Abnormal returns on M&As of Banks between 2007 and 2015 in the US and the factors that affect them

(1)

1

University of Amsterdam

Master Thesis

Abnormal returns on M&As of Banks between 2007

and 2015 in the US and the factors that affect them

Author: Supervisor:

Skoufos Ioannis Anastasios Dr. Liang Zou

A thesis submitted in fulfillment of the requirements for the degree of

MSc Business Economics: Finance Track in the Finance Group,

Amsterdam Business School

(2)

2

Statement of Originality

This document is written by Skoufos Ioannis Anastasios who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the

(3)

3

Acknowledgements

(4)

4

To my parents, Thanasis and Zoi.

(5)

5

Abstract

This research dives into M&As of banks that occurred in US between 2007 and 2015. In order to gauge the abnormal returns for acquirers and targets event study methodology is deployed and more specifically the market adjusted return model. The findings of this methodology shows that target shareholders enjoy significantly positive abnormal returns while bidder shareholders enjoy positive abnormal returns but not as substantial as target shareholders. Moreover, this thesis illustrates the effect of the size, the leverage, the profitability, the price to earnings ratio, the market to book ratio of the target firms on their cumulative abnormal returns. Finally the regression that was used for this scope showed that the size of the target and the market to book ratio, significantly affect target’s cumulative abnormal returns.

(6)

6

List of figures and tables

Table 2.1 Descriptive overview of event study results……….17

Table 2.1 continued………..18

Figure 3.1 US Public banks involved in Mergers and Acquisitions during 2007-2015………..….26

Table 4.1 M&A announcement day abnormal returns for targets……….34

Table 4.2 M&A announcement day abnormal returns for acquirers………35

Figure 4.1: Average abnormal returns during the event window (-10,+10)………...36

Figure 4.2: Cumulative average abnormal returns during the event window (-10,+10)………..37

Table 4.3 Summary statistics………38

Table 4.4 Regression………... …….40

(8)

8

1. Introduction

The recent financial crisis has had a wide-reaching influence on the economy. From 2007 up to the end of 2009 the volume of M&As was shrinking every year. During 2010, the amount of deals augmented at about 9% related to 2009 and the deals value increased at about 48%. The years after, there was a growing amount of consolidations at an accelerating rate from year to year.

The current thesis deals with M&As of banks that occurred between 2007 and 2015 in US, mainly the listed firms on stock exchange, and aims to study the existence of abnormal returns for the acquirers and the targets. To inspect whether there are abnormal returns or not, a certain methodology is used to examine abnormal activity in the stock market. Much research has been done concerning M&As and abnormal returns for the US market in the past. In many studies that have been carried out already, researchers have employed different types of methodologies to evaluate M&A event’s performance. The most known categories of these methodologies are event studies, efficiency studies and performance studies. In our case, event study was used and more specifically the market adjusted return model. The data of this study were retrieved from Thomson One and Datastream while data processing was made through Excel.

As literature has indicated so far the targets earn large, positive, and statistically significant abnormal returns. Overall, the data suggests that the average abnormal return target shareholders expect is between 10% and 15% something that comes in absolute agreement with the results of this research since the average abnormal return on the announcement day was found 12.46% and statistically significant at 1% level while all the event windows that were selected, were statistical significant and had large positive cumulative average abnormal return.

As far as the literature for the acquirers is concerned, the results are more mixed. There are studies that claim that acquirers destroy shareholder value when consolidation occurs (e.g. Hannan and Wolken, 1989; Baradwaj et al., 1990; Houston and Ryngaert, 1994; Kane, 2000; Cornett et al., 2003; Gupta and Misra 2007). On the other hand, there are studies which found that acquirers gained positive abnormal returns (James and Wier, 1987; Dubofsky and Fraser, 1989; Seidel 1995 and Kiymaz, 2004). The current thesis is mostly categorized in these studies since on the announcement day and one day after the announcement of the M&A, the abnormal returns for the acquirer are positive 0.21% and 0.23%. The cumulative average abnormal returns were also positive in all the event windows that were studied and most of the times statistical significant at

(9)

9 5% and 10% level.

The second part of this thesis is dedicated to explain which factors affect target’s abnormal returns. To achieve this, an implementation of a regression analysis was needed, to find out what factors affect the cumulative abnormal returns (CAR) around the M&A announcement. The reason behind the focus of regression on the target firm is that the positive abnormal return and the results on the bidder’s side are more ambiguous since they are neither as great nor as significant as target’s abnormal returns. The regression analysis was performed in Stata 13.

Previous literature has presented that the most significant factors that affect target’s abnormal returns are target’s growth rate, its size (Kane, 2000; De Long, 2001; Campa and Hernando et al, 2006; Gupta and Misra, 2007) and its market performance (Beitel et al. 2004, Lorenz et al 2006, Ismail and Davidson 2007). In addition, it has been shown that abnormal returns tend to be more positive when post-merger cost of debt is reduced (Penas and Unal, 2004). Another factor that influences the success of an M&A according to the literature is the method of payment. Opinions differ, since some have found that M&As financed with stock have lower returns than those financed with cash or a mixture of cash and stock (Baradwaj et al, 1991; Becher, 2000; Ismail and Davidson, 2007; Al-Sharkas and Hassan, 2010) while others claim absolutely the opposite (Choi et al, 2010). Efficiency of the target was also considered a crucial factor according to Akhavein et al (1997), Beitel et al (2004) and Beccali and Frantz (2009).

This study explored the most significant factors that may affect the target’s cumulative abnormal return as indicated in the literature. Some of them were indeed significant for the cumulative abnormal returns while others were not. More specifically, in contrast with the literature, target’s profitability was not found to affect target’s cumulative abnormal returns as long as the coefficient was not statistically significant. Moreover, by adding price to earnings ratio on the regression, which is the profitability of the targets in regard to their stock valuation, was not found to influence cumulative abnormal returns of the targets. In another conflict with the literature, the leverage of the target was not found to influence the success of an M&A for the targets since total debt to total equity variable was not found to be statistically significant. Furthermore, market to book ratio variable was found statistically significant at 1% significance level which means that target’s investment opportunity influence its cumulative abnormal return after the M&A. In addition, this study came in total agreement with the literature which states that the size of the target affect the M&A success and more specifically target’s cumulative abnormal

(10)

10 return, which was proved implementing the market value variable, which was found to be statistically significant at 1% significance level.

The following sections of this paper review the literature, which is presented in section 2, in order to get a good understanding of the issue. Section 3 provides data sources and presents methodology both for the event study and for the regression. In section 4, the analysis of the results is presented while section 5 discusses the implications of the results and provides concluding remarks.

(11)

11

2. Literature review

2.1 Definitions of M&As

Nakamura (2005) highlighted that using a broad definition of M&A could lead to confusion and misunderstanding as it implies everything from simple mergers to strategic alliance. A merger is an amalgamation of two or more entities to form a new company (Gaughan, 2002 and Jagersma, 2005), while an acquisition is the purchase of shares or assets on the target company to put over a managerial influence (Chunlai et al., 2003), not necessarily by mutual agreement (Jagersma, 2005). However, due to the fact that their meaning is almost the same, for research purposes, the terms ‘merger’, ‘acquisition’, ‘takeover’ and ‘consolidation’ can be used broadly. (Jarrel et al, 1988; Berkovitch and Khanna, 1991).

2.2 Trends of M&As

The first official implemented M&A in the US was between, Berks County Trust Company which was the acquirer and Schuylkill Valley Bank which was the target, on 1901 and from then countless M&As have been accomplished. The mitigation of the existing banks in US in an accelerating matter indicates that there is a momentum of consolidations at a hastening rate from year to year. Several key factors have been identified that can explain the recent fast pace of M&As among whom the most important are technological progress, improvements in financial condition, excess capacity or financial distress, international consolidation of markets and deregulation (Berger et al., 1999).

2.3 Motives for M&As

According to literature there are value and non-value maximizing motives for financial consolidation (Halpern, 1983).

a. Value maximization motive

The most important motive for M&As is maximization value, which means that the reason behind the consolidation is the maximization of value of the firm. There are two ways for banks to achieve that through consolidation. First, by increasing their market power in setting prices and second by increasing their operational efficiency. The most significant value motives are:

(12)

12

Economies of scale and scope

Perhaps the most commonly quoted source of potential gains from M&As is the exploitation of economies of scale. Most of the researches study on the existence of scale economies in retail commercial banking presenting a relatively flat U-shaped average cost curve, with a minimum approximately below $10 billion of assets, depending on the sample, country and time period analyzed. Berger et al. (1993) supports that the minimum optimal size for the US seems to be below half a billion dollars of assets for researches conducted on samples containing mainly small banks. As for studies conducted on samples with mostly large banks or that also consider risk as one of the relevant factors, the size was estimated to be between two and ten billion dollars of assets. As an outcome it is considered that smaller banks operate with a different technology than larger ones and that economies of scale are exhausted by larger banks at an early stage. It is also suggested that efficiency gains from the exploitation of scale economies disappear once a specific size is reached and that there might be diseconomies of scale above it, probably due to the complexity of managing large organizations or the difficulties that arise when a bank’s geographical coverage increases. However, this theory is disputed since it relies on data from the 1980s and early 1990s and from firms mostly below the size of the average bank in many countries. It might also have to be revised due to recent technological improvements that imply large fixed costs and thus have the potential for economies of scale even for larger banks.

Presumably the second most quoted reason for M&As is the exploitation of synergies, or economies of scope. Measuring the existence and extent of economies of scope is difficult given that, in theory, the benchmark should consist of single-product firms. The lack of such firms casts doubts on the reliability of results in this particular field. However, a research using American data has found little or no revenue scope economies between bank deposits and loans. Other research, has suggested that financial conglomerates are more revenue efficient than specialized institutions and that universal banks appear to be more cost and profit efficient than non-universal banks (Berger et al., 1999).

(13)

13

Revenue enhancement

This motive assumes that the company will be absorbing a major competitor and thus increase its power to set prices. Spiegel and Gart (1996) showed that as large banks move to consolidate through in market mergers and acquisitions there will be greater operating efficiency causing a high stream of income for the new entity. More details about revenue enhancement will be analyzed below.

Geographic Diversification

This is designed to smooth the earnings results of a company, which over the long term smoothens the stock price of a company, giving conservative investors more confidence to invest in the company. However, this does not always deliver value to shareholders. According to Spiegel and Gart (1996) the new entity may be able to efficiently and effectively cover a more diverse and wider geographic spread of actual and potential customers. Adkisson and Frazer (1990) argued that states in US that permit geographic expansion positively influence the premium paid by acquirers. The recent literature tends to find little or no evidence of cost efficiency improvements (Berger et al., 2001; Vander Vennet, 2002), although there is some evidence of improvements in profit efficiency and accounting returns (Vander Vennet, 2002; Elsas et al., 2006). Generally studies have found that it is more possible for large efficient banks to expand geographically (Berger et al., 2000; Focarelli and Pozzolo, 2001; Buch and DeLong, 2004; Berger, 2007; Havrylchyk and Jurzyk, 2007; Buch and DeLong, 2003; Hernando et al., 2009; Correa 2009). On the other hand, Wheelock and Wilson (2004) present that efficient banks are less likely to be acquired.

b. Non value maximizing motives

Furthermore, there are alternative clarifications for the consolidation phenomenon. Utility maximization is one reason that managers want to engage in M&As, to obtain more benefits or build an empire (Berger and Hannan, 1998; Bliss and Rosen, 2001; Hughes et al.; 2003). Another motive is that managers want to attain the status of ‘too-big-to-fail’ (TBTF) since it results in an implicit government guarantee which reduces risk and provides cost-of credit advantage.

(14)

14

Utility maximization

Managers want an increase in firm size since they expect that they will enjoy higher salary and other benefits. Berger and Hannan (1998) illustrates that market power which arises from large firm size, can result to higher prices and profits or to ‘quiet life’ for managers away from any kind of competition. On the other hand, Roll (1986) claimed that firm size can destroy firm value if managers overestimate their ability to manage a large institution and pay too much for targets. This phenomenon is called hubris and it is very common. According to Anderson et al. (2004), who examined US bank mergers, post-merger compensation for CEOs is positively related to the expected abnormal returns from merger as being calculated in the announcement date. In addition, Bliss and Rosen (2001) claimed that CEO compensation is increased when large mergers occur. Specifically, they found that CEOs that frequently engage in acquisitions receive 7% more of their compensation in cash than CEOs who do so less frequently. They claimed that this results from value creation or productivity improvements. Furthermore, Rosen (2004) found that when CEOs anticipate compensation improvements, from increased size, they tend to participate in large M&As. Moreover Hughes et al. (2003) found that in banks that managers have high levels of ownership, M&As have more possibilities to be happen. Gupta and Misra (2007) found that abnormal returns are different depending on the case that managers make acquisitions to enhance value or to deal with agency problems. Moreover, Hagendorff et al. (2007), explained that negative abnormal returns for acquirers in the US bank mergers of the 1980s and 1990s are linked with weak governance structures. Cornett et al. (2003) shows that focusing and diversifying mergers are more successful if the CEO of the acquiring bank has a greater stake in the company, either through direct equity holdings or through stock options. However, Hughes et al. (2003) and Hadlock et al. (1999) argued that if managers have a large holding, can lead to entrenchment and management blocking takeover attempts from other banks, even though a takeover would be value generating for the bank’s shareholders.

Safety net subsidies, systemic risk and government‘s role

Banks that grow very large become TBTF and have the chance to exploit safety net subsidies (Kane, 2000; Stern and Feldman, 2004; Mishkin 2006). O’Hara and Shaw (1990) were one of the first to examine the effect of TBTF on the banks that benefit from it and they found positive abnormal returns. Moreover, Ennis and Malek (2005) claimed that it is difficult to evaluate the

(15)

15 subsidies that are created. Shull and Hanweck (2001) found that the 10 largest banks in USA enjoyed the advantages of being TBTF and paid less for funds than smaller banks. Penas and Unal (2004) claimed that positive returns after M&As are related with TBTF status. In addition, Morgan and Stiroh (2005) noticed that when US banks were named as TBTF their bond’s rating increased compared to other banks’ rating. They also suggested that investors were more optimistic than credit raters about the probability to support banks with TBTF status. In contrast to the results of Buch and DeLong (2004) and Hagendorff et al. (2012), Dong et al. (2011) shows that banks expand from more heavily regulated markets into less regulated ones.

Other studies used the merger premium that is paid for large mergers as a proof of the safety net subsidy. Schmid and Walter (2009) found that significant premiums are paid in mega conglomerates between 1985 and 2004. Brewer and Jagtiani (2007) also found that higher premiums are paid for large targets banks. They showed that the eight acquiring banks of M&A transactions that led to new TBTF institutions were willing to pay a combined $14 billion to $17.1 billion in premiums to achieve that status. Hagendorff et al. (2012) showed that there is no conclusive evidence for banks paying a premium to extract safety net benefits.

Arguments that examine systemic risk are closely related to arguments linking consolidation to safety net subsidies. De Nicolo and Kwast (2002) claimed that systemic risk does not necessarily increase as a result of the consolidation trend. Trends in international consolidation and conglomeration are also likely to increase risks for large complex financial firms (De Nicolo et al. 2003). D’Souza and Lai (2006) found that financial sector consolidation may influence the amount of liquidity in money markets, depending on the post-merger allocation of capital within merging firms.

The government also plays a significant role in consolidation decisions through restricting the types of M&A permitted, and through approval or disapproval decisions for individual M&As. In addition, exploitation of TBTF is prevented or other policy goals are promoted. On the other hand, governments may encourage consolidation of troubled financial institutions (Berger et al., 1998).

2.4 Results from previous event studies

Previous event studies that examine the impact of mergers and acquisitions on acquirers, targets and combined firms find mixed results. Their main finding is that, target shareholders gained strong positive abnormal returns with statistical significance, while acquirer shareholders gained marginally negative returns, and the combined abnormal returns were statistically or

(16)

16 economically unimportant (e.g Houston and Ryngaert, 1994; Hudgins and Seifert, 1996; Pilloff, 1996; Subrahmanyam et al., 1997). However, more recent studies claim that the results for the acquirer are not negative. Table 2.1 (Kolaric and Schiereck, 2013) gives an overview of 46 event studies published between 1985 and 2012.

In the beginning, most studies claimed that acquirers destroy shareholder value when consolidation occurs (e.g. Hannan and Wolken, 1989; Baradwaj et al., 1990; Houston and Ryngaert, 1994; Kane, 2000; Cornett et al., 2003; Gupta and Misra, 2007). On the other hand, results from more recent studies differ from this perspective and suggest that bank mergers improve acquirer’s shareholder’s value. It is also shown that in the 1990s the results for the acquiring banks were better than in the 1980s. Houston et al. (2001) suggested that this may happen due to better access to information for acquirers or better market assessment of the value of M&A transactions than in previous decades and find cost reduction from mergers including those involving very large US banks. The fact that mergers and acquisitions became more profitable through years can be explained by De Long and De Young’s (2007) who concluded that M&As generate knowledge that can be exploited by later M&As. Both James and Wier (1987) and Dubofsky and Fraser (1989) showed that acquirers gained positive abnormal returns during the 1970s, especially if the acquirer was bidding for a failed bank in an FDIC auction. Similarly, Bertin et al. (1989) comes up with the same result using data from the 1980s. In a research carried out amongst government assisted mergers between 1989 and 1991, Seidel (1995) found positive abnormal returns for acquirers.

The results for targets are indisputably positive and statistically significant but there is no consensus among the examiners about the magnitude of the abnormal returns since they vary between 1.18% (Frame and Lastrapes, 1998) and 33,58% (Neely, 1987). In general, it is suggested that the average abnormal return expected by target stockholders is between 10% and 15%.

As far as combined abnormal returns are concerned the evidence from studies is mixed. There are studies that find them statistically or economically unimportant (e.g, Houston and Ryngaert, 1994; Piloff, 1996) and other ones that show that the combined returns are significant positive (e.g Becher 2000; Anderson et al. 2004; Becher and Campbell 2005; Al Sharkas and Hassan 2010). However, Toynee and Tripp (1998) were the first ones to claim that the combined entities show negative returns.

(17)

17

Table 2.1 Descriptive overview of event study results

Authors (year) Period Sample Event

Window

Results

Acquirer Target Combined

Desai and Stover (1985) 1976-1982 18 (0,+1) 0.96** NA NA

Trifts and Scanion (1987) 1982-1985 12 (-40,+20) -3.25 21.37** NA

Neely (1987) 1979-1985 26 (+1,+5) 2.82** 33.6*** NA

James and Wier (1987) 1972-1983 60 (-1,0) 1.07*** NA NA

James and Wier (1987) b 1973-1983 19 (-1,0) 2.36*** NA NA

Sushka and Bendeck (1988) 1972-1985 41 (-4,0) -0.74 NA NA

Bertin et al (1989) 1982-1987 33 (-4,0) 2.03*** NA NA

Hannan and Wolken (1989) 1982-1987 43 (-1,0) -3.78*** 11.1*** -0.99

Dubofsky and Frazer (1989) 1973-1983 101 (-1,0) -0.46** NA NA

Wall and Gup (1989) 1981-1983 23 (0,+3) -0.349 NA NA

Kaen and Tehranian (1989) 1979-1987 31 (-1,0) -0.58** NA NA

Baradwaj et al (1990) 1980-1987 23 (-2,+2) -2.07*** 18.8*** NA

Cornett and De (1991) 1982-1986 132 (-1,0) 0.85*** 9.96*** NA

Baradwaj et al (1991) 1981-1987 108 (-5,+5) -2.09*** NA NA

Cornett and Tehranian (1992) 1982-1987 30 (-1,0) -0.8** 8.0*** NA

Madura and Wiant (1994) 1985-1991 152 (0,+36) -27.1*** NA NA

Houston and Ryngaert (1994) 1985-1991 131 (-2,+2) -2.32*** 14.4*** 0.38

Zhang (1995) 1980-1990 107 (-5,+5) 0.4 6.96*** 3.50***

Waheed and Mathur (1995) 1963-1989 259 (0) -0.17 NA NA

Seidel (1995) 1989-1991 105 (-20,+20) 1.84*** NA NA

Piloff (1996) 1982-1991 48 (-10,0) NA NA 1.44*

Hudgins and Seifert (1996) 1970-1989 160 (-1,+1) -0.25 7.53*** NA

Subrahmanyam et al (1997) 1982-1989 225 (-1,+1) -0.9*** NA NA

Houston and Ryngaert (1997) 1985-1992 209 (-4,+1) -2.4 20.4 NA

Gleason and Mathur (1998) 1963-1996 135 (-1,0) -0.29 NA NA

Toyne and Tripp (1998) 1991-1995 68 (-1,0) -2.24 10.97 -0.7

Frame and Lastrapes (1998) 1990-1993 54 (-5,+5) -0.30** 1.183** NA

Kwan and Eisenbeis (1999) 1989-1996 56 (-1,0) NA NA 0.77*

Kane (2000) 1991-1998 110 (0) -1.5 11.14 0.83

Becher (2000) 1980-1997 558 (-5,+5) 0.4 6.96*** 3.50***

Houston et al (2001) 1985-1996 64 (-4,+1) -3.47*** 20.8*** 1.86***

DeLong (2001) 1988-1995 280 (-10,+1) -1.68*** 16.6*** 0.04

Hart and Apilado (2002) 1994-1997 22 (0) -0.415 6.02*** 0.68***

Cyree and DeGennaro (2002) 1989-1995 123 (-5,+5) 0.28 NA NA

DeLong (2003) 1988-1998 438 (-10,+1) -1.89*** 14.8*** 0.12

(18)

18

Table 2.1 continued

The table shows previous event studies for M&As in US. ***, ** and * indicate significance at the 1%, 5% and 10% levels respectively.

2.5 Factors of M&A success

There are many factors that influence the success of M&As. The most common way to find these factors is the execution of a regression. According to Beitel (2002) they can be separated into three groups (i) acquirer and target based characteristics, (ii) transaction specific factors, and (iii) environmental factors.

The characteristics of acquirers and targets can have significant effect on the success of M&A. First of all, the efficiency of M&A is proportional to the target‘s growth rate, its market performance (Beitel et al., 2004; Lorenz et al., 2006; Ismail and Davidson, 2007) and acquirer’s growth rate (Olson and Pagano, 2005). Brewer and Jagtiani (2007) found that bank takeover premiums tended to be higher for more profitable banks. Akhavein et al. (1997), Beitel et al. (2004) and Beccalli and Frantz (2009) argued that large efficiency differentials between the acquirer and the target lead to successful M&As. Moreover, Penas and Unal (2004) found that abnormal returns are more positive when post-merger cost of debt is reduced. Βeitel et al. (2004) and Hernando et al. (2009) also claimed that management skills can be transferred from the acquirer to the target. Although, Kane (2000), DeLong (2001), Campa and Hernando (2006) and Gupta and Misra (2007) argue that absolute target size and transactions’ size affect positively M&A success, Cybo-Ottone and Murgia (2000), Mußhoff (2007), Mititelu and Hunger (2009) claim the opposite. Furthermore, many studies claim that frequent acquirers are more profitable than non-frequent ones (e.g. DeYoung, 1997; Zollo and Singh, 2004; Haleblian et al., 2006; Kolaric and Schiereck, 2013). Altunbas and Marqués (2008) and Kim and Finkelstein (2009) highlighted that when the merging

Authors (year) Period Sample Event

Window

Results

Acquirer Target Combined

Kiyamaz (2004) 1989-1999 207 (-10,+2) 0.83* 4.26*** NA

Becher and Campbell (2005) 1990-1999 443 (-5,+1) -1.29*** 16.7*** 0.93*** Gupta and Misra (2007) 1981-2004 503 (-1,+1) -1.84*** 16.1*** 0.29** DeLong and DeYoung (2007) 1987-1999 216 (-10,+10) -3.09*** 14.9*** 0.26 Brewer and Jagtiani (2009) 1991-2000 8 (-1,+1) -0.4 15.5*** 6.6*** Kim and Finkelstein (2009) 1989-2001 2.204 (-5,+5) -0.3*** NA NA Al-Sharkas and Hassan (2010) 1980-2000 785 (-5,+5) -0.09** 7.59*** 2.26**

(19)

19 institutions have similar strategies, there are better chances for a successful M&A. Finally, M&As are more successful when the banks have similar operations (e.g. Houston and Ryngaert, 1994; Pilloff, 1996; Houston et al., 2001) and when there is high management ownership (Anderson et al., 2004).

Transaction specific factors can also significantly influence the success of M&As. Α factor for which opinions differ, is the geographic location of the banks involved in the merger. The majority of the literature suggests that mergers that are more geographically focused have higher returns than mergers that are across large geographic borders. More specifically, inter-state mergers seem to have negative effect on mergers (Trifts and Scanion, 1987; Toyne and Tripp, 1998) while intra-state mergers appear to be more positive (Trifts and Scanion, 1987; Zhang, 1995; Berger, 1998). On the other hand, other studies claim that the effects on mergers are negative for the intra-state mergers (Madura and Wiant, 1994; Chavaltanpipat et al., 1999; Al-Sharkas and Hassan, 2010). The influence of product focus versus product diversification is another investigated factor in regard to M&A performance but the results are not clear. Although, some studies claim that product focus alone (Beitel et al., 2004; Becher and Campbell, 2005) or combined with geographical focus (DeLong, 2001; Cornett et al., 2006; Mußhoff 2007) seem to have positive effect on M&A success, others argue that diversifying mergers lead to better results (e.g. Berger, 1998; Baele et al., 2007; Hagendorff and Keasey, 2009; Elsas et al. 2010). Another factor that may affect returns is the method of payment (cash, stock, or a mixture) used to finance the merger. It has been found that mergers financed using stock have lower returns than those financed with cash or a mixture of cash and stock (e.g. Baradwaj et al., 1991; Becher, 2000; Ismail and Davidson, 2007; Al-Sharkas and Hassan, 2010). However, Choi et al. (2010) argued that acquisitions paid in cash lead to negative abnormal returns. James and Wier (1987) stated that, when a large number of possible targets exists, there are more chances for the M&A to be successful. In addition, Becher (2000) found that if there are many bidders for a certain number of targets, the acquirer is able to achieve better returns. Baradwaj et al. (1990), dealt with the difference between hostile and non-hostile takeovers and concluded that targets of non-hostile takeovers gained more positive abnormal returns than non-hostile ones. Conversely, bidders of a hostile takeover obtain more negative abnormal returns than friendly bidders. Henock (2004) claimed that takeover speculation results in positive abnormal returns while defensive acquisitions designed to prevent takeovers leads to negative abnormal returns.

(20)

20 Environmental factors can also have a significant influence on the outcome of M&A transactions. Firstly, the final approval of Μ&Α has positive impact on the stock return of acquirers (Sushka and Bendeck, 1988) and targets (Hudgins and Seifert, 1996) because any uncertainties are removed. Some studies claim that deregulation has positive effect on bank M&A (Kwan and Wilcox, 1999) due to improvements in cost and profit performance for merging institutions while others point out that deregulation drives to negative results (Dubofsky and Fraser, 1989; Becher and Campbell, 2005). DeLong and DeYoung (2007) found a positive relation between the market reaction and ‘M&A wave’ since banks can learn by observing but Mußhoff (2007) claimed that this relation is negative because acquirers seem to be overpaying. It is shown by Hagendorff et al. (2008) that when American acquirers consider an international M&A, better abnormal returns are gained, especially when choosing countries that have lower protection levels than USA. Kiymaz (2004) presented that higher protection levels in the target country is positively related to the acquirer’s returns. Beccali and Frantz (2009) came to the same conclusion and Buch and DeLong (2004) suggested that these targets are the most attractive in international M&A.

2.6 External effects of M&As

According to Shull and Hanweck (2001), although U.S. studies using 1980s data find that consolidation resulted in market power effects, with lower deposit rates and higher loan rates in more concentrated markets, studies using 1990s data find weaker relationships between local market concentration and deposit rates. It is also claimed that large banks, and especially merging banks, allocate a lower proportion of their assets to small business loans compared to small banks, although these adverse effects appeared to be offset by an increased flow of credit to small businesses from small incumbent banks (Berger et al., 1998; Berger et al., 1999).

The price and availability of business credit

Studies have mainly focused on the effect of consolidation on the price and availability of business credit. De Young et al. (2004) claimed that small banks rely on soft information for the construction of enduring bank-borrower relationships. On the other hand, Berger and Black (2008) and Uchida et al. (2006) found that small banks use hard information to make business loans.

(21)

21 (Craig and Hardee, 2007) and capital constrained firms (Carow et al., 2006), while others claim that these market power effects can vary depending on the specific product in question (Park and Pennachi, 2007). On the other hand, Berger et al. (2007) found little difference in the small business lending behavior of small and large banks. Francis et al. (2008) argued that consolidation between large banks causes reduction of small business formation in local U.S. markets for two years after the merger, while consolidation between small or medium banks contributes to the increase of business creation.

The results from studies that examine third-party responses to bank mergers are either mixed or neutral. Avery and Samolyk (2004) found that the post-merger decline in lending to small business in the U.S. tends to be matched by increased credit from other incumbent banks. Berger et al. (2004) also found increase in post-merger credit supplied by alternative sources including newly chartered banks. Hauswald and Marquez (2006) showed that mergers enable banks to acquire proprietary information, and they use that information to soften lending competition and grow their market share. They claimed that as competition increases, investment in information acquisition declines, which leads both to lower loan rates and inefficient lending decisions. Whether credit availability declines or holds steady in the aftermath of bank mergers, it is suggested that credit becomes more expensive. Calomiris and Pornrojnangkool (2005) found higher merger spreads for medium-sized and mid-market borrowers, but no change in post-merger spreads for small-sized and mid-market borrowers. In addition, Garmaise and Moskowitz (2006) found that U.S. bank mergers typically do result in higher loan rates. Finally, Panetta et al. (2009) showed that the consolidated institutions make significant adjustments to loan rates, post-merger, to better reflect the riskiness of existing borrowers.

Depositors and other stakeholders

Bank mergers and banking market consolidation tend to impact some deposit rates more than others. Corvoisier and Gropp (2002) showed that increased consolidation between 1993 and 1999 resulted in less competitive pricing on demand deposits, but not on other types of deposits. Craig and Dinger (2009) found that checking account interest rates actually fell substantially in the two years following bank mergers. Focarelli and Panetta (2003) found that after consolidation, deposit rates decline but then they increase again.

(22)

22 mergers resulted in higher market rates on unsecured personal loans, but lower interest rates on automobile loans which were secured by liens. They outlined the importance of soft versus hard information. Stavins (2004) argued that industry consolidation should have a significant effect on the correspondent banking business, and found that the Federal Reserve System felt significant declines in check-processing volumes as industry consolidation intensified. Wu and Zang (2007) found that bank mergers result in higher analyst turnover, especially for the top forecast performers. Robinson et al. (2005) found that Community Reinvestment Act (CRA) protests and negotiated lending agreements that typically follow the announcement of large U.S. bank mergers had no significant effects on stockholder returns at the merging bank.

Merger activity can also have a significant influence on other market participants and potential entrants. Reinforcing the results of earlier studies, recent studies have found strong evidence that interstate and intrastate deregulation led to more new bank charters (Jeon and Miller 2007), that mergers encourage new bank charters (Keeton 2000, Seelig and Critchfield 2003), and that the presence of a large BHC stimulates entry into rural banking markets (Feinberg 2009).

2.7 Reasons of failure

The main reason for the significant amount of failures of M&As lies in the attempt of the firms to merge their different identities into a single one. When the reason of the consolidation is the corporate match and not strategic objective, consolidated banks are more likely to face the problem of conformity mismatch. Moreover, firms should focus more on cost reduction instead of focusing on the development of the firm as a whole. Inefficient communication from the management is another reason of failure, since it is difficult to retain key employees and to attract new ones. More specifically, the reasons of failure of M&As are the following:

Inappropriate Strategy

The strategic plans play a vital role in M&As since a business strategy may not be appropriate for the target company and this could lead to the failure of the M&A. A good strategic analysis before a merger is important but it does not guarantee its success.

Cultural differences

The cultural difference between the firms is one of the main reasons of M&A failure since communication between the consolidated firms becomes very difficult and this leads to

(23)

23 uncertainties. This is generally observed in the cross border mergers and acquisitions.

Lack of Experience and Knowledge

Lack of experience and knowledge can lead to a poor audit before acquisition are also reasons of failure since they can lead to a loss of valuable time and thus synergies are lost.

Over-optimism

Sometimes being over optimistic about the market conditions also results in the failure of merger deals. Management usually makes an attempt to discuss only the positive factors of the deals in order to win the votes of the shareholders to accept the overpriced deal, underestimating the negative factors.

External Environment

The external environment in the economy that surrounds the deal is quite complicated and has to be investigated carefully before an M&A so that a failure could be prevented.

2.8 Basic methodologies

As it was mentioned in the introduction, in order to measure whether the M&A transactions among banks are successful or not the most common methodologies are event studies, efficiency studies and performance studies. There are definitely more methodologies but they are not remarkable since they are rarely used. The basic idea of the event study is to find the abnormal return that is created from the M&A transaction. If it is positive the merger and acquisition is considered successful, otherwise it is considered damaging. Event studies have been used for more than 40 years. This methodology was first applied by Fame et al. on (1969) and since then it became the most famous method to evaluate the success of an M&A. On the other hand, efficiency studies make comparison between the combined institution and the efficient frontier, which is usually the most efficient company of the sample. Finally, performance studies characterize M&A to be successful if accounting ratios are improved.

(24)

24

2.9 Evaluation of the event study

All the methodologies that measure the success of M&A transactions among banks have some disadvantages. As far as event studies are concerned, single factor’s model drawback is that it has weak explanatory power. Another problem with event studies is that the source of any value creation cannot be easily identified, and must be exported from the data using a second statistical procedure. On the other hand, studies show that the results of more complex multi-factor models do not differ from these of the single factor (Zollo and Leshchinkskii, 2000; Cyree and DeGennaro, 2001). Other studies use GARCH (Frame and Lastrapes, 1998) or EGARCH (Al-Sharkas and Hassan, 2010) model and come to the same conclusion. Pilloff and Santomero (1998) put emphasis on their most important advantage which is the provision of a decent prediction of the expected future performance of the consolidated institutions.

(25)

25

3. Data and Methodology

3.1 Data

This section looks at the data used to conduct this study. The bank M&As transactions examined in the present study are collected from ‘Thomson One’ while information related to share prices and ratios are collected from ‘Datastream’.

The selected sample includes all completed M&A transactions of US banks, announced between 2007 and 2015. To be included in sample, it is required that both bidder and target were publicly traded all these years. The initial sample collected was comprised by 472 M&A deals. For both the bidder and the target, the following variables were collected: bidder DS (Datastream) code, target DS code, initial merger announcement date, bidder total return index, target total return index and the price index of Dow Jones Industrial.

From the preliminary sample, few acquirers and targets were deleted for technical reasons since either the DS code or key data were missing. As a result, the remaining bidders and targets are350 and 339 respectively.

On the second part of this research, the ratios that were collected from Datastream for the regression are: return on equity (ROE), market value which on Datastream is the share price multiplied by the number of ordinary shares in issue, the price dividend divided by the earnings per share (P/E), Market to Book value (M/B) which is defined as the balance sheet value of the ordinary (common) equity in the company divided by the market value of the ordinary (common) equity and total debt divided by common equity (Leverage). From this sample, only 283 firms remained and were taken into account for the regression since all the others were deleted for technical reasons (missing DS code) or for missing key data.

(26)

26

Figure 3.1: US Public banks involved in Mergers and Acquisitions during 2007-2015.

Above, figure 3.1 depicts the number of public banks in US that participated in M&As each year from 2007 to 2015. As can be seen clearly in figure 3.1, there was a decline in the number of M&As that occurred during 2007 to 2009 which may be due to the financial crisis at these years. On the other hand, from 2010, the number of M&As has been increasing till 2014 but in 2015 the above mentioned graph illustrates a descending pattern.

3.2 Methodology

3.2.1 Event study methodology 3.2.1.1 Abnormal returns

The aim of the present study is to analyze the shareholders’ value creation of bank M&A deals. For this reason, a standard “event study” methodology is adopted. As already mentioned, the key assumption underlying this method is the hypothesis that stock market prices fully and immediately incorporate all available information (market efficiency hypothesis). As a result, the announcement of an event such as an M&A deal leads to a rapid adjustment of the stock price connected with this event.

Traditional event study benchmark include (i) the market and risk adjusted return model, or short market model, (ii) the market adjusted return model and (iii) the comparison period mean

0 10 20 30 40 50 60 70 80 90 2007 2008 2009 2010 2011 2012 2013 2014 2015 M&A announcements

(27)

27 adjusted return model. The application of each benchmark includes two time series of return data for each security event: an estimation period or else estimation window for estimating the benchmark parameters and an event period or else event window for computing prediction errors based upon the estimated parameters. The prediction error represents abnormal returns. Typically the estimation period and event period do not overlap in order to avoid biasing the parameter estimates in the direction of the event effect.

(i)a. The market model was initially introduced by Dodd and Warner (1983) and Brown and Warner (1985) and it is the most recognized method in empirical studies. It is assumed that security returns follow a single factor market model,

𝑅_𝑗𝑡 = a_j+ β_jR_mt+ ε_jt

where 𝑅_𝑗𝑡 is the rate of return of the common stock of the jth firm on day t, 𝑅_𝑚𝑡 is the rate of a market index, 𝜀_𝑗𝑡 is a random variable that by construction must have an expected value of zero, and is assumed to be uncorrelated with 𝑅𝑚𝑡, uncorrelated with 𝑅𝑘𝑡 for k≠j, not autocorrelated and

homoscedastic while 𝛽𝑗 is a parameter that measures the sensitivity of 𝑅𝑗𝑡 to the market index.

The abnormal return is then calculated

𝐴𝑅_𝑗𝑡 = 𝑅_𝑗𝑡− (𝑎̂_𝑗+ 𝛽̂_𝑗𝑅_𝑚𝑡)

b. Market model with Scholes-Williams beta estimation. Using this model, betas are estimated by both ordinary least squares and the method of Scholes and Williams (1997).

c. Market model with GARCH or EGARCH estimation invokes a single factor model with GARCH (1,1) or EGARCH (1,1) errors.

d. Fama-French three factor model is

𝑅_𝑗𝑡 = 𝑎_𝑗+ 𝛽_𝑗𝑅_𝑚𝑡+ s_j𝑆𝑀𝐵_𝑡+ h_j𝐻𝑀𝐿_𝑡+ 𝜀_𝑗𝑡

which is similar to the simple factor market model but has additionally two factors, 𝑆𝑀𝐵_𝑡 which is the average return of small market-capitalization portfolios minus the average return of three large market-capitalization portfolios and 𝐻𝑀𝐿_𝑡 which is the average return of two high book-to-market equity portfolios minus the average return of two low book-to-book-to-market equity portfolios

(28)

28 𝑅_𝑗𝑡 = 𝑎_𝑗+ 𝛽_𝑗𝑅_𝑚𝑡+ s_j𝑆𝑀𝐵_𝑡+ h_j𝐻𝑀𝐿_𝑡+ u_j𝑈𝑀𝐷_𝑡+ 𝜀_𝑗𝑡

Where 𝑈𝑀𝐷𝑡 is the average return of two high prior return portfolios minus the average return of

two low prior return portfolios.

ii) Market adjusted return model is computed by subtracting the observed return of the market index for day t,𝑅_𝑚𝑡 from the rate of return of the common stock of the jth firm on day t:

𝐴𝑅_𝑗𝑡= 𝑅_𝑗𝑡− 𝑅_𝑚𝑡

iii) Comparison period mean adjusted returns are computed by subtracting the arithmetic mean return of the common stock of the jth_{firm computed over the estimation period, 𝑅}

𝐽

̅̅̅, from its return on day t:

𝐴𝑅𝑗𝑡 = 𝑅𝑗𝑡− 𝑅̅̅̅, 𝐽

According to Halpern (1983), the announcement date is the most appropriate date to gauge the impact of an event. If information leaks before this date, abnormal returns that are generated by the merger would be observed before the event date. Halpern (1983) also argued that at the date of announcement, the stock prices of the acquirer firm will adjust accordingly to reflect the probability of the success of the deal, and the time period to accomplish the merger. Announcement day depicts the first trading day when the news of merger or acquisition reaches the market. Accurate information of this day is important in order to observe the reaction of stock market to unexpected information. As a result in this study, the date of announcement has been set as day 0. A few days around the date of announcement are included in the estimation and the event window because it is not sure that the day 0 was chosen correctly. Estimation window is used to obtain the parameters of the market model in order to calculate expected return on equities when there was not a merger announcement. In other words, it is used to obtain these parameters at the pre-event time period. For this study the estimation window was from -120 to -11 days. MacKinlay (1997) suggested that in this way, one can take information leakage, slow market reaction and effects of end of trading days into account.

The event window which shows the number of days over which we are supposed to measure the possible abnormal returns that are caused by the M&A was set to 21 days that contains the announcement date, 10 days before and 10 days after it. The decision for the event window is very important since a long window could result in the risk that the possibility of finding any considerable empirical evidence might be diluted. On the other hand, with a short event window there might be a possibility that we do not observe the effect of the event if the information is

(29)

29 available after the closing of the market and therefore, it fails to reach the public until the next working day. Furthermore, if the information is leaked out the day before the date of announcement, the effect will be observed on the day before the event day.

After setting these windows, the daily rate of returns of acquirers and targets was calculated using the following formula: Ret= (𝑅𝐼𝑡-𝑅𝐼𝑡−1)/ 𝑅𝐼𝑡−1 where 𝑅𝐼𝑡 represents the closing price of the

current day and 𝑅𝐼_𝑡−1 depicts the closing price of the previous day. The rate of return for the market index (Dow Jones) was estimated using the following formula: 𝑅𝑒𝑡_𝑖𝑛𝑑= (𝑃𝐼_𝑡-𝑃𝐼_𝑡−1)/ 𝑃𝐼_𝑡−1 where 𝑃𝐼𝑡 is the market closing price of the current day and 𝑃𝐼𝑡−1 is the market closing price of the

previous day.

The next step was to find the abnormal returns for both bidders and targets. In this event study market adjusted returns model was used as described above (𝐴𝑅𝑗𝑡 = 𝑅𝑗𝑡− 𝑅𝑚𝑡). This equation

covered both estimation and event window from day -120 to day +10.

The average abnormal return (or average prediction error) 𝐴𝐴𝑅_𝑡 is the sample mean:

𝐴𝐴𝑅𝑇 = ∑ 𝐴𝑅𝑗𝑡 𝑁

𝑗=1

/𝑁

where t is defined in trading days relative to the event day (e.g t=-60, means 60 trading days before the event).

Over an interval of two or more trading days beginning with day T1, and ending with day T2

the cumulative average abnormal return is 𝐶𝐴𝐴𝑅_𝑇1,𝑇2 = 1 𝑁∑ ∑ 𝐴𝑅𝑗𝑡 𝑇2 𝑡=𝑇1 𝑁 𝑗=1 3.2.1.2 T-Test statistic

After retrieving the CAAR, two t-tests were performed in order to show that the abnormal returns statistically differ from zero. The implication of null hypothesis is that the event has no significant effect on the distribution of the stock returns. If the null hypothesis is not accepted, it can be concluded that the acquisition announcement produced new information and on the basis of market perceptions, the acquisition created or destroyed the value of the firm. More specifically H0: banking M&As do not have a positive or negative effect on value of banks (CAAR=0)

(30)

30 Statistical significance at the 1%, 5% and 10% levels was studied and null hypothesiswas rejected when t-statistic was greater than 2.571, 1.96 and 1.64 respectively. Differently the null hypothesis was accepted.

There are many event study test statistics which include Patell test, standardized cross-sectional test, time-series standard deviation test, cross-cross-sectional standard deviation test, skewness-adjusted transformed normal test, generalized sign test, rank test and jackknife test.

For this particular study time-series standard deviation test also called the ‘crude dependence adjustment’ test (Brown and Warner, 1980) was used. This test uses a single variance estimate for the entire portfolio and it does not account of unequal return variances across securities. On the other hand, it avoids the potential problem of cross-sectional correlation of security returns. The estimated variance of AARt is

𝜎̂_𝐴𝐴𝑅2 _{= ∑ (𝐴𝐴𝑅}

𝑡− 𝐴𝐴𝑅̅̅̅̅̅̅)2 𝛦2

𝑡=𝐸1

/(𝑀 − 2)

where the market model parameters are estimated over the estimation period of M= E2-E1+1 days and

𝐴𝐴𝑅

̅̅̅̅̅̅ = ∑ 𝐴𝐴𝑅𝑡 𝐸2

𝑡=𝐸1

/𝑀

The portfolio test statistic for day t in event time is 𝑡 = 𝐴𝐴𝑅_𝑡/𝜎̂_𝐴𝐴𝑅

Assuming time-series independence, the test statistic for 𝐶𝐴𝐴𝑅_𝑇1,𝑇2 is 𝑡 = 𝐶𝐴𝐴𝑅𝑡/(𝑇2− 𝑇1+ 1)1/2𝜎̂𝐴𝐴𝑅

Many studies use the time series standard deviation test (for example see Dopuch, Holthausen and Leftwich, 1986 and Brickley, Dark and Weisbach, 1991).

The second test that was performed was cross-sectional standard deviation test. The portfolio test statistic for day t in event time is

𝑡 = 𝐴𝐴𝑅_𝑡/(𝜎̂𝐴𝐴𝑅𝑡 √𝑁 ) where 𝜎̂_𝐴𝐴𝑅2 𝑡=1/(N-1)∑ (𝐴𝑅𝑖𝑡 𝑁 𝑖=1 − 1 𝑁∑ 𝐴𝑅𝑗𝑡) 𝑁 𝑗=1 2

(31)

31 The estimated variance of 𝐶𝐴𝐴𝑅_𝑇1,𝑇2 is

𝜎̂_𝐴𝐴𝑅2 _𝑡=1/(N-1)∑𝑁_𝑖=1(𝐶𝐴𝐴𝑅_𝑖,𝑇₁_,𝑇₂ −1

𝑁∑ 𝐶𝐴𝐴𝑅𝑗,𝑇1,𝑇2)

𝑁

𝑗=1 2

The test statistic for 𝐶𝐴𝐴𝑅𝑇1,𝑇2 is

𝑡_{𝐶𝐴𝐴𝑅} = 𝐶𝐴𝐴𝑅_𝑇1,𝑇2/(𝜎̂𝐴𝐴𝑅𝑡 √𝑁 )

Brown and Warner (1985) report that the cross-sectional test is well-specified for event date variance fluctuations but not very powerful. Boehmer, Musumeci and Poulsen (1991) report that the standardized cross-sectional test is more powerful and equally well specified. Cowan (1992) reports that the generalized sign test also is well specified for event date variance fluctuations and more powerful than the cross-sectional test.

3.2.2 Methodology of the regression

The second research purpose of this thesis is to find out what factors in M&As affect the target companies’ abnormal returns and how much they affect them. The factors that were chosen have already been mentioned in data section.

The first factor is return on equity (ROE) and was selected because it would be useful to know if target’s profitability affects its abnormal returns. In addition, market value of targets (MV) was employed to find out if the size of the target influences the M&A success for targets. The price per earnings (P/E) ratio was also selected to explain how the profitability of the targets in regard to their stock valuation influences the success of the M&A. In addition, market to book ratio (M/B) was selected to find how the investment opportunity of the targets contributes to their post-merger abnormal returns. The last but not least factor that is investigated in this study is total debt to common equity which is a debt ratio to measure a company‘s financial leverage, so it is used to explain how the leverage of the targets influences their abnormal returns on the date of the M&A announcements. Due to the fact that the targets are observed during an event period, cumulative average abnormal returns are used.

The method that will be used for this study is regression analysis. Regression analysis is a method that is based on obtaining massive data, and use statistical method to formulate the relationship between dependent and independent variables as a model function. This method is frequently used in other event studies addressing the effect of individual variables on stock returns. In this research, the dependent variable is the cumulative abnormal return (CAR), in the event window (-1, +1). The CAR choice refers to a small period around the announcement date in order

(32)

32 to exclude the impact of information leakage. Independent variables are all these factors that were mentioned above. All these ratios were selected for one year (t-1) before the announcement year of the M&A (t), to show how they influence the CAR.

Depending on the relationship between the dependent and the independent variable, the model function can be linear or not. To simplify the research, it is assumed that this model function should be linear. The model function is expected to be formulated as:

𝐶𝐴𝑅_𝑖 = 𝑎 + 𝛽₁𝑅𝑂𝐸_𝑖+ 𝛽₂𝑙𝑛𝑀𝑉_𝑖 + 𝛽₃𝑙𝑛(𝑃 𝐸)𝑖, + 𝛽4𝑙𝑛( 𝑀 𝐵)𝑖+𝛽5( 𝑇𝐷 𝐸)𝑖+ 𝜀𝜄

Where,( 𝐶𝐴𝑅_𝑖 ) is the cumulative abnormal return of stock i over the event window (-1,+1), (α) is the constant term, (𝑅𝑂𝐸_𝑖) is the return on equity of the target , (𝛽₁) is the coefficient of the return on equity, (𝑙𝑛𝑀𝑉𝑖) is the natural logarithm of the market value of the target, (𝛽2) is the coefficient

of the natural logarithm of the market value, (𝑙𝑛𝑃

𝐸)𝑖 is the natural logarithm of the price per

earnings ratio of the target, (𝛽3) is the coefficient of the price per earnings ratio, (𝑙𝑛 𝑀

𝐵)𝑖 is the

natural logarithm of market to book ratio of the target, (𝛽₄) is the coefficient of the market to book ratio, (𝑇𝐷

𝐸 )𝑖 is total debt to common equity ratio of the target or leverage and (𝛽5) is the coefficient

of total debt to common equity ratio of the target.

All the coefficients represent the magnitude that the corresponding independent variable affects the dependent variable. The tool that is used to help with regression analysis is Stata 13. In addition, the calculation of the statistical significance is performed.

Statistical significance is investigated at 1%, 5% and 10% levels as widely used among empirical studies and the critical values are equal to 1.645, 1.96 and 2.576 respectively. In order to check the statistical significance of one variable we must compare the t statistics, that we got from the regressions, with the critical values and when they are above those values we can say they are significant at the corresponding level.

(33)

33

4. Analysis of results

4.1 Results of event study

Table 4.1 reports the findings associated with announcement day 0 for targets. The upper panel reports daily average abnormal returns, number of positive and negative daily abnormal returns, daily cumulative average abnormal returns and the associated t-statistics for the cross-sectional standard deviation test and the time-series standard deviation test for event days -10 to +10 relative to announcement day 0. The lower panel of Table 4.1 displays the cumulative average abnormal returns and associated t-statistics for the cross-sectional standard deviation tests and time-series standard deviation tests for the event windows 1,+1), 2,+2), 5,+5), 10,+10), (-5,0) and (0,+5) around event day 0 as well as the number of positive and negative cumulative abnormal returns. These different combinations of event windows were used as a robustness check for my analysis whereas the main window of interest is (-1,+1).

Table 4.1 shows that the M&A announcements generate market response which is generally positive in the 21-day event window. The cross-sectional standard deviation t-test is statistically significant at conventional levels on event days 0 and +10. The average abnormal returns on these days are 12.46% and 0.29% respectively. On event day 0 the average abnormal return is significant at 1% level because the t-statistic is greater than 2.576. On the other hand, on event day +10 significance exists on 5% level because t-statistic is equal to 2.12. Similarly the time-series standard deviation t-test detects statistically significant daily average abnormal returns on event days -9, -8, -3, 0, +1 and +5. The average abnormal returns are positive these days. More specifically on event day -9 the average abnormal return is 0.56%, on day -8 is -0.4%, on day -3 is 0.44%, on day 0 is 12.46%, on day +1 it is 9.66% and on day +5 is 0.5%. The average abnormal returns for the other event days are not statistically significant since t-statistics are not greater than t-critical values. It is also noteworthy that the proportion of positive and negative abnormal returns on the announcement day is 233 and 107 respectively and one day after the announcement there are 227 positive and 108 negative abnormal returns while for the other event days the same proportion is more balanced. This is also a proof that targets’ abnormal returns react positively to the M&A announcement. The CAARs range from 12.78% for event window (-5,0) to 23.95% for event window (-10,+10). The other CAARs of the event windows (-1,+1), (-5,+5),

(34)

34

Table 4.1 M&A announcement day abnormal returns for targets

Event Day AAR (%) Positive: Negative CS STDEV TS STDEV CAAR (%) t- test t-test -10 0.20 155:175 1.29 1.18 0.20 -9 0.56 168:166 1.08 3.34*** 0.76 -8 -0.40 152:180 -1.61 -2.35** 0.36 -7 0.17 161:161 0.69 1.01 0.53 -6 0.22 158:169 0.84 1.33 0.76 -5 0.15 154:179 0.34 0.91 0.91 -4 0.08 154:177 0.28 0.47 0.99 -3 0.44 168:164 1.53 2.65*** 1.43 -2 -0.27 149:182 -1.52 -1.61 1.16 -1 -0.08 151:179 -0.23 -0.48 1.08 0 12.46 233:107 8.81*** 27.41*** 13.54 1 9.66 227:108 7.25 21.25*** 23.20 2 -0.11 164:165 -0.34 -0.65 23.09 3 0.20 164:164 1.46 1.20 23.29 4 0.10 172:159 0.58 0.59 23.39 5 0.50 163:166 1.62 2.98*** 23.89 6 -0.23 157:169 -1.50 -1.39 23.66 7 0.13 169:161 1.38 0.79 23.79 8 -0.06 156:169 -0.77 -0.38 23.72 9 -0.06 162:169 -0.56 -0.33 23.67 10 0.29 177:157 2.12** 1.70* 23.95

The table shows event days, daily average market-adjusted abnormal returns (AARs), number of positive and negative daily ARs, cross sectional standard deviation test (CS STDEV t-test) and time series standard deviation test (TS STDEV t-test) values for the daily AARs, daily cumulative average abnormal returns (CAARs), and CAARs along with the associated test statistics for the intervals (-1,+1), (-2,+2), (-5,0),(0,+5), (-5,+5) and (-10,+10) relative to the event day 0. Event day 0 is the announcement day of the merger and acquisition. The study period is from 2007 to 2015. ***, ** and * indicate significance at the 1%, 5% and 10% levels respectively.

(-5,0) and (0,+5) were found 22.04%, 21.66%, 12.78% and 22.81% respectively. According to both tests CAARs are statistically significant at the 1% level for all the event windows that were investigated since t-statistic is greater than 2.576. The positive market reaction to M&A announcements is consistent with studies that have been held so far and suggests that M&As create positive abnormal returns for shareholders of the target.

Event periods CAAR (%) Positive: Negative CS STDEV TS STDEV t- test t-test (-1,+1) 22.04 278:62 12.39*** 27.99*** (-2,+2) 21.66 272:68 11.93*** 21.31*** (-5,0) 12.78 226:114 7.96*** 11.48*** (0,+5) 22.81 278:62 12.83*** 20.49*** (-5,+5) 23.13 270:70 12.18*** 15.34*** (-10,+10) 23.95 272:68 12.05*** 11.50***

(35)

35

Table 4.2 M&A announcement day abnormal returns for acquirers

Event Day AAR (%) Positive:

Negative CS STDEV TS STDEV CAAR (%)

t- test t-test -10 0.10 157:181 0.79 0.74 0.10 -9 -0.03 165:178 -0.27 -0.22 0.07 -8 -0.24 142:200 -2.17** -1.85* -0.17 -7 -0.06 166:165 -0.47 -0.42 -0.23 -6 -0.04 177:162 -0.39 -0.32 -0.27 -5 0.17 178:164 1.54 1.32 -0.10 -4 0.02 167:175 0.20 0.18 -0.07 -3 -0.04 161:181 -0.38 -0.32 -0.12 -2 -0.02 169:173 -0.23 -0.16 -0.14 -1 -0.02 165:175 -0.18 -0.14 -0.16 0 0.21 164:186 0.70 1.58 0.05 1 0.23 178:166 1.46 1.74* 0.28 2 0.25 185:152 1.85* 1.88* 0.52 3 0.16 173:165 1.24 1.21 0.68 4 0.08 168:172 0.59 0.61 0.76 5 0.05 161:178 0.53 0.38 0.81 6 0.01 155:180 0.08 0.08 0.82 7 -0.05 171:169 -0.44 -0.36 0.77 8 0.24 175:162 2.24** 1.80* 1.01 9 0.01 160:179 0.05 0.03 1.01 10 0.01 173:170 0.09 0.07 1.02

The table shows event days, daily average market-adjusted abnormal returns (AARs), number of positive and negative daily ARs, cross sectional standard deviation test (CS STDEV t-test) and time series standard deviation test (TS STDEV t-test) values for the daily AARs, daily cumulative average abnormal returns (CAARs), and CAARs along with the associated test statistics for the intervals (-1,+1), (-2,+2), (-5,0), (0,+5), (-5,+5) and (-10,+10) relative to the event day 0. Event day 0 is the announcement day of the merger and acquisition. The study period is from 2007 to 2015. ***, ** and * indicate significance at the 1%, 5% and 10% levels respectively.

Table 4.2 presents exactly the same features as table 4.1 but for the acquirers. Using the cross-sectional standard deviation t-test statistic, it is concluded that the days that have statistically significant average abnormal returns are day -8 at 5% level, day +2 at 10% level and day +8 at 5% level. The values of these abnormal returns are -0.24%, 0.25% and 0.24% respectively. On the other hand, time-series standard deviation t-test detects statistically significant average abnormal

Event periods CAAR (%) Positive:

Negative CS STDEV TS STDEV

t- test t-test (-1,+1) 0.41 176:174 1.18 1.83* (-2,+2) 0.64 173:177 1.72* 2.18** (-5,0) 0.32 169:181 1.07 0.99 (0,+5) 0.97 186:164 2.14** 3.03*** (-5,+5) 1.08 181:169 2.32** 2.49** (-10,+10) 1.02 183:167 2.01** 1.71*

(36)

36 returns on event day -8 at 10% level, day +1 at 10% level, day +2 at 10% level and day +8 at 10% level and the abnormal returns were -0.24%, 0.23%, 0.25% and 0.24% respectively. In contrast with targets, the proportion of positive and negative abnormal returns of acquirers is balanced even at the announcement day when the majority of the acquirers had negative abnormal returns. The CAAR is 0.41% for (-1, +1) event period but not statistically significant at any level according to cross sectional standard deviation t-test. On the other hand, according to time series standard deviation t-test it is considered significant at 10% level. For event period (-2, +2) the CAAR for the acquirers is 0.64% and statistically significant at 10% for cross sectional and 5% for time series standard deviation t-test. For event period (-5, +5) the CAAR is 1,08% and statistically significant at 5% level for both cross sectional standard deviation test and time series standard deviation t-test. Furthermore for (-10, +10) the CAAR is 1,02% and statistically significant for cross sectional standard deviation test at 5% level and for time series standard deviation test at 10% level. As far as event window (-5, 0) is concerned it has no statistical significance. On the other hand, the CAAR for the event window (0, +5) is 0.97% and statistically significant at 5% level with cross sectional standard deviation test and at 1% with time-series standard deviation t-test. Comparing the acquirers and the targets and based on the results of both Table 4.2 and Table 4.1 it can be claimed that targets had by far greater abnormal returns than acquirers and statistically more significant.

Figure 4.1: Average abnormal returns during the event window (-10,+10)

-0,02 0,00 0,02 0,04 0,06 0,08 0,10 0,12 0,14 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10