95.70

Target Growth 133 4.32 1.97 22.77 263.75 -12.51 11.24 128.46

Target Leverage 133 171.95 64.77 765.35 5496.07 -3583.52 0.94 25.09 Acquirer Market Value (€ mln) 62 22,029.15 9,384.50 30,433.78 168,785.00 141.99 2.76 9.12 Acquirer Profitability 62 6.43 5.77 9.09 37.37 -16.44 0.87 3.89

Acquirer Growth 62 3.13 2.06 6.54 51.71 0.00 7.03 52.22

Acquirer Leverage 62 184.09 63.44 595.94 3968.89 -1532.50 4.12 27.93 Table 4 reports the statistical significance of the target and acquirer CAR. As the target and acquirer CAR both follow a non-normal distribution (Appendix 1, table A), I also perform a

(24)

non-parametric test, namely the Cowan (1992) rank-test. The average target CAR is 7.30% and significant, which is substantially higher than previous EU studies. Campa and Hernando (2004) find a CAR of 3.93% for European targets. For a worldwide sample, Aktas et al. (2011) find a slightly higher target CAR of 9.60%. The average acquirer CAR of 2.69% is statistically significant with a p-value of 0.02 for the parametric test and insignificant for the non-parametric test. This acquirer CAR is high compared to previous studies who often find zero, negative or slightly positive gains (Jensen and Ruback, 1983; Campa and Hernando, 2004). Also, this acquirer CAR is higher than the negative acquirer CAR of -1.16% of Aktas et al. (2011).

Table 4. Abnormal returns test statistics

This table reports additional descriptive statistics concerning the acquirer and target three-day cumulative abnormal returns (CAR). The full sample contains completed deals with a European target, during the period 2004-2016 with a value above one million euro and an available target ESG-rating. The selected deals have 50% of ownership for the acquirer after the deal. N shows the number of observations. The t-statistic corresponds to the Student t-test and the Cowan rank-test.

Target CAR (N = 130) Acquirer CAR (N = 54)

T-test Cowan T-test Cowan

Mean 7.30% 2.69%

t-Statistic 6.55 1.41 2.01 -0.89

p-Value 0.00 0.08 0.02 0.19

4.2.2 Variables correlations

Because I run a different regression for the premium on the sample with target specific control variables and on the sample with acquirer specific control variables, I also present two separate correlation tables regarding the acquisition premium. Furthermore, the variables for the regressions of the target and acquirer CAR are also presented in two different correlation tables. The correlation tables are shown in Appendix 2. All correlations which are higher than 0.6 are marked bold.

(25)

5. Results

First, I examine the difference in means of the premium, the target and the acquirer abnormal returns between subsamples of high and low CSR targets under the preliminary analysis by means of t-tests. Second, I investigate the effect of the targets CSR rating on the premium, target and acquirer abnormal returns using OLS-regressions.

5.1 Preliminary analysis

As discussed in the methodology section, the sample is divided in high and low CSR targets according to the sample median of the CSR score. I check whether non-parametric tests are needed with the Jarque-Bera test. Table A in appendix 1 reports the results of the Jarque-Bera test and indicates that the premium and target and acquirer CAR are non-normally distributed (p-value 0.00). Therefore, I use parametric, as well as non-parametric significance tests to investigate whether the two sub-samples differ significantly from each other.

5.1.1. Shareholder gains for high CSR vs. low CSR targets

In table 5, I report the average premium (panel A), target CAR (panel B) and acquirer CAR (panel C) for subsamples based on the CSR ratings of the target firms. Below I discuss whether there is a significant difference in the means of the premium, target and acquirer CAR between the subsamples of high and low CSR targets.

Panel A of table 5 shows the average acquisition premiums for subsamples of high and low CSR targets. The difference in means tests indicate that the acquisition premium does not depend on whether the target has a high or low ESG, social or corporate governance score. However, the Mann-Whitney test indicates that the acquisition premium is significantly higher for the subsample with a high environmental score.

Panel B of table 5 reports the average target CARs for subsamples of high and low CSR targets. Based on those results, the target CAR seems unrelated to whether the target has a high or low ESG, environmental, social and corporate governance rating. However, the mean of the target CAR for high CSR targets is higher than for low CSR targets.

(26)

CAR appears to be negatively related to the targets corporate governance score, with the difference in means statistically significant at 5%. Furthermore, the equality in mean of the acquirer CARs between the high and low environmental subsamples is rejected at the 10% level, with a higher acquirer CAR for the high environmental subsample.

Table 5. T-tests - difference high CSR vs. low CSR targets

This table reports the average premium, target CAR and acquirer CAR for subsamples based on the ESG rating of the firms in the second column. High and low ESG firms are divided based on the sample median of the ESG score. The sample of deals covers completed deals with a European target, during the period 2004-2016 with a value above one million euro. The selected deals have 50% ownership for the acquirer after the deal. This table represents the outcomes of the two sample t-test and the Mann Whitney tests. The t-statistics and p-values present the values from a test of mean differences between the subsamples.

Student T-test Mann-Whitney test

Mean No. T -statistic P-value T -statistic P-value

Panel A: Premium ESG score Low rating 28.16% 66 0.873 0.385 -0.257 0.399 High rating 44.23% 67 Environmental score Low rating 32.12% 66 -0.442 0.659 -1.679 0.047 High rating 40.32% 67

Corp governance score

Low rating 42.58% 66 -0.679 0.498 -0.081 0.468 High rating 30.02% 67 Social score Low rating 26.75% 66 1.027 0.307 -0.414 0.339 High rating 45.62% 67

Panel B: CAR Target

ESG score Low rating 6.27% 65 0.489 0.626 -0.375 0.354 High rating 7.31% 65 Environmental score Low rating 5.84% 65 0.892 0.375 -0.198 0.422 High rating 7.74% 65

Low rating 5.39% 65 1.322 0.189 -0.826 0.204 High rating 8.20% 65 Social score Low rating 5.80% 65 0.931 0.354 -0.296 0.384 High rating 7.78% 65

Panel C: CAR Acquirer

ESG score Low rating 2.96% 27 1.005 0.322 -0.112 0.455 High rating 0.81% 27 Environmental score Low rating 1.81% 27 -0.072 0.943 -1.479 0.070 High rating 1.96% 27

(27)

5.2 Cross-sectional regression analysis

Before running the regressions, the assumptions of the OLS are checked. The null hypotheses of homoscedasticity and no serial correlation cannot be rejected for all regression models. Therefore, I do not correct for heteroscedasticity and serial correlation. Subsequently, I reject the null hypotheses of normality in the residuals by the Jarque-Bera test in the regressions of all samples. I check the robustness of these regressions in section 5.3, by running the regressions again on an adjusted (i.e. winsorized) sample that shows no signs of non-normality in the residuals. Below I discuss the effect of the targets CSR score on the premium, the target CAR and the acquirer CAR.

5.2.1 The impact of the target’s CSR score on the acquisition premium

Table 6 and table 7 report the empirical results of the tests of the hypotheses regarding the impact of the targets’ ESG scores on the acquisition premium with target and acquirer specific control variables. None of the regressions indicate a significant relationship between the acquisition premium and CSR performance.

The effects of the ESG and social scores on the acquisition premium are insignificantly positive. The effect of the corporate governance score on the premium is negative in both cases, when considering target specific and acquirer specific control variables. Also, the findings are insignificant. The environmental score shows a positive sign when considering target specific control variables and a negative sign when considering acquirer specific control variables. However, both results indicate no statistical significance. Hence, the ESG, environmental, corporate governance and social score do not seem to have a significant effect on the acquisition premium, regardless of whether target or acquirer specific control variables are included in the regression. This contradicts the research of Malik (2014) who finds a significantly positive effect of the target’s CSR rating on the acquisition premium. However, Malik (2014) uses a sample of U.S. deals and this study examines a sample of European deals. Also, Malik (2014) uses the KLD ratings from the MSCI ESG database as a measure for CSR performance whereas this study uses the ESG scores from the ASSET4 database.

(28)

my expectations in both regressions. Moreover, the industry dummy is significantly positive when considering target specific control variables. The signs of the hostility, payment and competition dummy on the premium are inconsistent with my expectations in both tables.

Based on the abovementioned results I reject hypothesis H1, that target firms with higher CSR ratings receive higher acquisition premiums. Those results are consistent with the preliminary test results, which indicate no difference in the acquisition premium when comparing a sample of high and low ESG target firms. This insignificant effect of the CSR performances on the acquisition premium may imply that the market value of the target shares captures a large portion of the value of the firms CSR performance and investors do not expect large synergies from the CSR performance of the target firm.

Table 6. Multiple regression: explaining Premium with the Target ESG rating

This table represents the results of a multiple OLS regression with the acquisition premium as dependent variable and target specific control variables. The sample contains 133 completed deals with a European target, during the period 2004-2016 with a value above one million euro and an available target ESG-rating. The selected deals have 50% of ownership for the acquirer after the deal. The first column represents the independent and control variables in the model. The other columns from left to right represent the regression results of the ESG score, environmental score, corporate governance score and social score, respectively. The first number shows the estimated coefficient and the second number represents the t-statistic. The intercept and geographical dummies are not shown in the table for brevity. *, **, *** indicate statistical significance at the 0.10, 0.05 and 0.01 levels.

Variable Independent Variables ESG score 0.001 0.314 Environmental score 0.000 0.117

Corporate Governance score -0.005

(29)

Table 7. Multiple regression: explaining Premium with the Target ESG rating

This table represents the results of a multiple OLS regression with the acquisition premium as dependent variable and acquirer specific control variables. The sample contains 62 completed deals with a European target, during the period 2004-2016 with a value above one million euro and an available target ESG-rating. The selected deals have 50% of ownership for the acquirer after the deal. The first column represents the independent and control variables in the model. The other columns from left to right represent the regression results of the ESG score, environmental score, corporate governance score and social score, respectively. The first number shows the estimated coefficient and the second number represents the t-statistic. The intercept and geographical dummies are not shown in the table for brevity. *, **, *** indicate statistical significance at the 0.10, 0.05 and 0.01 levels.

Variable Independent Variables ESG score 0.000 -0.036 Environmental score -0.001 -0.102

Corporate Governance score -0.007

-0.984 Social score 0.003 0.408 Control Variables Acquirer MV 0.000 0.000 0.000 0.000 -0.417 -0.424 -0.235 -0.549 Acquirer ROA 0.023 0.023 0.022 0.020 1.014 1.003 0.985 0.828 Acquirer PTB -0.022 -0.023 -0.020 -0.020 -0.810 -0.815 -0.736 -0.707 Acquirer DTE 0.000 0.000 0.000 0.000 -0.223 -0.234 -0.529 -0.162 Hostility dummy -0.472 -0.459 -0.243 -0.627 -0.347 -0.341 -0.182 -0.457 Industry dummy 0.531 0.531 0.519 0.528 1.007 1.007 0.994 1.003 Payment dummy -0.091 -0.097 -0.268 -0.026 -0.220 -0.239 -0.624 -0.063 Majority dummy 0.404 0.403 0.485 0.383 0.894 0.899 1.074 0.853 Competition dummy -0.343 -0.340 -0.408 -0.293 -0.568 -0.584 -0.704 -0.497 F-statistic 0.346 0.347 0.427 0.360 R-squared 0.086 0.086 0.104 0.089 Number 62 62 62 62

5.2.2 The impact of the target’s CSR score on the target CAR

I find no significant results for the effect of target CSR on the acquisition premium, as a measure of target gains. However, since it may be the case that target CSR has a different effect on the target CAR, as a measure of target gains, I present the effect of CSR on target CAR in table 8. Table 8 reports the empirical results of the tests that examine the impact of the targets’ ESG scores on the target CAR with target specific control variables. I find that only the environmental and social performance of the target firm positively influence the target CAR.

(30)

significant. Hence, based on the ESG and corporate governance score, I find no support for hypothesis H2. These results are in line with the findings of Aktas et al. (2011), who find that the wealth effects to target shareholders do not seem to depend on the targets own CSR rating. Furthermore, I do find evidence for a significant positive effect of the environmental score and social score on the target CAR, which indicates I cannot reject hypothesis H2, that states that the targets CSR rating has a positive effect on the target abnormal returns. An increase in the environmental or the social rating of 1 unit, lead to an increase in the target CAR of 0.1%. This contradicts the findings of Aktas et al. (2011), who find no effect of the environmental and social rating of the target on the target gains.

Table 8. Multiple regression: explaining Target CAR with the Target ESG rating

This table represents the results of a multiple OLS regression with the target CAR (-1,1) as dependent variable and acquirer specific control variables. The sample contains 62 completed deals with a European target, during the period 2004-2016 with a value above one million euro and an available target ESG-rating. The selected deals have 50% of ownership for the acquirer after the deal. The first column represents the independent and control variables in the model. The other columns from left to right represent the regression results of the ESG score, environmental score, corporate governance score and social score, respectively. The first number shows the estimated coefficient and the second number represents the t-statistic. The intercept and geographical dummies are not shown in the table for brevity. *, **, *** indicate statistical significance at the 0.10, 0.05 and 0.01 levels.

Variable Independent Variables ESG score 0.000 1.252 Environmental score 0.001 * 1.913

Corporate Governance score 0.001

(31)

This suggests that, the higher the environmental and social score of the target, the higher the expected synergy. Moreover, these findings seem to indicate that European investors weight environmental and social performance more positively than corporate governance since the effect of the corporate governance rating is insignificant.

Concerning the control variables in table 8, although not significant, the signs of the target’s market value, return-on-assets and price-to-book ratio and the payment and competition dummy are in line with my expectations. Moreover, the payment dummy has the expected positive sign and is significant. On the other hand, the signs of the target’s debt-to-equity ratio and the hostility and majority dummy are not in line with previous literature.

Overall, the estimates in table 8 are partially consistent with hypothesis H2, which states that the targets CSR rating has a positive effect on the target abnormal returns. These findings partially contradict the preliminary analysis, which indicates that the target CAR is independent of its own CSR rating.

5.2.3 The impact of the target’s CSR score on the acquirer CAR

Besides the effect of the target’s CSR score on the target gains as discussed above, it is also interesting to investigate the effect of the targets CSR score on the acquirer gains. Table 9 reports the empirical results of the tests that examine the impact of the targets’ ESG scores on the acquirer CAR with acquirer specific control variables. None of the regressions indicate a significantly positive relationship between the acquirer CAR and CSR performance.

As reported in table 9, the coefficients of the target ESG rating, environmental rating and the social rating on the acquirer abnormal returns are positive but insignificant. Therefore, I find no support for hypothesis H3. Additionally, the coefficient of the corporate governance score on the acquirer abnormal returns is negative and statistically significant at 10%, which also does not support hypothesis H3, that the targets corporate governance rating has a positive effect on the target abnormal returns. In contrast, Aktas et al. (2011) are able to confirm the positive effect of those CSR measures on the acquirer returns. This contradictory result of Aktas et al. (2011) could be due to the fact that they use a worldwide sample and a different measure, i.e. the IVA rating, for CSR performance.

(32)

hostility, payment and majority dummy are not consistent with my expectations based on previous literature. Moreover, the negative sign of the majority dummy is significant.

So, the estimates in table 9 indicate that the targets CSR rating has no positive effect on the acquirer abnormal returns. These findings imply that investors do not seem to value CSR investments (i.e. the acquisition of a high CSR target) positively. For the ESG, corporate governance and social rating these findings are in line with the preliminary analysis. However, the preliminary analysis implies that high environmental targets show higher acquirer CARs.

Table 9. Multiple regression: explaining Acquirer CAR with the Target ESG rating

This table represents the results of a multiple OLS regression with the acquirer CAR (-1,1) as dependent variable an target specific control variables. The sample contains 62 completed deals with a European target, during the period 2004-2016 with a value above one million euro and an available target ESG-rating. The selected deals have 50% of ownership for the acquirer after the deal. The first column represents the independent and control variables in the model. The other columns from left to right represent the regression results of the ESG score, environmental score, corporate governance score and social score, respectively. The first number shows the estimated coefficient and the second number represents the t-statistic. The intercept and geographical dummies are not shown in the table for brevity. *, **, *** indicate statistical significance at the 0.10, 0.05 and 0.01 levels.

Variable Independent Variables ESG score 0.000 -0.647 Environmental score 0.000 0.450

Corporate Governance score -0.001 *

(33)

5.3 Robustness checks

Winsorizing

Non-normality of the error terms is detected in all the OLS-regressions. To see if the non-normality of the error terms has an influence on the results the data is not trimmed, but winsorized at the 5th and 95th percentile to avoid deleting useful data (Powel, 2004). Any data above the 95th percentile is replaced with the value at the 95th percentile, and any value below the 5th percentile is replaced with the value at the 5th percentile. Winsorizing is applied for all control variables, the premium and the target and acquirer CAR. The ESG scores are not winsorized because those are scored on a scale from zero to 100 and therefore those variables contain no outliers. When running the OLS-regressions again, the error terms show no signs of non-normality. The results of this analysis are presented in tables G-J in appendix 3. No large differences between the OLS-regressions on the original data and the winsorized data are observed. Hence, I conclude the non-normality of the error terms has no influence on the results.

Different event windows CAR

Does the target’s CSR performance influence target and acquirer gains? Evidence from mergers and acquisitions