Do Acquisitions Generate Abnormal Returns? Evidence from the Deregulated Electric Utility Industry

1

Do Acquisitions Generate Abnormal Returns?

Evidence from the Deregulated Electric Utility

Industry

University of Groningen

Uppsala University

Faculty of Economics and Business Department of Business Studies

MSc International Financial Management

MSc Business and Economics

Ronald Steensma

S2413744

rost5824

Supervisor: Dr. W. Westerman

Co-assessor: Dr. N. Selmane

June 2018

Abstract

2

1. Introduction

As a manager, would you cherish or averse acquisitions in the electric utility sector? Naturally, the fundamental aim of acquisitions is to generate shareholder value, however, in many scenarios acquisitions are not value enhancing (Alexandridis et al. 2010). In this sense, having knowledge about acquisitions is important, since it helps managers to define corporate strategy, and it gives policymakers substance to underpin strategies to enhance economic prosperity. The number of studies that examine acquisitions in the electric utility sector is parsimonious compared to other sectors (Jandik and Makhija, 2005, Kishimoto et al., 2017), but, more importantly, only a few academics have focussed on acquisitions in the electric utility sector after the sector liberalized, while due to liberalization the landscape for corporate control has dramatically changed from 1992 onwards (Leggio and Lien, 2000, Becker-Blease et al., 2008, Kishimoto et al., 2017).

In a regulated market, artificial monopolies can be created by the government. In exchange for compliance with specified rules, such as fixed electricity rates, particular companies have the privilege to serve the market without being exposed to regular market forces (Kishimoto et al., 2017). Due to regulation, the market for corporate control was characterized by long and rigid approval processes and regulators had the authority to return savings that originate from acquisitions to ratepayers (Leggio and Lien, 2000). Consequently, acquirers perceived little incentive to take the risk of engaging in acquisitions and the market for corporate control did not receive major managerial attention (Domanico, 2007).

3 changed as a consequence of deregulation (Leggio and Lien, 2000, Becker-Blease et al., 2008, Kishimoto et al., 2017).

Motivated by the lack of consensus on the value creation of acquisitions in the deregulated electric utility industry, this study is conducted. A few repeatedly cited authors as Leggio and Lien (2000) Becker-Blease et al. (2008) and Kishimoto et al. (2017) already examined the deregulated market, however, this paper diverts from these authors by using a relatively recent sample that ranges further than 1996, 2001 and 2013 respectively. This study also covers a geographic area which is broader than only US acquirers as in Leggio and Lien (2000) and Becker-Blease et al. (2008) and by examining a broader European area as suggested for future research by Kishimoto et al. (2017). Ultimately, the cumulative abnormal returns (CARs) of 714 acquisition announcement between 1997 and 2017 of acquirers stemming from the American and European electric utility industry are calculated. CARs are determined over a three-day and eleven-day event window by the market model and the mean adjusted return model.

This study documents an insignificant average eleven-day CAR of 0.2% for acquiring firms in the deregulated electric utility industry. European bidders generate a significant shareholder return 0.94% which outperforms the average CAR of -0.02% stemming from American firms, meaning that the market for corporate control provides on average a fruitful investment avenue for European bidders. For the aggregate sample, it is found that the geographic diversification strategies do not outperform each other, whereas industry-focussed deals outperform industry-diversified deals. Moreover, additional substance should be given to the effect of diversification. American managers should take into consideration that, cross-border acquisitions within the American region destruct shareholder value. Thereby, European managers should engage in geographic or industry-focussed deals, because these deals generate significant shareholder value.

In the next section literature regarding M&A in the electric utility industry is reviewed combined with the hypotheses of this study. Especially literature regarding acquisitions in the deregulated electric utility industry is considered. In the following section, the used methodology of this study is described combined with descriptive statistics. Subsequently, the results are discussed and by the use of additional literature these results are compared with the proposed hypotheses. The last section consists of the conclusion combined with managerial implications and a limitation is discussed.

2. Literature Review and Hypothesis Building

2.1 Value creation of acquisitions

4 emanate from: efficiencies in production, economies of scale in purchasing and procurement, the elimination of duplicate energy delivery services and from the integration of corporate functions such as administration and marketing (Becker-Blease et al., 2008, Kwoka and Pollit, 2010). When investors believe such synergies can be obtained, this might be reflected in the stock prices when an acquisition is announced.

Regulation used to impede valuable deals to occur, which resulted to lower acquisition gains in the electric utility industry compared with more liberalized industries (Bartunek et al., 1993) Leggio and Lien (2000) analyzed CARs of 76 mergers of publicly traded electric utility firms stemming from the US between 1983 and 1996. With regard to the US Energy Policy Act of 1992 (Congress U.S. 1992), this sample is an illustration of the transition between a regulated and deregulated market for corporate control. For the deregulated period (pre-1992) M&A did not significantly enhance value for the bidding firm. In contrast, a negative and significant mean bidder return of -0.92% is found for the deregulated period between 1994 and 1996 (Leggio and Lien 2000). The authors argue that most lucrative acquisition candidates were already acquired in the period between 1992 and 1994, which caused that limited synergies where available between 1994 and 1996.

Becker-Blease et al. (2008) corroborate the findings of Leggio and Lien (2000). They analyzed 70 proposed US M&A deals between 1992 and 2001 and found based on three-day CARs a significant mean bidder return of -1.29%. The authors also document negative deal-to-close CARs and a five-year past acquisition operating performance that is worse than a non-merging control portfolio. Kishimoto et al. (2017) analyzed 377 acquisitions of publicly traded acquiring firms originating from the US, Canada, UK, France, and Germany. Their results definitely divert from the view of Becker-Blease et al. (2008) and Leggio and Lien (2010) that acquisitions in the deregulated electric utility industry are value destructive. Kishimoto et al. (2017) show based on a significant three-day mean CAR of 0.86% that the stock market reacts positively to acquisition announcements in the deregulated electric utility industry. Kishimoto et al. (2017) also document increased long-term operating performance for the three years after the deal closed.

In conclusion, Leggio and Lien (2000) and Becker-Blease et al. (2008) show that acquisitions in the electric utility industry destruct shareholder value, whereas Kishimoto et al. (2017) show that acquisitions are positively associated with the generation of shareholder value. Motivated by these mixed findings on the value creation of acquisitions in the deregulated electric utility industry the following two hypotheses are suggested:

 H1a: Acquisitions in the deregulated North American and West European electric utility industry increase shareholder value for the acquiring firm.

5

2.2 Geographic diversification

When diversifying geographically, acquiring firms can obtain peculiar benefits. For instance: the firm can reach broader product markets, gain from differences in product prices, extract gains from the targets legal and tax environment and lastly it provides the opportunity to exploit finance opportunities originating from the target country (Becker-Blease et al., 2008, Li et al., 2016, Frésard et al., 2017). On the other hand, it is questionable to what extent these cross-border synergies can be captured due to increasing managerial complexities originating from the new environment (Bodnar et al., 1997). In addition, geographical diversification is associated with risk dispersion. It would make a company less sensitive to investment opportunities and it lowers idiosyncratic risk which generally undermines the financial position of a diversified investor (Martin and Sayrak, 2003, Li et al., 2016). Becker-Blease et al. (2008) analyzed with a US sample if diversifying into a state where the acquiring firm was not active yet affected the value creation of mergers. The authors found a mean bidder CAR of -1.33% for focussed deals and a mean bidder CAR of -2.67% for diversified deals implying that an acquisition within the same a region destructs less value for the acquiring firm. Kishimoto et al. (2017) found that the stock market reacts approximately equally positive to domestic and cross-border acquisition announcements. A mean three-day CAR of 0.85% for domestic acquisitions and a slightly higher CAR of 0.86% for cross-border acquisitions is found by the authors. Becker-Blease et al. (2008) and Kishimoto et al. (2017) show mixed results regarding the effect of geographic diversification on the generation of shareholder value. Therefore the following two hypotheses are proposed.

 H2a: Geographic-focussed acquisitions in the deregulated North American and West European electric utility industry generate more shareholder value (destroy less shareholder value) compared with geographic-diversified acquisitions.

 H2b: Geographic-focussed acquisitions in the deregulated North American and West European electric utility industry generate less shareholder value (destroy more shareholder value) compared with geographic-diversified acquisitions.

2.3 Industry diversification

6 Li et al., 2016). Thereby, stock prices reflect the operating performance of a company. This reflection becomes more opaque when a firm diversifies into other industries, which affects the investment efficiency of the shareholder (Li et al., 2016).

Jandik and Makhija (2005) argue that in the regulated electric utility industry, focussed firms were due to regulatory restrictions unable to freely invest excess cash, while diversified firms could bring idle capital into more promising deregulated industries. Due to this valuable investment avenue diversifying deals where implied to create more shareholder value compared to focussed deals (Bartunek et al., 1993). Due to deregulation of the electric utility industry, this diversification premium disappeared and investment opportunities emerged for focussed firms (Jandik and Makhija, 2005). Leggio and Lien (2000) corroborate this view of Bartunek et al. (1993) and Jandik and Makhija (2005), they document for their regulated subsample higher CARs for diversified deals compared to focussed deals. This diversification premium is not documented for their deregulated subsample, where focussed deals generate significantly higher CARs compared with diversified deals. The findings of Leggio and Lien (2000) are contradicted by Becker-Blease et al. (2008) and Kishimoto et al. (2017). The authors find a mean CAR of -1.98% and 0.65% respectively for focussed deals and a mean CAR of -1.40% and 1.66% respectively for industry diversified deals meaning that diversified deals generate higher returns than focussed deals. Due to these opposite findings of Leggio and Lien (2000), Becker-Blease et al (2008) and Kishimoto et al. (2017) regarding the effect of Industry diversification on CARs the following two hypotheses are proposed.

 H3a: Industry-focussed acquisitions in the deregulated North American and West European electric utility industry generate more shareholder value (destroy less shareholder value) compared with industry-diversified deals.

 H3a: Industry-focussed acquisitions in the deregulated North American and West European electric utility industry generate less shareholder value (destroy more shareholder value) compared with industry-diversified deals.

2.4 North American and West European regions

7 pioneer of the European region (Kishimoto et al. (2017). Deregulation started later in other European countries and the pace of deregulation was low compared to the America region, with as consequence that the European market has for a long time been characterized by regulated, national, dominant, vertically integrated monopolies (Domanico, 2007).

Kishimoto et al. (2017) explicitly address the difference in CARs between the American and European regions. The authors report a significant mean CAR of 0.84% and 0.92% respectively, which suggests that the stock market responds slightly more positive to acquisition announcements stemming from European acquirers compared to acquisition announcements stemming from American acquirers. For the American region, domestic acquisitions generate higher returns than cross-border acquisitions, which is in contrast to the European region where cross-border acquisitions generate higher returns. For both regions, industry-focussed deals generate significant shareholder returns, while this is not the case for industry-diversified deals.

In spite of the documented CARs of both regions, no significant difference in the generation of shareholder value, between the two regions is documented. Besides Kishioto et al (2017), no literature is found that specifically addresses the difference in the generation of shareholder value between North American and West European acquiring firms in the electric utility industry. Therefore the following two hypotheses are suggested.

 H4a: Acquisitions in the deregulated electric utility industry generate more shareholder value if the acquiring firm stems from the North American region compared with acquisitions where the acquiring firm stems from the West European region.

 H4b: Acquisitions in the deregulated electric utility industry generate less shareholder value if the acquiring firm stems from the North American region compared with acquisitions where acquiring firm stems from the West European region.

3.

Methodology

3.1 Deal selection and distribution

8 or greater and deals with a relative deal size larger than 0.5% of the market capitalization of the acquiring firm are included, since it is assumed that such deals are of sufficient size to receive attention from the stock market (Kishimoto et al., 2017). To ensure that only acquisitions are reflected the restriction is imposed that 100% of the shares of the target are acquired during the transaction.

3.1.1 Geographic sample criteria and distribution

In line with Kishimoto et al. (2017), the United States and Canada reflect the American region, and the United Kingdom, France, and Germany reflect the European region. An extension is made for the European region since this region can be perceived broader than three countries. Domanico (2007) argues that since 1987 the European Union has aimed to create an internal electric utility market by imposing policies and by encouraging member states to pursue likewise practices. Therefore, countries that have been a member of the European Union since 1997, which equals the start of the sample period of this study, are also considered for the European region. These countries are: Austria, Belgium, Denmark Finland, Greece Ireland Italy, Luxembourg, Netherlands, Portugal, Spain and Sweden (European Union, 2018). To be able to measure the effect of geographic diversification, no restrictions are imposed on the country of the target firm.

Appendix 1 documents the number of acquisitions ordered by acquiring country. 77.55% of the acquisitions in the sample stem from the American region and 22.55% from the European region, meaning that the American subsample is of high importance for the overarching outcomes of this study. The United States is by far the country with the highest acquisition frequency, covering 63.73% of all acquisitions. For the European region, the United Kingdom is the country where most acquisitions are accomplished.

Table 1. Deal selection criteria and availability

Selection Criteria Deal availability

Industry Filter 18498

Country Filter 11206

Acquiring firm = publicly traded 5494

Announcement date between: 01/01/1997-31/12/2017 4190

Deal status: Completed 2932

Deal Value: ≥ 10 Mil US$ 1515

Shares acquired in transaction = 100% 871 Deal value ≥ 0.5% of the market capitalization of the acquiring firm 802 Available deals after imposed selection criteria 802

Dropped deals due to unavailability of abnormal returns 88 The final amount of used acquisition deals 714

9 Figure 1 depicts the acquisition frequency categorized by acquiring region. The high intensity of acquisition deals in the American region is particularly observed for the start of the sample period with a peak of 49 acquisitions in 1999. This peak is followed by a dramatic drop between 2001 and 2003, which is attributable to the California energy crisis. From 2004 till 2009 the amount of acquisitions is approximate proportionally distributed, and an increase in acquisition frequency is observed for the years 2010 and 2014. For Europe, a different conjecture in acquisition intensity is observed over time. A parsimonious amount of acquisitions are observed between 1997 and1999 followed by an increase between 2000 and 2002 and between 2004 and 2007. From 2007 onwards, the number of European acquisitions cascade slightly down, except for the years 2014 and 2017. The low acquisition frequency in the European market during the start of the sample period is attributable to the high extent of regulation around this time period (Domanico, 2007). The peaks in acquisition frequency are largely attributable to directives imposed by the European Parliament (Domanico, 2007).

Figure 1. Acquisition frequency ordered by the region of the acquiring firm.

Note: this figure depicts the number of acquisitions in the electric utility industry between 1997 and 2017, where the acquiring firm is located in North America or West Europe. The North American region reflects the United States and Canada. The West European region reflects Austria, Belgium, Denmark Finland, Greece Ireland Italy, Luxembourg, Netherlands, Portugal, Spain, and Sweden. The acquisition data is derived from the ThomsonOne (T1) Mergers and Acquisitions database.

3.1.2 Industry sample criteria and distribution

The primary business interest of all included acquiring firms is the electric utility industry. More specifically, and in line with Becker-Blease et al. (2008), firms whose primary business interest concerns: electric services (SIC 4911), natural gas transmission (SIC 4922), natural gas transmission and distribution (SIC 4923), natural gas distribution (SIC 4924), gas production and/or distribution (SIC 4925), electric and other services combined (SIC 4931), or gas and other services combined (SIC 4932) are considered. In line with Kishimoto et al. (2017), also firms with a primary business interest in cogeneration and alternative energy sources (SIC 499A) are included. In order to determine the effect of industry diversification, no restrictions are imposed on the primary business interest of target firms. 0 10 20 30 40 50 60 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17

10 Table 2 documents the primary business industry of the acquiring and target firms. The primary business interest of most acquiring companies concerns electric services (SIC 4911) and natural gas transmission (SIC4922) with a frequency of 40.8% and 28.0% respectively. Due to the imposed deal selection criteria, no acquiring firms are classified in the non-electric utility industry. Electric services is also the most common primary industry for target firms, with a frequency of 28.7% respectively. In total 62.6% of the target firms are classified in the electric utility industry. 11.6% of the target firms stem from SIC 1311, which represents establishments that are primarily engaged in operating oil and gas field properties, meaning that, in line with Becker-Blease et al (2008), a substantial amount of the acquirers aim to obtain direct access to natural resources as oil and gas.

3.2 Event study

According to the efficient market hypothesis, the effect of an event should be immediately reflected in the value of a firm. Therefore, an event study around the announcement date of the acquisition is appropriate to conduct, for finding the effect of acquisition announcements on stock prices. The stock prices of companies around the acquisition announcements are all retrieved from DataStream. The seminal of MacKinlay (1997) is used as the base for the conducted event study.

The effect of an acquisition announcement on stock prices is expressed in abnormal returns, which is the stock return in the event window, minus the expected (normal) return during that same day in the event window. The expected return is the return if the event does not take place. The calculation for abnormal returns is expressed in equation 1, where the abnormal return (AR) of stock I in the event period T is calculated by the actual return (R) of stock I at time T minus the expected return of stock I at time T. Xt is the conditioning information for the calculation of the expected return. For robustness reasons and in line with MacKinlay (1997) two conditioning methods are applied. These are the market model (MARK) and the constant mean return model (MEAN).

Table 2. Firms categorized by primary industry

Acquiring firm Target firm

Primary industry Number of acquisitions

% Number of acquisitions

%

Electric services (SIC 4911) 291 40.8 205 28.7 Natural gas transmission (SIC 4922) 200 28.0 86 12.0 Natural gas transmission and distribution (SIC 4923) 41 5.7 34 4.8 Natural gas distribution (SIC 4924) 43 6.0 42 5.8 Gas production and/or distribution (SIC 4925) 0 0 4 0.6 Electric and other services combined (SIC 4931) 79 11.1 19 2.7 Gas and other services combined (SIC 4932) 5 0.7 2 0.3 Alternative energy sources (SIC 499A) 55 7.7 55 7.7 Total electric utility industry 714 100 447 62.6 Other than electric utility industry 0 0 271 37.4

Total 714 100 714 100

11

ARIT=RIT-E(RIT|Xt) (1)

Equation 2 shows MARK, where the abnormal return AR of stock I at time T is calculated by the actual return R of stock I at time T minus the expected return based on the market model. RMT is the return on the market portfolio at time T. α and β are the parameters of the model based on an OLS regression, where alpha represents the intercept of the model and beta the slope of the stock with respect to the market portfolio. To reflect the market, several broad-based indices are used. In line with Kishimoto et al. (2017), the S&P 500 is used for the US market, the S&PTSX index for the Canadian market, the FTSE 100 index for the British market, the CAC 40 index for the French market and the DAX 30 index for the German market. For all residual European countries, the STOXX Europe 600 index is used.

ARIT=RIT-(α+βI*RMT) (2)

The second conditioning model, MEAN, is illustrated in equation 3, where the abnormal return AR of stock I at time T is calculated by the actual return of stock I at time T minus the average stock return µ of stock I in the estimation period.

ARIT=RIT-µI (3)

For both conditioning methods, the same estimation window and event window are applied. In line with recommendations in MacKinlay (1997), an estimation window of 120 days is chosen. Alexandridis et al. (2010) argue that a period of ten days prior to an acquisition announcement is sufficient to control for information asymmetry and information leakage. To prevent that the normal performance model is influenced by the acquisition announcement, the estimation window is determined to range from 131 days before the acquisition announcement till 11 days before the acquisition announcement.

Two different event windows are considered. Firstly, a three-day event window ranging from one day before until one day after the announcement date (CAR11). This event window is commonly used in the literature and inter alia applied in Becker-Blease et al. (2008) and Kishimoto et al. (2017). In addition, an eleven-day event window ranging from five days prior to the announcement date until five days after the announcement date (CAR55) is considered. This broader estimation window is applied to capture effects of information asymmetry and information leakage of the week prior to the event. It also reflects the value that is gradually expressed in the acquirer’s share price, for instance, due to short-term proceedings related to the acquisition announcement.

12 equation 4 for the three-day event window and in equation 5 for the eleven-day event window. Based on the two discussed conditioning models and the two discussed event windows, four different models are ultimately retrieved. These are: CAR11MARK day market model), CAR11MEAN (three-day constant mean return model), CAR55MARK (eleven-(three-day market model) and CAR55MEAN (eleven-day constant mean return model).

∑ ( ∑ (

MacKinlay (1997) argues that the advantage of applying multifactor models for event studies is generally limited. However multifactor models show higher accuracy when the sample originates from a common industry (MacKinlay, 1997), which is largely the case in this study. Therefore it is assumed that the market model is more sophisticated than the constant mean return model. In this line of reasoning, MARK is applied as the leading model and MEAN is mainly included for robustness reasons.

3.3 Measuring diversification

The effect of diversification strategy on CARs is measured by the use of dummy variables, where the diversification dummy takes the value of 1 when the acquisition is classified as diversified and the value of 0 otherwise For both geographic and industrial diversification, a narrow measure and a broad measure of diversification are applied.

For geographic diversification, the narrow measure is illustrated in equation 6.1 and implies that an acquisition is (country) focussed when the acquiring country equals the target country. An acquisition is classified as (country) diversified when the acquiring country and target country are not the same. The broad measure of geographical diversification is illustrated in equation 6.2 and implies that an acquisition is (region) focussed if the acquiring and target firm are located in the same region and (region) diversified when the acquiring and target firm are located in a different region. For this broad measure, two regions are determined. Canada and the United States represent the American region and Austria, Belgium, Denmark Finland, Greece Ireland Italy, Luxembourg, Netherlands, Portugal, Spain and Sweden represent the European region. All remaining countries are not classified into a region.

G {

( G {

13 For industry diversification, the narrow measure is illustrated in equation 7.1 and assumes that an acquisition is (SIC) focussed when the acquirer’s and target’s primary SIC code is exactly the same, and the acquisition is classified as (SIC) diversified if the acquirer’s and target’s primary SIC differ. The broad measure of industry diversification is illustrated in equation 7.2 and classifies an acquisition as (power) focussed when the primary business interest of the target firm falls into one of the electric utility SIC codes.When the target firm’s primary SIC code does not equal one of the electric utility SIC codes the acquisition is classified as (power) diversified.

I {

(

I {

(

Table 3 shows the sample distribution regarding diversification motives of American and European acquirers. In terms of geographic diversification, in 76.1% of the deals, the target firm is domiciled in the same country as the acquiring firm and in 90.9% of the deals, the target and acquiring firm are domiciled in the same region. For European acquirers, it is relatively more common to acquire a company that is domiciled in another country or region compared to their American counterparts. This difference is reasonable because, compared to the American region, the European region exists in more and smaller countries that are due to a common regulatory framework more stimulated to interconnect with each other. In addition Appendix 1 indicates that fewer acquisition opportunities exist within the West European boundaries which might encourage European firms to engage in cross-continental acquisitions. In terms of industry diversification, in 32.1% of the deals the acquiring and target firm are classified within the same primary industry and in 67.9% of the deals, the

Table 3. Diversification by region.

Diversification motive North America (N) West Europe (N) Total (N) Total (%) Geographic diversification Country-focussed acquisitions 463 80 543 76.1 Country-diversified acquisitions 90 81 171 24.0 714 100 Region-focussed acquisitions 526 123 649 90.9 Region-diversified acquisitions 27 38 65 9.1 714 100 Industry diversification SIC-focussed acquisitions 173 56 229 32.1 SIC-diversified acquisitions 380 105 485 67.9 714 100 Power-focussed acquisitions 348 99 447 62.6 Power-diversified acquisitions 205 62 267 37.4 714 100

14 acquiring and target firm are both classified within the electric utility industry. Relatively, no major difference can be found between the American and European subsample in terms of industry diversification motives.

3.4 Control variables

Three categories of control variables related to announcement returns are analyzed, these are: deal-specific control variables, firm-deal-specific control variables, and country-deal-specific control variables. Also, two time-specific controls are outlined. The deal-specific control variables are extracted from the ThomsonOne SDC Mergers and Acquisitions database, company-specific control variables are extracted from DataStream, and country-specific control variables are, except for a few indicated cases, extracted from Worldbank. Data from ThomsonOne and Worldbank are winsorized with 1% at each tail, data from DataStream are winsorized with 2.5% at each tail because more odd outliers are observed in data that originate from DataStream.

3.4.1 Deal-specific control variables

Firstly the payment method is included as deal specific control variable. In general, managers prefer to finance acquisitions with stock if they believe the stock is overvalued, and prefer to finance acquisitions with cash if they believe the stock is undervalued. Due to information asymmetry, the method of payment signals information to investors (Travlos, 1987). In this line of reasoning, it is expected that all-stock based payments result in relatively low CARs (Andrade et al., 2001, Alexandridis et al., 2010) and that all-cash based payments result in relatively high CARs (Asquith et al. 1983., Alexandridis et al., 2010). In line with Becker-Blease et al. (2008) a stock dummy

(STOCK_D) is included that takes the value 1 if the deal is 100% equity-financed and a value of 0

otherwise. Also, a cash dummy (CASH_D) is included that takes the value 1 if the deal is 100% cash-financed and a value of 0 otherwise.

Secondly, a deal size control called relative deal size (RDEALS) is included. Both positive and negative significant effects of relative deal value on announcement returns have been found over the years (Asquith et al., 1983, Becker-Blease et al., 2008, Alexandridis et al., 2010). Asquith et al. (1983) argue that relative deal size should be seen as a scaling variable of bidder returns, the higher the relative deal size the more positive or more negative bidder returns will be. The relative deal size is in line with Becker-Blease et al. (2008) measured as the deal value divided by the market capitalization of the acquiring firm at the end of the fiscal year prior to the announcement date of the acquisition.

15 stems from a higher liquidity discount for privately held targets. A dummy variable is included that takes the value 1 if the target firm is privately held and the value 0 otherwise.

Bidder competition and deal attitude are also considered but excluded as a control variable. Deal attitude is excluded because only two of the 714 deals are labeled as non-friendly takeovers. Bidder competition is excluded because only three deals had more than one bidder.

3.4.2 Firm-specific control variables

Firstly firm size is included as a firm-specific control. In general, large firms have lower acquisition returns compared to small firms. This mainly stems from the tendency of large firms to obtain fewer synergies, to pay with stock, to offer higher bidding premiums, and to make hubris-motivated decisions (Moeller et al., 2004). In line with Kishimoto et al. (2017), the natural logarithm of the total assets of the acquiring firm, the fiscal year prior to the acquisition announcement (A_LNSIZE) is used to control for firm size.

Secondly, a control for prior operating performance is included for the acquiring firm. Hence, poor managerial performance prior to an acquisition might negatively influence announcement returns (Morck et al., 1990). As proxy for prior performance the return on assets (A_ROA) and the return on equity (A_ROE) is used, calculated as the EBITDA of the fiscal year prior to the acquisition announcement divided by the total assets and repetitively by the total market capitalization of the fiscal year prior to the acquisition announcement (Kishimoto et al., 2017).

Thirdly the price to book ratio of the acquiring firm is included (A_PTB), which is according to Alexandridis et al. (2010) also a proxy for Tobin’s q. Price to book ratio is an important proxy for the valuation of acquisitions (Dong et al., 2006) but the effect on announcement returns has been ambiguous over the years. Rau and Vermaelen (1998) argue that managers of firms with a high price to book ratio tend to overestimate their abilities to manage acquisitions and are more likely to be influenced by hubris-related motives. In addition, a high price to book ratio might be a signal to investors for overvalued equity (Moeller et al., 2005), especially when the payment is stock-based (Dong et al., 2006). On the other hand, a high price-to-book ratio can be considered as a proxy for expected growth, profitability and investment opportunities (Dong et al., 2006, Becker-Blease et al., 2008). The price to book ratio is calculated by the market value of a share four weeks prior to the acquisition announcement, divided by the book value of a share four weeks prior to the acquisition announcement.

16 greater announcement returns. Cash holding is measured by the total assets of the firm in the fiscal year prior to the acquisition announcement divided by the cash and cash equivalents of the fiscal year prior to the acquisition announcement (Kishimoto et al., 2017).

Fifthly, the leverage of the acquiring firm is included (A_LEV) because in general, managers of highly leveraged firms are more monitored and incentivized by shareholders and creditors to improve firm performance, which makes it less likely for these managers to engage in bad acquisitions (Jensen, 1986). This reasoning is in line with Maloney et al. (1993) who found that highly leveraged firms are expected to have higher bidder returns. Leverage is measured as the total debt the fiscal year prior to the acquisition announcement divided by the total amount of capital the fiscal year prior the acquisition announcement.

It is also considered to control for the target’s prior firm performance, the target’s price to book ratio and the age of the acquiring firm. However, these controls are omitted from further analysis due to data scarcity.

3.4.3 Country-specific control variables

Firstly, In line with Rossi and Volpin (2004), measures are included to reflect macroeconomic conditions. Firstly, the gross domestic product per capita of the acquiring (A_LNGDP) and target firm (T_LNGDP) is included (Worldbank1, 2018), because GDP is the most widely accepted measure to reflect macroeconomic conditions. Based on the years 1997 to 2017, the average GDP is calculated for each country, and subsequently, the natural logarithm is taken of this average. In addition, a proxy for change in macroeconomic conditions is included by using the percentage GDP growth (Worldbank2, 2018), of the year of the acquisition announcement for the acquiring (A_%GDPG) and target country (T_%GDPG). No GDP growth data was available for the year 2017, therefore the percentage of GDP growth of 2016 is used for acquisitions that were announced in 2017. Lastly, the difference in macroeconomic conditions between the acquiring and target firm is included (A_GDP/T_GDP), calculated as the GDP of the acquiring country divided by the GDP of the target country.

Secondly, announcement returns are expected to be lower in countries where the market for corporate control is highly competitive (Alexandridis et al., 2010). Therefore a dummy variable is created

(COMPET_D) that takes the value 1 if the target firm stems from a highly competitive country and

the value 0 otherwise. Based on Alexandridis et al. (2010), highly competitive countries are: the United States, Canada, and the United Kingdom.

17 with respect to management (Rossi and Volpin 2004), based on a country score between zero and five. Antidirector rights are included for both the acquiring (A_SHAREHP) and target (T_SHAREHP) country. A dummy variable that takes the value 1 if the target firm originates from a common law country and 0 otherwise (COMMONL_D) is used as a legal proxy. The antidirector scores and common law origin of countries are retrieved from Porta et al. (1998).

Lastly, in line with Frésard et al. (2017), a control is included for acquiring and target countries that speak the same primary language. A dummy (Language_D) is included that takes the value of 1 if the primary spoken language of the target and acquiring country is the same and a value of 0 otherwise. Primary languages are extracted from the ThomsonOne(T1) Mergers and Acquisitions database.

3.4.4 Time-specific control variables

Two time dummies are constructed to control for major economic events. To control for the California energy crisis a dummy (CALIF_D) is included that takes the value of 1 if the target stems from the US and when the announcement year is 2001. The dummy takes a value of 0 otherwise. In addition, a dummy is included to control for the financial crisis (CRISIS_D) which takes a value of 1 if the announcement year is 2008 or 2009 and a value of 0 otherwise.

3.5 Omitted variables

To test for multicollinearity between independent variables, a Pearson correlation matrix is made. If variables suffer from multicollinearity, they can be predicted from each other with a certain degree of accuracy, which affects the predictive power and sensitivity of the model. For omitting variables, a multicollinearity threshold of 0.7 is applied. The Pearson correlation matrix can be found in Appendix 2. Ultimately, four variables are omitted, these are: T_LNGDP, A_%GDPG, T_SHAREHP, and COMP_D.

T_LNGDP and A_%GDPG are omitted because the variables correlate highly with A_GDP/TGDP and T_%GDPG respectively. This high correlation is attributable to the number of domestic acquisitions in the sample. Thereby the high correlation is reasonable because the two omitted variables represent just as the two non-omitted variables a measure of macroeconomic conditions and economic growth. The variables T_SHAREHP and COMP_D show a correlation higher than the threshold with each other and with COMMONL_D. This is sequacious because the level of shareholder protection and the extent of competitiveness of the market for corporate control are usually high in common law countries. The omitted control variables are used as robustness and can be found in Appendix 3.

3.6 Empirical method

18 regression equations, the dependent variable is the eleven-day CAR based on the market model (CAR55MARK). The control variables that are not omitted as result of the multicollinearity test are used as regressors and εit represents the error term. Equation 8a does not take fixed effects into consideration. Equation 8.2 shows three separate regressions: firstly with time fixed effects, secondly country fixed effects and thirdly industry fixed effects. This is done to determine whether CAR55MARK is influenced by fixed time, country and industry determinants.

To assure that the OLS regression is the best linear unbiased estimator the Gauss Markov assumptions as outlined in Brooks (2010) are considered. The error term is found to be heteroskedastic based on a Breusch-Pagan / Cook-Weisberg heteroskedasticity test and based on a White heteroskedasticity test (Breusch and Pagan, 1979, White, 1980, Cook and Weisberg, 1982). Consequently, robust standard errors are used for the regressions shown in equation 8.1 and equation 8.2.

CAR55MARK= α +β STO K_D + β SH_D + β3P IV TE_D + β DE LS + β _LN SIZE + β A_ROA + β _ OE + β8 _PTB + β9A_ SHH + β _LEV + β _LNGDP + β A_GDP/T_GDP + β 3T_ %GDPG + β _SH EHP + β L NGU GE_D + β16COMMONL_D + β ISIS_D + β 8CALIF_D + ε (8.1) M K α +β STO K_D + β SH_D + β3P IV TE_D + β DE LS + β _LN SIZE + β _ O + β _ OE + β8 _PTB + β9 _ SHH + β _LEV + β _LNGDP + β _GDP/T_GDP + β 3T_ %GDPG

+ β _SH EHP + β L NGU GE_D + β OMMONL_D + β ISIS_D + β 8 LIF_D + 1 ∑

fixed effects| 2 ∑ xed effects | 3 ∑ x 3 + ε (8.2)

Also several mean comparison t-tests are conducted in this study to determine if the found CARs significantly differ from zero as illustrated in equation 9.1 and to determine if CARs significantly differ between subsamples as illustrated in equation 9.2.

H0: µ1=0; H : µ (9.1) H0: µ1- µ2=0; Ha: µ1- µ2 (9.2)

4. Results

4.1 Descriptive statistics and regression results

19 dramatic drop in A_PTB is observed for the years 2002, 2009 and 2013 as a consequence of economic turndown in the industry. The A_GDP/T_GDP measure is higher than one. This implies in line with Frésard et al. (2017) that bidders have the tendency to acquire firms in countries that are less developed than their home country. The average score on shareholder protection for both the acquiring and target countries is higher than 4.5, which is attributable to the high proportion of common law countries in the sample. Lastly, 4.1% of the acquisitions are done by US firms during the California electricity crisis and 8.4% of the deals are conducted during the financial crisis.

Table 4 also illustrates the coefficients and the corresponding t-statistics of the control variables with regard to the found CARs. The stock dummy is the only deal specific control variable that significantly influences CAR55MARK. A positive coefficient of 0.018* is derived, meaning that if the deal is 100% stock-financed, CARs are expected to be 1.8% higher compared with deals that are not 100% stock-financed. This positive association between stock-financed acquisitions and CARs contradict generally accepted findings of Travlos (1987) and Alexandridis et al. (2010). Nevertheless, Eckbo and Thorburn (2000) also show that all-stock payments result in higher bidder returns, and argue that for some countries the extent of adverse selection is low, which mitigates the information asymmetry effect as described in Travlos (1987) and Alexandridis et al. (2010). An alternative

Table 4. Descriptive statistics and OLS regression of the control variables (CARR55MARK dependent variable)

Variable Observations Mean Std. Dev. Min Max Coefficient T-statistic Deal specific control variables

STOCK_D 714 0.077 0.267 0 1 _{0.018 1.70*}

CASH_D 714 0.239 0.427 0 1 _{0.004 0.81}

PRIVATE_D 714 0.349 0.477 0 1 0.005 1.06

RDEALS 702 0.410 0.738 0.007 4.028 0.006 0.96

Firm-specific control variables

A_LNSIZE 702 21.536 1.720 17.367 24.400 -0.002 -1.10 A_ROA 675 0.097 0.047 -0.058 0.188 0.093 1.25 A_ROE 690 0.186 0.123 -0.051 0.550 -0.008 -0.27 A_PTB 672 2.339 1.560 0.67 8.940 0.000 0.14 A_CASHH 677 0.053 0.072 0 0.352 0.084 1.73* A_LEV 700 0.475 0.197 0 0.829 0.000 -0.01

Country-specific control variables

A_LNGDP 714 10.620 0.148 10.121 10.712 0.041 1.82* A_GDP/T_GDP 713 1.0826 0.640 0.654 6.244 -0.005 -1.10 T-%GDPG 713 0.024 0.017 -0.029 0.051 -0.185 -0.86 A_SHAREHP 714 4.576 1.163 0 5 _{0.003 0.96} LANGUAGE_D 714 0.906 0.292 0 1 _{-0.003 -0.29} COMMONL_D 714 0.854 0.353 0 1 -0.031 -2.47**

Time-specific control variables

CRISIS_D 714 0.084 0.278 0 1 -0.028 -2.21** CALIF_D 714 0.041 0.198 0 1 -0.030 -2.21**

20 explanation for the higher return of stock-financed deals corresponds with the found A_PTB which is considerably high and shows periods of prolonged growth. Such A_PTB characteristics sign towards cycles where the market is overheated, where stock prices are expected to rise, and where cash would likely be invested in stocks. Hence, in an overheated market, all-stock based payments might generate higher CARs because it can be considered by the stock market as the most efficient payment consideration.

The cash holding of the acquiring firm positively affects CARs, a coefficient of 0.084* is found. This is in contrast with the free cash flow theory of Jensen (1986) and the corresponding findings of Harford (1999) that cash-rich firms make worse acquisition decisions. The positive coefficient is in line with Kishimoto et al. (2017) and shows that the free cash flow problem might not be severe in the deregulated electric utility industry. Besides, the industry fixed effects model in Appendix 5 shows that firm size negatively influences CARs. In line with Moeller et al. (2004), it is found that small firms generate significantly higher CARs then their larger counterparts.

Country variables also influence CARs. A positive coefficient for A_LNGDP is found. This entails that CARs are expected to increase when the level of development of the country where the acquiring firm stems from increases. Lower CARs of 3.1%** are expected if the target firm is domiciled in a common law country. This lower return is in line with findings in Alexandridis et al. (2010) who argue that in common law countries, the market for corporate control is highly competitive, which consequently urges acquiring companies to pay high takeover premiums. Appendix 5 corroborates the view that the acquisition of targets that stem from common law countries results in lower CARs because, T_SHAREP, a variable that highly correlates with COMMONL_D, is also negatively associated with CAR55MARK.

Lastly, acquisitions where the target firm stems from the North American region during the California energy crisis, resulted on average in 3.0%** lower CARs. Additionally lower CARs of 2.8%** are observed for announcements in 2008 and 2009. These findings are in line with Kishimoto et al. (2017) and in line with the negative impact of the California energy crisis and the economic crisis on returns in the electric utility industry.

4.2 Announcement returns

21 be due to information asymmetry and information leakage the week prior to the event and due to value that is gradually expressed in the acquirer’s share price, for instance, due to proceedings related to the acquisition announcement.

Hypothesis 1A and 1B raise the question if acquisitions in the American and European electric utility sector are value enhancing or value decreasing. Based on the positive means as illustrated for both event periods and for both conditioning models a tendency might arise to assume that acquisitions in the electric utility sector increase shareholder value. However, based on the calculated t-statistics, none of the average CARs differ significantly from zero. Consequently, neither hypothesis 1A nor hypothesis 1B is accepted.

The documented CARs are partly in line with the findings of Kishimoto et al. (2017), who also document positive CARs, however, their reported CARs are significantly higher than zero. Leggio and Lien (2000) and Becker-Blease et al. (2008) report significant negative bidder returns which contradict my findings. This disparity might be due to a major difference in sample criteria. The sample of Leggio and Lien (2000) and Becker-Blease et al. (2008) only exists of US acquirers and represents the deregulated period before 1996 and 2001 respectively. In order to make a more representative comparison with Leggio and Lien (2000) and Becker-Blease et al. (2008), a US subsample before 2001 and after 2002 is created. Table 6 illustrates the two US subsamples based on CAR11MARK. CARR11MARK is chosen as the independent variable because this measure corresponds the most with the applied proxy for CARs in Leggio and Lien (2000) and Becker-Blease et al. (2008). The years 2001 and 2002 are in line with Kishimoto et al. (2017) omitted for this comparison to circumvent the effect of the California energy crisis.

The US subsample before 2001 does not show major differences in CARs compared with the subsample after 2002. Slightly positive but insignificant CARs are observed for the US sample before 2001. This implies that findings before 2001 are just as the general sample not in line with the findings in Leggio and Lien (2000) and Becker-Blease et al. (2008). This discrepancy might be Table 5. Cumulative abnormal returns

Applied model N. Obs. Mean T-Stat. St. Dev Min Max

CAR11MARK 714 0.094% 0.603 4.15% -10.282% 11.619%

CAR11MEAN 714 0.073% 0.444 4.382 -10.862% 12.439%

CAR55MARK 714 0.197% 0.851 6.185% -13.220% 17.435%

CAR55MEAN 714 0.267% 1.085 6.573% -14.502% 17.308%

22 attributed to the sample of Leggio and Lien (2000) and Becker-Blease et al. (2008) that includes merger deals which might result in different bidder returns compared to my sample that purely consists of acquisitions. Besides, the applied sample period in Leggio and Lien (2000) and Becker-Blease et al. (2008) reaches further back in time than 1997. Kishimoto et al. (2017) elaborate that M&A was not a major strategic concern in the regulated industry and that managers had limited M&A knowledge just after the sector liberalized which made them during the start of the sample period of Leggio and Lien (2000) and Becker-Blease et al. (2008) unable to engage in efficient M&A deals.

4.3 Diversification and announcement returns

Besides the CARs in general, the effect of diversification strategy is analyzed. Hypothesis 2 questions if country and region-focussed acquisition generate different CARs compared with country and region-diversified acquisitions. In the same line, hypothesis 3 questions if SIC and power-focussed acquisitions generate different CARs compared with SIC and power-diversified acquisitions. Table 7 illustrates the effect of geographical and industrial diversification on CARs.

4.3.1 Geographic diversification and announcement returns

In terms of geographic diversification, mixed results are observed between models. Based on CAR11MARK, CAR55MARK and CAR55MEAN is assumed that country-focussed deals generate higher CARs than country-diversified deals, which is in opposition to the results of CAR11MEAN. Based on CAR55MARK, country-focussed deals are expected to generate an average CAR of 0.328% and country-diversified deals are expected to generate a CAR of 0.074%. More mixed results are found for region-focussed and region-diversified deals. Based on CAR11MARK, CAR11MEAN, and CAR55MEAN it is assumed that region-diversified deals create higher CARs than region-focussed deals, which is in contrast to the results of CAR55MARK. Based on CARMARK55 and thus in contrast to other models, region-focussed deals are expected to generate on average a CAR of 0.210% and region-diversified deals a CAR of 0.068%.

Table 6. General CARs and CARs by diversification strategy for two US subsamples (CAR11MARK dependent variable)

Entire Sample Country-focussed Region-focussed Country- diversified Region-diversified SIC-focussed Power-focussed SIC-diversified Power-Diversified The United States before 2001

Number 131 119 123 12 8 30 79 101 52

Mean 0.121% 0.099% 0.011% 0.342% 1.823% -1.365% -0.643% 0.563% 1.283% Tstat. 0.301 0.239 0.026 0.208 1.044 -1.620* 0.013 1.246 2.322**

The United States after 2002

N 282 270 278 12 4 94 170 188 112

Mean 0.108% 0.099% 0.088% 0.301% 1.508% 0.697% 0.525% -0.187% -0.525% Tstat 0.410 0.364 0.329 0.355 1.161 1.443* 1.456* -0.599 -1.424*

23 Hypothesis 2 questions if geographic-focussed deals create more or less shareholder value compared with geographic-diversified deals. Based on the first t-statistics in table 7 no evidence is found that a specific geographic diversification strategy generates significant shareholder value. Thereby, the second t-statistics in table 7 show no significance, meaning that no proof is found that geographic-focussed and diversified deals outperform each other. Therefore neither hypothesis 2a nor hypothesis 2b is accepted.

The finding that the stock market responds indifferntly to geographic diversification strategies corresponds with the findings of Kishimoto et al. (2017) who report negligible differences in value creation between domestic and cross-border acquisitions. Becker-Blease et al. (2008) report significant disadvantages of geographic diversification. Table 6 shows that US firms sporadic diversify geographically and that geographic diversification tends to result in higher CARs compared with geographic-focussed deals, which also contradicts with Becker-Blease et al (2008). The inconclusive findings of the aggregate sample are in line with repeatedly cited work of inter alia Ferris et al. (2010) who show an indifferent effect of geographic diversification on CARs and with findings of Doukas and Travlos (1988) and Morck and Yeung (1991) who show that geographic diversification by itself does not generate shareholder value.

Table 7. CARs by geographic and industry diversification.

Country- focussed Region- focussed Country-diversified Region-diversified SIC-focussed Power- focussed SIC-diversified Power- diversified CAR11MARK N 543 649 171 65 229 447 485 267 Mean 0.103 0.087% 0.063% 0.158% 0.346% 0.250% -.0256% -0.169% Tstat1 0.580 0.532 0.198 0.324 1.218 1.242 -0.138 -0.698 Tstat2 -0.132 -0.132 1.118 1.306* CAR11MEAN Mean 0.047% 0.067% 0.152% 0.123% 0.264% 0.225% -0.018% -0.182% Tstat1 0.256 0.393 0.445 0.231 0.873 1.048 -0.090 -0.725 Tstat2 -0.096 -0.096 0.802 1.201 CAR55MARK Mean 0.314% 0.210% -0.175% 0.068% 0.503% 0.314% 0.053% 0.002% Tstat1 1.2121 0.8714 -0.3440 0.0820 1.2187 1.059 0.189 0.004 Tstat2 0.1763 0.1763 0.907 0.652 CAR55MEAN Mean 0.328% 0.250% 0.074% 0.432% 0.572% 0.440% 0.123% -0.022% Tstat1 1.199 0.970 0.135 0.533 1.263 1.384* 0.420 -0.057 Tstat2 0.440 -0.213 0.853 0.908

24

4.3.2 Industry diversification and announcement returns

For industry diversification, the documented announcement returns are more in line between event periods and conditioning models. Based on all four models, SIC-focussed acquisitions generate higher CARs than diversified acquisitions. The three-day event period even illustrates that SIC-diversified deals result in negative CARs. Based on CAR55MARK SIC-focussed deals generate an average CAR of 0.503% while SIC-diversified deals generate an average CAR of 0.053%. The broader measure of industry diversification shows corresponding results. All four models show that power-focussed deals generate a higher average CAR than power-diversified deals. Moreover, all models except for CAR55MARK document a negative average CAR for power-diversified deals. Based on CAR55MARK focussed deals generate on average a CAR of 0.314% and power-diversified deals generate on average a CAR of 0.002%.

Hypothesis 3 questions if industry-focussed deals create more or less shareholder value compared to industry-diversified deals. CAR55MEAN shows based on a corresponding t-statistic of 1.384* shown in Table 7 that power-focussed acquisitions in the deregulated American and European region generate significant shareholder value. The second t-statistic measures if industry diversification strategies outperform each other. Based on a second t-statistics of 1.306* as shown at CAR11MARK it is assumed that power-focussed deals create more shareholder value than power-diversified deals. In this line of reasoning hypothesis 3a is accepted and therefore hypothesis 3b is automatically rejected.

The acceptance of hypothesis 3a initially contradicts with findings of Becker-Blease et al. (2008) who in contrast report that industry-diversified deals generate higher CARs than industry-focussed deals. However, when my US subsample in Table 6 is considered, a sample that highly corresponds with the sample of Becker-Blease et al. (2008), matching results are found. The initial contradiction in results stems from a discrepancy in the effect of industry diversification for US acquirers between the period before 2001 and the period after 2002. My US subsample before 2001 illustrates in line with Becker-Blease et al. (2008) that industry-diversified deals generate higher CARs than industry-focussed deals. My US subsample after 2002 corresponds with the aggregate sample. This discrepancy in the effect industry diversification for US firms between time periods is largely attributable to the California electricity crisis (Kishimoto et al., 2017).

25 industry. Besides, the findings correspond with repeatedly cited work of for instance Morck et al. (1990), Lang and Stulz (1994), DeLong (2001), and Dos Santos et al. (2008) who show that the stock market reacts more positively to industry-focussed deals compared with industry-diversified deals. There are many reasons why the stock market believes that industry-diversified acquisitions are inefficient investments. Jensen (1986) argues that this stems from agency costs, Berger and Ofek, (1995) attribute this to overinvestment and cross-subsidization.

4.4 North American and West European regions

The previous sections considered the Amercian and European electric utility sector as one aggregate region, however, there are many reasons to assume that announcement returns and the effect of diversification differ by the region where the acquiring firm is established.

4.4.1 North America, West Europe, and announcement returns

Table 8 shows the average CAR for the American and European region separately. For the American region, CAR11MARK, CAR11MEAN, and CAR55MARK show slightly negative CARs while CAR55MEAN shows a slightly positive average CAR. Based on CAR55MARK it is concluded that acquisitions in the North American electric utility industry generate on average a CAR of -0.02%. Based on the corresponding first t-statistics, none of the average CARs significantly differ from zero. Therefore, no evidence is found that acquisitions stemming from the American electric utility industry are value enhancing or value decreasing. For the European region, all documented mean CARs are positive with the eleven-day event window illustrating higher returns than the three-day event window. CAR55MARK shows that acquisitions in the European electric utility industry generate on average a CAR of 0.94%. In addition, based on the first t-statistic of CAR11MARK, CAR55MARK and CAR55MEAN it is found that acquisitions in the European electric utility sector significantly increase shareholder value. Because of these findings, more substance is given to the rejection of hypothesis 1A and hypothesis 1B, which initially entailed that acquisitions in the electric utility Table 8. CARs by region

N.obs. Mean T-stat1 St. Dev Min Max T-stat2

North America CAR11MARK 553 -0.003% - 0.014 4.239% -10.282% 11.619% -1.149 CAR11MEAN 553 -0.027% - 0.142 4.427% -10.862% 12.439% -1.125 CAR55MARK 553 -0.020% - 0.074 6.262% -13.220% 17.435% -1.737** CAR55MEAN 553 0.112% 0.403 6.527% -14.502% 17.308% -1.169 West Europe CAR11MARK 161 0.424% 1.413* 3.807% -10.282% 11.619% CAR11MEAN 161 0.415% 1.248 4.216% -10.862% 12.439% CAR55MARK 161 0.941% 2.034** 5.872% -13.220% 17.435% CAR55MEAN 161 0.800% 1.510 * 6.722% -14.502% 17.308%

26 industry neither enhance nor destroy shareholder value. Based on table 8 it can be concluded that this rejection of hypothesis 1A and 1B only counts for the American region. If the European region is considered individually, hypothesis 1A can be accepted meaning that acquisitions stemming from European acquirers do create shareholder value.

Hypothesis 4 questions if acquisitions, where the acquirer stems from the North American region, generate more or less shareholder value than acquisitions where the acquirer stems from the West European region. The previous paragraph showed that acquisitions with American acquirers are neither value enhancing nor value decreasing while acquisitions with European acquirers significantly generate shareholder value. The row on the right-hand side of Table 8 shows the second t-statistic which corresponds with the difference in mean CAR between both regions. Based on CAR55MARK and a corresponding T-statistic of -.1737**, it is concluded that acquisitions stemming from the European region generate significantly higher CARs compared with acquisitions stemming from the American region. In this line of reasoning, hypothesis 4B is accepted and therefore hypothesis 4a is automatically rejected. The found CARs for the American and European subsample correspond largely with findings of Kishimoto et al. (2017), who report that acquisitions by European acquirers generate significant shareholder value and generate higher CARs than American acquirers. However, Kishimoto et al. (2017) report that American acquirers also generate significant shareholder value, while this is not reflected in this research.

Three reasons for the difference in CARs between the American and European region are documented. Firstly, Appendix 6a and Appendix 6b show that 98% of the targets of American acquirers stem from a common law country, for European bidders, this is only 41%. Table 4 significanlty shows that CARs are expected to be 3.1% lower when a common law target is acquired. Markides and Oyon (1998) and Alexandridis et al. (2010) corroborate this view and show that bidder CARs are lower for samples such as the American subsample where many target firms have a common law origin. These lower returns stem from the highly competitive market for corporate control in common law counties which consequently urges acquiring companies to pay high takeover premiums.

27 implied effect as shown in Moeller et al. (2004), Officer (2007) and Frésard et al. (2017) does not hold in the OLS regression as illustrated in Table 4.

Thirdly, CARs might differ for American and Europen acquirers because for both regions the deregulation of the electric utility industry took place over a different period of time. Leggio and Lien (2000) illustrate that most lucrative acquisition candidates are quickly acquired after the sector deregulates, which consequently causes that limited synergies are available for the following years. In the US deregulation started in 1992, therefore many valuable acquisitions stemming from American acquirers are not covered in this study. For most European countries deregulation took place in a later stadium, this entails that many of the by Leggio and Lien (2000) called lucrative acquisitions are considered for European acquirers, which possitvely contributes to the found average CAR of the Euoropean subsample compared to the American subsample.

Besides, Table 9 provides, based on CAR11MARK and CAR55MARK a more detailed view of the contribution of each acquiring country towards the documented average CAR for the two regions. Canadian acquirers document the lowest CAR of -1.0% respectively (CAR55MARK), while US acquirers generate on average a positive CAR of 0.2% (CAR55MARK). In this sense, the low CARs of the American region, compared to the European region is for a major part attributable to Canadian acquiring firms. However, it should be taken into consideration that acquiring firms of both Canada and the US document substantial lower CARs than the European average. Moreover, 82.3% of the American sample stems from firms domiciled in the US, which causes that the US sample has a big influence of the average CAR of the American region. For the European region based on CAR55MARK only a negative average CAR is documented for Italian acquirers, however when CAR11MARK is taken into consideration, also deals where the acquiring firm originates from the United Kingdom demonstrate a negative CAR. Deals where the acquiring firm stems from France, Spain, or the rest of Europe show a CAR of 1.8%** and 1.9%** (CAR55MARK) and 3.2% *** (CAR11MARK) respectively, and contribute therefore positively to the European average.

Table 9. CARs by acquiring nation

US Canada UK France Germany Italy Spain R. Europe

N. obs 455 98 58 14 20 26 23 14 CAR11MARK Mean 0.072% -0.350% -0.341% 2.339% 0.199% -0.123% 0.622% 2.856% T-statistic 0.357 -0.899 -0.649 1.720* 0.242 -0.206 1.006 3.2333*** CAR55MARK Mean 0.195% -1.016% 0.163% 4.119% 1.350% -0.221% 1.998% 0.732% T-statistic 0.666 -1.607 0.221 1.804** 1.291 -0.213 1.916** 0.383

28

4.4.2 North America, West Europe, and diversification

In addition to the difference in CARs between the American and European region, Table 10 shows how diversification strategy influences CARs for both regions. For the American sample, country-diversified deals result in a significant negative average CAR of -1.1%* and the second t-statistic of 1.740** shows in line with Moeller and Schlingemann (2005) and Kishimoto et al. (2017) that for American bidders country-focussed deals generate higher CARs than country-diversified deals. Remarkable is that this effect of geographic diversification is not shown for region-focussed and region-diversified deals, suggesting that for American acquirers the form of geographic diversification matters. This finding is in line with Markides and Oyon (1998) who show that US acquirers create less value if they target Canadese firms compared to European firms. For the European region, the found average CAR is only significant if the acquisition is country or region-focussed and the average CAR is even negative if the deal is region-diversified. However, based on the second t-statistics no significant difference in value creation is found between geographic-focussed and diversified deals. Due to these findings, more substance is given to the rejection of hypothesis 2A and hypothesis 2B, which initially entailed that the applied geographic diversification strategy does not influence CARs. Based on Table 10 can be concluded that hypothesis 2a can be accepted for the North American region when the narrow measure of geographic diversification is applied, meaning that country-focussed deals outperform country-diversified deals for American acquirers.

Table 10. CARs categorized by diversification strategy and acquiring region

Country- focussed Region- focussed Country- diversified Region-diversified SIC- focussed Power- focussed SIC- diversified Power- Diversified North America CAR11MARK Number 463 526 90 27 173 348 380 205 Mean 0.070% -0.046% -0.375% 0.852% 0.165% 0.073% -0.079% -0.130% Tstat1 0.355 -0.251 -0.831 1.081 0.495 0.310 -0.369 -0.464 Tstat2 0.910 -1.075 0.628 0.543 CAR55MARK Mean 0.184% -0.035% -1.069% 0.279% 0.225% 0.116% -0.131% -0.249% Tstat1 0.640 -0.128 -1.555* 0.238 0.471 0.339 -0.408 -0.588 Tstat2 1.740** -0.254 0.6191 0.6623 West Europe CAR11MARK Number 80 123 81 38 56 99 105 62 Mean 0.297% 0.658% 0.549% -0.334% 0.904% 0.874% 0.168% -0.295% Tstat1 0.730 1.925** 1.242 -0.542 1.706** 2.284** 0.463 -0.624 Tstat2 -0.4188 1.409* 1.171 1.912** CAR55MARK Mean 1.065% 1.257% 0.819% -0.082% 1.361% 1.009% 0.717% 0.832% Tstat1 1.910** 2.578*** 1.107 -0.071 1.676** 1.714** 1.272 1.104 Tstat2 0.264 1.231 0.6609 0.1862