The Influence of Relative Size Difference and
Relatedness on Acquisition Premium
Michal Valachovič
May 2008
UNIVERSITY OF GRONINGEN
Faculty of Economics and Business
MSc BA Finance
The Influence of Relative Size Difference and
Relatedness on Acquisition Premium
ABSTRACT
Acquisition premium is the percentage difference between the trading price of the target's stock before the announcement of the acquisition and the price per share paid by the acquiring firm. This thesis examines, on the sample of 222 public to public deals conducted between 1st January 2000 and 31st December 2007, the influence of relative size difference, expressed by the market capitalization and the number of employees, and relatedness of involved companies on the acquisition premium. The relative size difference measured by market capitalization proves to have significant impact on the acquisition premium. However, my analysis reveals that this relationship is positive, which is in clear contradiction to the initial hypothesis. When the number of employees is used as the measure of size, the relative difference does not have any impact on acquisition premium. Furthermore, my analysis does not find any significant influence of relatedness of the target and acquiring company on the acquisition premium.
JEL Code: G34
Keywords: Acquisition premium, Relative size difference, Bargaining power,
PREFACE
Throughout the last year I have been studying the MSc BA program with a specialization in Finance at the University of Groningen. For the last part of this master I have been writing the master thesis. This thesis primarily examines two determinants of the acquisition premiums, the phenomenon which has intrigued me ever since I encountered the field of Mergers & Acquisitions from the academic perspective.
The writing of this thesis has proved to be an interesting and challenging process. I would especially like to thank my thesis supervisor, Dr. Auke Plantinga, who provided me support throughout the entire process. He gave me not only guidance, but also plenty of ideas, and inspiration, which made this process even more exciting. I would also like to thank my parents who provided me with support and an optimal background, which made my studying at the University of Groningen much easier.
Michal Valachovič Groningen, May 2008
Author’s student number and contact:
Student number: S 1740024
TABLE OF CONTENT
I. Introduction... 5
A. Relative size difference between acquirer and target ... 6
B. Related vs. Unrelated Acquisitions... 7
II. Previous Works ... 10
A. Determinants of acquisition premiums ... 10
B. Sources of value creation in M&As... 14
III. Data and Methodology ... 17
A. Data collection and processing ... 17
B. Methodology ... 24
IV. Results ... 28
V. Discussion ... 35
A. Impact of relative size difference... 35
B. Insignificance of relatedeness ... 36
VI. Conclusions... 39
References ... 40
List of Tables... 43
I. Introduction
In financial theory it is widely believed that the market value of a company is an unbiased estimate of the present value of expected future cash flows accruing to currently outstanding shares. However, empirical data documenting Mergers & Acquisitions (M&As) activity show the firms are, on average, paying substantial premiums over the market value in order to acquire another company. Bradley and Korn (1979), for example, report that for a sample of acquisitions made over time period from 1975 to 1978, premiums over market value averaged 53 % and ranged between 23 % and 115 %. Porrini (2006) finds, that acquirers paid on average 53,5 % premium over share price on 4 weeks prior to acquisition announcement in the acquisitions between 1990 and 2002, where 10 % of acquirers paid premium over 100 %.
Existence of this phenomenon has significant influence on the structure of the market of M&As. According to some, acquisition premium decreases the incentives of the potential acquiring firm to conduct the acquisition and therefore leads to suboptimal volume of corporate takeovers1. On the other hand, Burkart (1998) explains that higher acquisition premium, induced by competition, translates into higher ownership concentration and so improves corporate governance by decreasing moral hazard and is thus beneficial.
As stated in Flanagan and O’Shaughnessy (2003), the success of M&As is not only based on the ability of acquiring firms to successfully manage target firms but also upon the ability to complete the deals at a price that does not fully appropriate the potential rents. Thus, it is very important to investigate the drivers of premiums.
However, what determines the height of the acquisition premiums? The previous studies provide a high number of various possible explanations: synergies created by an acquisition, competition among bidders, free-riding behavior of the target’s shareholders, bargaining power, medium of payment, managerial hubris, empire building and many more. This thesis primarily examines two possible determinants of acquisition premiums, (1) relative size difference between acquiring and target company, and (2) impact of synergies produced by relatedness of acquisitions. The possible implications of these on acquisition premium are discussed in the subsections below.
A. Relative size difference between acquirer and target
Main focus of the first part of this research is the impact of relative size difference between participating companies on acquisition premium. I believe that relatively larger acquirers are able to pay smaller premium for the relatively smaller targets. This assumption is based on three implications of relative size difference on the acquisition characteristics: (1) relatively larger companies have higher bargaining power which enables them to negotiate more favorable premium, (2) relatively smaller deals are less likely to be induced by managerial hubris or empire-building, and (3) the larger the target company is relative to the acquirer, the higher the potential is for value enhancement in the acquisition and such value potential should be reflected into higher premium. Let’s investigate these assumptions in more detail.
Bradley and Sundaram (2004) argue that power asymmetries are more likely to be present in transactions involving a large acquirer and small target rather than the merger of equals. Large companies have a better position in negotiation over price or conducting due diligence. Smaller firms are less likely to have strong, well-connected or influential boards, less likely to have sophisticated financial institutions as shareholders, and less likely to be represented by sophisticated investment banks and lawyers at the bargaining table. All the factors should work to the advantage of relatively large investors. Bradley and Sundaram (2004) also note that acquisitions of small targets by large acquirers are often referred to as “bear hugs” suggesting that small targets are powerless to stop a determined acquirer with very deep pockets. This note should however be interpreted with caution, because such determined acquisition could be done “at all costs” and therefore be associated with higher premium.
It is also possible to assume that acquisition of relatively smaller targets is less likely to be connected with hubris, or value destructing empire-building and managerial self-dealing, which often leads to significant overpayment for the target. Smaller targets are less prominent and therefore are less likely to be linked with glamour, increase in social status, etc.
According to the reasoning provided above, the following hypothesis is formulated:
Hypothesis 1: Higher relative size difference between acquiring and target company is
associated with lower acquisition premium.
economies of scale and scope, compared with targets which are relatively small to the bidder. I adopt the approach of Hayward and Hambrick (1997) which claims that premiums are a major statement by acquirer’s managers of how much additional value they can extract from the target firm. In other words, I assume that higher potential for post-acquisition value enhancement is translated to higher premium. Therefore the finding of Seth (1990) can be interpreted as another supporting argument for Hypothesis 1.
B. Related vs. Unrelated Acquisitions
The second focus of this thesis is on the relatedness of the acquisition and its impact on the acquisition premium. I suppose that related acquisitions are associated with higher premium, mainly because they possess substantially higher potential for value creation.
Is there simple classification of M&As in which one class is associated with higher value enhancement than the other? Consulting companies selling M&A advice, such as McKinsey, Boston Consulting Group, and Bain & Co, are united in the opinion that the concentration on core business is, in general, a more profitable M&A strategy than diversification2. Haring and Rivet (2004) put it explicitly in their article in Wall Street Journal: “When companies abandon their core and start looking for alternative platforms of growth, it often signals trouble”. They argue that focus strategy has more potential to create value because of experience in the industry and that orientation on core business, indeed, proves to be a more competitive strategy. This implies that related acquisitions should be connected with higher value enhancement. Following paragraphs explain why related acquisitions suppose to have more value enhancing effects than unrelated ones and why this kind of acquisitions should be connected with higher acquisition premium.
It is widely believed that the primary rationale justifying M&A activity is the synergy effect3 which states (applied to corporate market environment) that the value of combined companies is higher than the value of standalone businesses. There are various reasons which could cause such effect, empirical studies4 most commonly list the the following five possible factors: (1) economies of scale, scope, and other operational synergies, (2) more effective management, (3) extended market power, (4) coinsurance in conglomerate acquisitions and (5) diversification of risk. Factors (1) and (3) are clearly more salient in case of related
2 See Frick and Torres (2002), Haring and Rivet (2004), or Cools et al. (2004) for further evidence and
discussion.
3 Other commonly used arguments justifying the M&A activity are for example underpriced target hypothesis or
undermanagement hypothesis, they are out of scope of this thesis. For more, see for example Varaiya (1987).
acquisitions. I argue that involvement in the related industry is crucial for market power expansion and also creates more opportunities for operational synergies. On the other side, synergies created by (4) and (5) are connected with diversification of industry risks and therefore unrelated acquisitions. More ambiguous is the interpretation of (2). It is not clear whether managerial capabilities brought from other industries have higher value enhancing effects, than management practices coming from the related field of business. I believe that the distinction of factors mentioned above supports the claim that related acquisitions create more value enhancement than unrelated ones. I argue that improved operational efficiencies, such as substantial improvement in competitive position and market power as a change of market structure in favor of the combined company, have more value-enhancing potential than decreased risk accompanied by more favorable costs of capital. Moreover, factor (5) as a valid source of synergy, is at least questionable, Seth (1990) concluded that there is no evidence that diversifying acquisitions achieve such level of risk reduction that cannot be duplicated by stockholders on their own5.
Indeed, numerous empirical studies6 point out disadvantages of conglomerate building, that for some companies, negative synergies from, for example, overhead costs and bureaucratic inertia, even exceed any other value enhancements created by the acquisition. As Black (1989) concludes, related synergies may be a source of gain in related, mainly horizontal, mergers, but one is left groping for a plausible source of observed shareholders’ gains from conglomerate mergers.
As already mentioned, I believe that the expected value enhancement will positively influences the maximum that a bidder will pay for the target7. This assumption is supported also by Roy (1989), which states that the target’s stockholders recognize the potential for synergies and approximate their value. From this estimate they infer the bidder’s reservation price and bargain for the maximal possible premium. And that leads to the formulation of a second hypothesis:
Hypothesis 2: Firms acquiring targets in related industry pay higher acquisition
premium.
5 Black (1989) also discovered that conglomerates retain substantial unsystematic risk and that there is no
evidence that managers in diversified companies undertake higher risk projects.
6 These studies are analyzing the conglomerate merger wave in the late 1960s and early 1970s. See for example
Steiner (1975), Mueller (1977), Ravenscraft and Scherer (1987), Black (1989), and Berger and Ofek (1996)
Some authors claim that bidders acquiring related companies have information advantage derived from detailed knowledge of the industry and so can negotiate better terms for the deal, including a lower price.8 There also exists evidence for inverse effects of information advantage, described in Flanagan and O’Shaughnessy (2003), who claim that when an acquirer is closely related to the target, the shareholders of the target can infer the acquirer’s reservation price9 and successfully demand a premium close to this price regardless of whether or not there are competing bidders. This is not the case for shareholders of not closely related target. I believe that this detailed knowledge of the industry reduces industry related information asymmetries equally for both involved parties, acquiring and target company, and should therefore have no significant influence on the acquisition premium.
This thesis is organized as follows: Section II is dedicated to the review of literature, Section III describes the data and methodology, Section IV presents and interprets the results of the analysis, Section V discusses major findings of this thesis in more detail, and Section VI concludes.
8 Russo and Perrini (2005) for example claim that the detail knowledge of the industry might help the managers
of the acquiring firm to assess the right premium for the target.
II. Previous Works
Acquisition premiums have long been perceived as the neglected topic of research in otherwise the very popular field of Merger & Acquisitions. However, over the years, numerous empirical and theoretical papers have been dedicated to determinants of acquisition premiums. This section summarizes main findings of such studies published in major business journals since the late 1980s. The latter part of this section shortly discusses concepts of value creation in M&As connected to the main hypotheses of this thesis and complements the arguments presented in the introduction.
A. Determinants of acquisition premiums
Varaiya (1987) predicts that acquisition premium are positively related to (1) the magnitude of the acquiring firm’s estimate of acquisition gains, and (2) the acquired firm’s relative bargaining power. Acquiring firm`s acquisition gains are determined by the level of a target’s under-pricing, and/or under-management. There are two main factors influence the bargaining power of target, these are: degree of competition (number of bidders) and anti-takeover amendments (corporate charter and legislation). Varaiya found, in a sample of 77 acquisitions from 1975 – 1980, both strong support for (2) and mixed support for (1).
Effects of a form of payment, degree of resistance, and type of offer on the target’s abnormal acquisition announcements’ returns are the primary focus of research by Huang and Walking (1987). This research was conducted on all acquisitions with front-page announcements in the Wall Street Journal from the period of April 1977 until September 1982, totaling 326 deals. Their results stress the importance of interdependencies among these factors. Even though previous studies suggest that tender offer targets earn higher abnormal returns than merger targets, Huang and Walking (1987) show that this difference becomes insignificant after controlling for payment and degree of resistance. They also found that resisted offers are associated with insignificantly higher returns than unresisted offers and that abnormal returns are associated with cash offers are significantly higher than those associated with stock offers.
resistance to his offer, (3) smaller for firms that had received acquisition proposals during the previous two years, (4) larger the later in the sample period the acquisition occurred10, (5) smaller for firms having a tax-loss carry forward, and (6) increased with the percentage of total acquisition price represented by cash, which was most salient in horizontal mergers. Kaufman (1988) considers findings from (1) till (4) as factors that represent the relative bargaining power of the participating firms.
Until now, the debate about the influence of relatedness on the acquisition premiums has not come to any clear conclusions. Singh and Montgomery (1987) found, in a sample of 103 deals with a value greater than 100 million USD in the period 1975 – 1980, that related acquisitions create higher dollar gains than unrelated ones and that acquired firms in related acquisition have substantially higher gains than acquired firms in unrelated acquisitions11. Hitt et al. (1991) also explain that managers might prefer to pay higher premiums when their companies are similar to their targets. On the other hand, Russo and Perrini (2005) argue that because of relatedness managers might be more likely to assess the right premium to be paid if they are going to pursue a related business. According to their opinion, acquiring managers have the advantage of: (1) knowing the strategic factors characterizing a specific industry, (2) sharing with the target companies the same difficulties, but especially the advantage of (3) a somewhat easier post-acquisition integration process. They also believe that managers would be willing to invest more resources in unrelated acquisitions, since they might want to diversify their activity to provide new resources to the firm. However, Russo and Perrini (2005) recognize the complexity involved in unrelated acquisitions and conclude that it might not affect the premium of the deal.
An interesting relationship is described in Flanagan and O’Shaughnessy (2003)12. They found that in the presence of only one bidder the acquisition premium is significantly higher in the case of core-related acquisitions, because it is less difficult for target’s shareholders to infer bidder’s reservation price and bargain higher premium. They imply that it might be the reason why numerous previous studies failed to find that related acquisitions have positive influence on acquiring firms` shareholders` wealth. However, they also discovered that in presence of multiple bidders there is no significant difference between the premium paid in core-related and not core-related acquisition, and the winner’s curse is even
10 This finding is connected to change of legislation regulating M&A activity in the USA. Especially the
Williams Amendments to Securities Act of 1934 (passed in the late 1960s) have been important by shifting more bargaining power to the target company.
11 This finding clearly implies higher acquisition premium in case of related acquisitions.
12 Flanagan and O’Shaughnessy (2003) examine 285 tender offers of US-based manufacturing firms that
magnified. This suggests that the impact of multiple bidders is so strong that it overwhelms the negative impact of the two firms not being core-related, and also that competing bidders influence premium more in the case of unrelated acquisitions.
Palia (1993) has examined merger premiums of all US bank mergers from the period of 1984 to 1987, with a total of 137 deals. He concludes that merger premium is related to characteristics of both acquirer and target, and also to regulatory environments in both acquirer and target bank states. Palia (1993) also found that the separation of ownership and control in acquirer and target banks have significant impact on merger premiums.
The role of interorganizational transfer of routines, and practices in determining the acquisition premium is examined13 by Haunschild (1994). She found that firms tend to pay the same premium in acquisitions as do their interlock partners14. The same relationship proves to be the case for companies using the same investment bank. Such an effect is even magnified in the conditions of uncertainty, since companies are more inclined to look to others for clues or suggestions about appropriate behavior in uncertain, ambiguous situations. They also perceive imitation of behavior of others as a way to limit the costs connected with information acquisition.
Hayward and Hambrick (1997) study the role of chief executive officer hubris in explaining the large size of some premiums paid for acquisitions. For a sample of 106 large acquisitions, they found that four indicators of CEO hubris are highly associated with the size of premiums paid. These are: acquiring company’s recent performance, recent media praise for the CEO, a measure of CEO`s self-importance, and the composite factor of these three variables. The relationship is further strengthened when board vigilance is lacking - represented by high proportion of inside directors and when the CEO is also the chairman of the board. They also show that as the wealth of an acquirer’s shareholders on average decreases following an acquisition, the greater the CEO hubris and acquisition premium, and the greater the shareholder losses.
Moeller (2005) examines the impact of corporate governance structures on the acquisition premiums by using the sample of 5077 deals with announcements dated between 1st January 1990 and 31st December 1999. He presents evidence that high target shareholder control, proxied by low target CEO ownership, low fraction of inside directors, and the presence of large outside block holders, is associated with higher takeover premiums in
13 Haunschild (1994) uses data from 453 acquisitions that occurred during period 1986 to 1993.
14 Interlock partners are independent companies for which director of one company sits in the board of another
friendly takeovers. However such finding is in contrast to studies15 of takeovers in the hostile environment in the 1980s which have shown negative relationship between target shareholder control and takeover premiums. Moeller (2005) explains this contradiction by the change of business environment. In the 1980s, environment with takeover defenses almost not in existance. Even if target CEOs opposed the deal, an acquirer was able to induce target shareholders to fight CEO`s resistance by large premium. The more powerful the CEO, the higher the premium was required to motivate shareholders to overturn the CEO`s opposition. In the 1990s, effective takeover defenses were widely available and hostile offers had only a small chance of success. The more powerful the CEO would be in this scenario, the more likely he is to negotiate about his personal benefits (e.g. position in the merged company, retirement compensation, etc.) instead of maximizing the acquisition premium16.
There are many other factors that proved to have significant influence on the acquisition premium. For example Slunsky (1991) claims agency factors, i.e. the degree of alignment of interests between shareholders and managers of an acquiring company, can contribute to explaining the variance of premium evidently more than real and financial synergies together17. Porrini (2006) discovered, by examining 481 acquisitions between 1988 and 1998, that the presence of investment bankers advising acquirer is positively correlated with the acquisition premium, even after controlling for the presence of investment bankers advising target. Such relationship is explained by the agency conflict between acquirers and their bankers.
Further research concentrates on the behavior of interested firms in the bidding contest. Varaiya and Ferris (1987) found evidence for so called “winner’s course” which suggests that the winning bidder tends to overestimate the target’s value, i.e. to pay excessive premium, so that the subsequent returns for the winning bidder are significantly negative. This topic is approached by Black (1989) from a slightly different perspective. He developed “Overpayment Hypothesis”, which states that for many takeovers, target shareholders gain partly because the bidder pays too much. He argues that any company, from which investors expect poor investments,18 can overpay for another company and this overpayment, if not too great, will not cause the bidders’ stock to drop because investors already expect the bidder to waste the money, one way or another. This also has impact on a bidding contest, because it
15 Theoretical model developed by Shleifer and Vishny (1986), and empirical evidence provided by Song and
Walking (1993)
16 Intervention is of course possible, but it is costly. And also here holds, the more powerful CEO makes the
intervention more costly and difficult.
enables “poor companies”, which are expected to invest available resources in negative NPV projects, to outbid “good companies”, for which such expectation is not incorporated in the share price.
B. Sources of value creation in M&As
Black (1989) identifies synergies as an important source of gains in acquisitions and has distinguishes three classes: operational synergies, improved management by the target, and financial or managerial synergy used due to more efficient use of capital or management talent. Seth (1990) conducted empirical research in order to identify sources of value creation in different types of acquisitions. She assumes five possible sources of value enhancement in M&As: (1) Market power, (2) Economies of scale, (3) Economies of scope, (4) Coinsurance in conglomerate acquisitions, (5) Diversification of risks in acquisitions. In a sample of 63 combined entities (28 of them classified as related acquisition, 35 as unrelated), she concluded that coinsurance is the main source of value creation in case of unrelated acquisitions. In case of related ones, she found that it is the relative size difference of the target to the bidder (this variable proxies for sources associated with important potential changes in operating decisions). Her data also show that different sources of value creation operating in related and unrelated acquisitions create similar magnitude of synergy. Seth (1990) suggests that her results imply that in determining the acquisition strategy it may be unwise to focus on simple predictions. Instead, she argues that it would be more useful to examine potential for specific sources of value creation on which to base the acquisition decision. Shelton (1988) examines a sample of 218 mergers made by randomly selected bidders during 1962 to 1983 classified by changes in product market opportunities of the bidder firms. His analysis shows that acquisitions which permit the bidder access to new but related markets create the most value with the least variance.
The stream of literature which examines the linkage between diversification and performance is extensive. See for example Lang and Stulz (1994), Comment and Jarrel (1995), John and Ofek (1995), Berger and Ofek (1995), and Daley et al. (1997). All these studies report benefits for firms that increase their focus by decreasing their non-core businesses.
targets as the most profitable M&A strategy19. They argue that growth via many small acquisitions is more value enhancing than by limited number of large deals for the following reasons: (1) they are easier to integrate, (2) they are more likely to be in a related business, (3) they are more likely to benefit acquirers from learning-by-doing, (4) they are less likely to be a reason of hubris and empire-building, (5) they are more likely to be able to exploit information asymmetries, and (6) they are are more likely to be acquired with cash rather than stock. However, Bradley and Sundaram (2004) do not find any clear evidence that this kind of acquisition is more profitable than others20.
Moeller et al. (2004) show, in a sample of 12023 acquisitions by public firms, from 1980 to 2001, that larger companies tend to overpay, and make, on average, less profitable acquisitions. One of the possible explanations for this is that large companies acquire large targets, their data also imply that acquisitions of smaller companies are more profitable for the acquirer.
Seth et al. (2000) examine motives of cross-border acquisitions done by U.S. firms. The motivation of these companies to conduct acquisitions, even at the cost of sharing the perceived gain with the target company (paying a premium over market price), can be based on three possible explanations: (1) the Synergy hypothesis proposes acquisitions take place when the value of a combined firm is greater than individual firms’ value. Managers are assumed to be motivated to create economic value in the best shareholder’s interest and have the ability to judge accurately the value potential of the combined firms. (2) the Hubris hypothesis states that bidding firm managers make mistakes in evaluating the targets and the acquisition premium merely reflects the random error, and (3) the Managerialism hypothesis suggests that managers embark on acquisitions to maximize their own utility at the expense of shareholders. They show that synergy hypothesis is the predominant explanation, but the hubris hypothesis coexists with the synergy in acquisitions that are characterized by positive total gains. Their evidence is also consistent with the managerialism hypothesis for the acquisitions with negative total gains.
To the best of my knowledge, there is no study which directly examines influence of relative size difference on the acquisition premium or target’s abnormal return. However, relative size is used in some studies in order to explain the abnormal returns for the acquirer. This variable is often significant, but surprisingly, its sign varies across studies. For instance,
19 See for example Frick and Torres (2002), Haring and Rivet (2004), or Cools et al. (2004).
20 Bradley and Sundaram examine extensive sample of 12476 acquisitions undertaken by publicly listed US
the relative size difference is positive in Asquith et al. (1983), or in Moeller et al. (2004), but negative in Travlos (1987) 21.
21 Asquith et al. (1983) examines 156 bids from 1963 until 1979, data sample of Travlos (1987) contains 167
III. Data and Methodology
A. Data collection and processingThis thesis examines 222 acquisitions. The set of deals was derived from the database Zephyr, and afterwards substantially reduced because of limited availability of crucial data concerning involved companies in Datastream. Zephyr selected deals according to four basic criteria: (1) acquisition deals completed in the time period from 1st January 2000 until 31st December 2007, (2) acquirer and target located in Western Europe or North American Free Trade Area (NAFTA), (3) minimal stake acquired in the deal 30 %22, (4) both target and acquirer are listed companies23. Zephyr offers 2795 deals which comply with all the above mentioned criteria. For the majority of targets which have been delisted following the M&A transaction, data, such as daily stock prices, number of shares outstanding, and number of employees, are no longer available on Datastream. Hence, the unavailability of such data reduced the dataset to 257 deals. Thirty-five deals show certain inconsistencies caused by mistakes in data recording or strong influence of specific factors which could cause bias in further analysis, so these are also excluded from dataset.
As already mentioned above, the dataset contains data from two databases: Zephyr and Datastream. Zephyr provides basic data about the M&A deals. For the purposes of this thesis, I have used the following variables: ID of acquirer and target24, announcement day of the deal, completion date, method of payment, country of origin, and US SIC codes of acquiring and acquired company. Such data are available for all deals with the exception of method of payment, which is available only for 143 deals. Datastream provides daily stock prices, value of S&P 500 Composite Index, number of shares outstanding, and number of employees of involved companies. Both stock prices and value of the index are adjusted for dividends and capital issues in the relevant period.
Market capitalization, the primary measure of size, is calculated as a share price 50 days before the acquisition announcement multiplied by a number of common shares outstanding on that date. Number of employees, alternative measure of size, is reported only on yearly basis. The value used in the analysis is the number of employees at the end of the year which ended at least 50 days before the acquisition announcement.
22 Minimal acquired stake of 30 % is a criterion for including only deals, in which acquirer gains significant
control over the acquired company.
23 As explained below, both involved companies should be quoted, because daily stock price data is required for
further analysis.
24 For the purpose of identification of companies is used issuer’s ISIN code, in some specific cases also company
TABLE I
Characteristics of involved companies
This table summarizes characteristics of the companies involved in the acquisition deals in the data sample. Market capitalization is the market value of all common shares outstanding 50 days prior to the acquisition announcement. No. of Employees is the value at end of the year which ended at least 50 days before the acquisition announcement. M&A Experience is the number of acquisitions conducted by acquirer in the period from 01/01/2000 until 31/12/2007 (fulfilling relevant criteria). In the lower part of the table the relative size difference between acquirer and target is described. Acquirer/Target is a simple fraction. Log Ratio is a ratio calculated according to formula (1): log ratio = log(acquirer)- log (target). Results are reported for two samples, full sample includes all deals, reduced data sample reports only deals with positive acquisition premium
(AP).
Full Sample Reduced Sample
Mean Median Std. Deviation Mean Median Std. Deviation
TARGET
Market capitalization (mil. USD) 1961,14 415,17 4055,52 1695,01 258,70 3810,31
No. of Employees 4581 864 10526 5404 824 12010
ACQUIRER
Market capitalization (mil. USD) 18645,26 2774,11 38377,24 17192,86 2531,02 35571,33
No. of Employees 41836 8599 79761 35454 8429 64833 M&A Experience 8,00 5,00 7,60 7,99 5,00 7,74 RATIOS Market capitalization Acquirer/Target 120,66 5,23 798,46 154,53 6,59 929,53 Log Ratio (SDC) 0,869 0,718 0,864 0,924 0,819 0,892 No. Of Employees Acquirer/Target 59,76 5,51 259,24 52,36 4,67 251,09
As shown in Table I the companies included in the sample have rather large market capitalizations. The difference between mean and median suggest that the average value of market capitalization is strongly influenced by a smaller number of very large companies. This effect is exceptionally strong in case of ratio (Acquirer/Target), which compares the market capitalization of acquirer and target. This difference, when expressed as simple fraction, yields rather large and dispersed results. The use of such a variable, as ratio for relative size difference between acquiring and target company, could cause imprecision in further analysis, so I prefer to use the log ratio, which yields smaller and less dispersed values.
⎟⎟ ⎠ ⎞ ⎜⎜ ⎝ ⎛ = T A MC MC SDC log10 (2)
Where SDC is an abbreviation denoting relative size difference in the market capitalization, MCA is the market capitalization of the acquirer, and MCT is the market
capitalization of the target. Relative difference in number of employees, alternative measure of size (denoted as SDE), is calculated the same way.
There is no clear approach towards calculation of acquisition premium. According to Haunschild (1994), a premium is the percentage difference between the trading price of the target's stock before the announcement of the acquisition and the price per share paid by the acquiring firm. I believe that it is crucial to consider the share price well before the acquisition announcement. Financial markets respond also to the rumors about the acquisition, caused by information leaks, and the stock price of the target at the date of official announcement (or few days before) might be already significantly influenced by the expectations of this event. This would create bias in the calculation of the actual acquisition premiums. A more precise approach would be to take the rumor date (indicator also provided by Zephyr) instead of the announcement date. The possible problem of this approach is that, in case of numerous acquisitions, the rumor date is equal to the announcement date, which in my opinion indicates imprecision of this measure. Therefore I prefer, for the sake of data consistency, to define the premium as the percentage difference between the trading price of the target's stock 50 days prior to the announcement of the acquisition25 and the stock price one day before
25 Similar approach was adopted by Officer (2003), who considers stock price 43 days prior to bid announcement
completion26. In order to eliminate other possible influences on the stock price movements, premiums are adjusted by returns of S&P 500 Composite Index for the relevant period (that is from 50 days before the announcement date until one day prior to completion). Influence of index returns is substantial, which can be explained by considerable length of the time-period in question and high stock market volatility27. The average index return in the period used for premium calculation is 2,34 %, however, it ranges from -22,53 % up to 22,73% for each individual case.
The formula used for the calculation of acquisition premium is the following:
⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − − ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ − = − − = − 1 1 50 1 50 1 A C A C SP SP P P AP (1)
Where AP is the acquisition premium for particular deal, Pc-1 is the closing stock price
of the target company on the day before the completion of the acquisition, PA-50 is the target
stock price 50 days prior to announcement, SPc-1 is the value of S&P 500 Composite Index
one day prior to acquisition completion, and SPA-50 is the value of S&P 500 Composite Index
50 days before the announcement of acquisition. TABLE II
Distribution of acquisition premiums
This table presents the distribution of acquisition premiums in 222 acquisitions (full sample). Size of Premium Percent of Sample
Negative 26,58% 0 % to 20 % 43,24% 20,1 % to 40 % 18,02% 40,1 % to 60 % 6,31% 60,1 % to 80 % 3,60% Greater than 80 % 1,35%
Acquisition premiums from the deals in my data sample are described in Tables II and III. Rather surprising is the fact that premium weighted by market capitalization of the target is substantially smaller than equally weighted average. Another unexpected finding is that there are 59 deals for which the acquisition premium is negative. I am not aware of any
26 Datastream provides the stock price at the end of the day. Because precise price paid per stock in the
transaction is not available for the most of the deals, I believe that price at the end of the day prior to completion of the deal is the best available value, fairly close to the actual price per share paid by acquirer.
27 Well known period of high stock market volatility, so called Tech Bubble and its following burst, is also
Table III Deal characteristics
This table presents the basic deal characteristics. Premium nominal is the relative difference in the share price between the price on 50 days prior to announcement and at the completion date. S&P 500 Return is the cumulative index return in that period. Premium real (AP) is the acquisition premium which accounts also for the return of S&P 500 index, it is calculated by using formula (2): AP = Premium nominal - S&P 500 Return. AP is used in this thesis for further analysis. Weighted AP is the AP weighted by the market capitalization of the target. Complperiod represents the number of days between the acquisition announcement and completion of the deal. Results are reported in two data samples, full sample includes all deals, reduced sample reports only deals with positive
acquisition premium (AP).
Full Sample Reduced Sample
Mean Median Std. deviation Mean Median Std. deviation
Premium
Premium nominal 18,06% 9,16% 26,73% 25,15% 15,97% 27,87%
S&P 500 Return 2,34% 2,74% 6,34% 1,95% 2,24% 6,94%
Premium real (AP) 15,72% 6,76% 26,51% 23,20% 13,18% 27,29%
Weighted AP 7,36% 14,44%
Other
commonly accepted intuitive explanation why there should negative be stock price reaction for the target company after the acquisition announcement28. Negative premiums can only probably be explained by idiosyncratic factors29, which cannot be accounted for in the analysis. Alternative explanations could also reason that expectations of the acquisition had been incorporated in the share price also 50 days prior to the acquisition announcement, but the actual acquisition offer has fallen short of these expectations. In the analysis, I conduct tests on two samples: full sample and reduced sample. The full sample contains all 222 deals, whereas the reduced sample contains 163 deals, because deals with negative premium have been excluded. I assume that inclusion of reduced sample can deliver more consistent results. On the other hand, I consider also inclusion of full sample as useful, because it may reveal important information, such as what kind of deals are likely to have very low acquisition premium.
As described by Flanagan and O’Shaughnessy (2003), even though the concept of relatedness receives a great deal of attention in the M&A literature, there is no clear and commonly used model of how to assess the relatedness of a transaction. I use the US SIC codes of acquiring and acquired company. Deal is considered to be related in case that acquirer and target do not differ in the first two digits of the US SIC code30. This is a loose measure with certain limitations, mostly because the extent of relatedness may vary greatly. Numerous companies are assigned multiple SIC codes, and companies that do not differ in the first two digits of one of the SIC codes do not necessarily have related operations, either at the vertical or horizontal level (with potential for economies of scale and scope), and/or do not serve the same customers. Unfortunately, I do not have enough information to determine the relatedness of every deal on a case by case basis. Therefore, I also apply a more strict measure for classification of relatedness where acquisition is considered to be related only in case both companies belong to same industrial group, i.e. they do not differ in all four digits of the US SIC code. Given that this criterion divides the data sample into two more equal subgroups and strict measure seems to be more credible proxy for possible synergy effects, I prefer to use strict measure as the primary variable for relatedness.
There are also numerous factors which are used as control variables. At least two of them, namely acquirer’s experience and legislative investor protection, call for word of
28 However it is important to note that the negative acquisition premium is in most cases not caused by the
decrease of stock price itself, but the rise in price is smaller than the market index returns in the relevant period.
29 One of the possible explanations could be, for example, that due diligence revealed unexpected and negative
information.
further explanation. In order to determine the acquirer’s M&A experience I use also transactions outside of the data sample, because of a very limited set of deals included in the data sample. This variable is defined as the number of acquisition deals by a particular acquirer included in Zephyr which: (1) were completed between 1st January 2000 and 31st January 2007, (2) where the target was located in Western Europe or North American Free Trade Area (NAFTA), and (3) where the minimal acquired stake was 30 %. The number of deals is considerably higher than one included in the original dataset, this fact can be explained by the exclusion of the fourth search criteria, which implies that the target company does not have to be quoted.
Investor protection legislation classifies countries by their level of regulation and legislative norms concerning M&A activity. With the example of Marshall et al. (2007), I consider countries governed by Common Law and Scandinavian Civil Code to have strong investor protection, generally perceived as in favor of M&A activity, and countries governed by German or French Civil Code to have low investor protection, generally perceived as limiting towards M&A activity. Countries regarded to have strong investor protection for the purposes of further analysis are Canada, Denmark, Finland, Norway, Sweden, UK, and USA31. For the purposes of further analysis, I have used only the level of investor legislative protection in the target’s country of origin.
TABLE IV
Correlation Matrix - Full sample
This table presents the correlation coefficient among the independent variables used in further analysis. Results are reported for the full sample which includes all deals.
SDE STRICT
REL
LOOSERE
L
CROSSBOARD M&A EXPERI
E NCE CASH STRO NGINV P ROT COMPLPE R IOD SDC 0,446 0,086 0,046 0,154 0,391 0,159 -0,111 0,138 SDE 1,000 -0,043 -0,016 0,068 0,395 0,085 -0,104 0,023 STRICTREL 1,000 0,638 0,282 0,132 0,036 0,011 0,141 LOOSEREL 1,000 0,321 0,083 0,000 -0,052 0,054 CROSSBOARD 1,000 0,289 0,126 -0,249 0,112 M&A EXPERIENCE 1,000 0,174 -0,195 0,026 CASH 1,000 0,000 0,125 STRONGINVPROT 1,000 0,082 COMPLPERIOD 1,000
As Table IV reveals, there is no evidence for multicollinearity among the explanatory variables. The correlation coefficient higher than 0,4 occurs only in two cases, between SDC and SDE, and between loose and strict measures of relatedness. However, both are just alternative measures for the same factor and are not to be used in the same equation.
TABLE V Classification of deals
This table provides overview of various deals' classifications which are used in further analysis. Acquisition is regarded as related according to Strict measure in case that acquirer and target have assigned the same US SIC code (all four digits), according to Loose measure
they do not differ in first two digits of the US SIC code. Method of payment describes the medium in which the acquisition price was paid. Other methods of payment include shares, loan notes, debt assumed and their combination. Deals are classified as crossboarder when acquirer and target are not based in the same country. Strong investor protection means that target is based in the country with either Common Law or Scandinavian Civil Code legislation (generally perceived as in favour of M&A activity). Results are reported in two data samples,
full sample includes all deals, reduced sample reports only deals with positive acquisition premium (AP).
Full sample Reduced sample
TOTAL No. of deals 222 163
RELATEDNESS Strict measure related 118 90 Unrelated 104 73 Loose measure related 158 116 Unrelated 64 47 METHOD OF PAYMENT Cash 111 88 Other 32 45 Undisclosed 79 30 CROSSBOARDER Crossboarder 108 80 Domestic 114 83
LEGISLATIVE INVESTOR PROTECTION
Strong 48 36
Weak 174 127
B. Methodology
In order to design most appropriate models, I have included numerous control variables, which according to financial theory, and intuition, could influence the size of the acquisition premium. I have tested the determinants of the acquisition premium by estimating the following equation using OLS:
D COMPLPERIO ROT STRONGINVP CASH EXPERIENCE CROSSBOARD STRICTREL SDC AP 7 6 5 4 3 2 1 1 β β β β β β β α + + + + + + + = (3)
Where AP is the acquisition premium calculated by using the formula (2); SDC is the relative difference of market capitalizations between acquiring and acquired company, calculated by formula (1); STRICTREL is a dummy which is one if companies are assigned the same US SIC code and zero otherwise; CROSSBOARD is a dummy which is one for deals where acquirer and target are not based in the same country and zero otherwise; EXPERIENCE is the number of M&A deals completed by acquirer from 1st January 2000 until 31st December 2007; CASH is a dummy which is one if method of payment used in the deal was cash and zero otherwise (it includes also deals for which the method of payment is unavailable); STRONGINVPROT is a dummy which is one for deals in where target’s country of origin has strong investor protection, zero otherwise; COMPLPERIOD is the number of days between the announcement of the acquisition and its completion.
Virtually the same procedure is followed for the alternative measure of size, the number of employees. The alternative equation follows:
D COMPLPERIO ROT STRONGINVP CASH EXPERIENCE CROSSBOARD STRICTREL SDE AP 7 6 5 4 3 2 1 1 β β β β β β β α + + + + + + + = (4)
Where SDE is the relative difference of number of employees between acquiring and acquired company, calculated by formula (1), all other variables remain the same as in equation (3).
D COMPLPERIO ROT STRONGINVP CASH EXPERIENCE CROSSBOARD LOOSEREL SDE AP 7 6 5 4 3 2 1 1 β β β β β β β α + + + + + + + = (5)
LOOSEREL is a dummy which is one if participating companies do not differ in first two digits of their US SIC code.
After estimating equation (3) – (5) by using OLS, I apply the Ramsey test (1 fitted term) in order to determine the correct functional form. The White hetescedasticity test (no cross-term) is also used in order to indicate the presence of either heteroscedasticity among residual. If these tests imply some deficiencies, different functional form or other method than OLS will be applied.
All variables which are not significant on the 10 % confidence level in equations (3), (4), and (5) will be dropped from the model in order to exclude the variables with no explanatory power, with exception of SDC and STRICTREL, which are the primary focus of this research. This allows for the examination of the influence of these factors in more detail. As explained in Brooks (2002), exclusion of irrelevant variables improves the efficiency of estimators. Variables, which would otherwise have been marginally significant, may no longer be so in the presence of irrelevant variables.
The criterion of relatedness divides the sample of deals into two subgroups: related and unrelated acquisitions. I apply Chow’s breakpoint test to investigate whether explanatory variables used in this analysis have the same explanatory power over both subgroups, i.e. whether there are no structural change in the researched relationship. This approach is rather unconventional, mainly because Chow’s breakpoint test is mostly used for time-series analysis. Nevertheless, its characteristics allow it to be used also for this kind of cross-sectional analysis. The Chow test is applied for equations (6) and (7), listed below. Note that the dummy for relatedness is dropped, as the subgroups are sorted according to this criterion. The test is applied for both, strict and loose measure of relatedness.
IV. Results
The results from estimation of equations (3) - (5) are displayed in Table VI for the full sample and in Table VII for the reduced sample. There are four explanatory variables which are statistically significant32 in the full sample: relative size difference in market capitalization (SDC), acquirer’s M&A experience, method of payment, and completion period. All these variables are significant also for reduced sample with exception of acquirer’s M&A experience. Value of adjusted R2 can be considered to be fairly high for the given model. It is natural that the value acquisition premium is not fully explained by seven factors included in the model, but determined also by deal fundamentals and idiosyncratic specifics. Adjusted R2 is slightly higher for the full sample than for the reduced one.
A striking finding is that coefficient for relative size difference in market capitalization, significant at a 5 % confidence level, is positive (in both equations (3) and (5)). This means that larger size difference between acquirer and target is associated with higher premium. Such a finding is in clear contradiction with Hypothesis 1, and its implications are discussed in the section below.
The independent variable with highest explanatory power in the models is completion period, thus the number of days between the acquisition announcement and its completion. Coefficient is positive and statistically significant, even at a confidence level of 1 %. It shows that the longer it takes to complete the acquisitions, or i.e. for a target’s shareholders to accept the offer, the higher the premium is paid. I assume that it is possible to interpret the length of the completion period as some form of target’s resistance to acquisition and therefore the most plausible and probable explanation is that targets apply takeover defensives, or other measures to make the acquisition more difficult for the acquirer, which in the end results in higher acquisition premium33. However, it is considered necessary to mention that the coefficient for completion period is rather small and therefore it has no substantial economic importance.
The dummy for cash as a method of payment is also positive and significant (on 5 % level of confidence) for both samples. This finding supports the conclusions of numerous previous studies34 which also discovered that cash as a method of payment in acquisition is associated with higher acquisition premium. The effects are consistent with a tax explanation
32 Thus, significant on at least 10 % confidence level.
TABLE VI
Cross-sectional analysis - Equations (3), (4), and (5) - full sample
This table presents the results of the cross-sectional analysis as defined by equations (3), (4), and (5), where acquisition premium (AP) is the dependent variable. The independent variables are SDC - the relative difference in market capitalization of acquiring and target company, SDE - the relative difference in the number of employees of acquiring and target company; STRICTREL - a dummy for relatedness according
to strict measure; LOOSEREL - a dummy for relatedness according to loose measure; CROSSBOARD - a dummy for crossboarder deals; M&A EXPERIENCE - the number of acquisitions conducted by acquirer in the period from 01/01/2000 until 31/12/2007 (and fulfilling relevant criteria); CASH is a dummy for deals in which cash was used as a method of payment; STRONGINVPROT - a dummy for deals where target is based in the country with legislation in favour of M&A activity; COMPLPERIOD - the number of days between acquisition announcement and its completion. Jarque-Bera test, Ramsey RESET test and White test are presented in CHI2 p-values. Respectively, ***, **, and * denote
statistical significance at the 1 %, 5 %, and 10 % level.
Equation (3) Equation (4) Equation (5)
Coefficient t-stat Coefficient t-stat Coefficient t-stat
α1 0,052 1,468 0,072* 1,944 0,076* 1,875 SDC 0,053** 2,419 x x 0,052** 2,381 SDE x x -0,004 -0,215 x x STRICTREL -0,001 -0,028 -0,002 -0,051 x x LOOSEREL x x x x -0,041 -1,078 CROSSBOARD -0,025 -0,674 -0,021 -0,572 -0,012 -0,337 M&A EXPERIENCE -0,005* -1,926 -0,002 -0,813 -0,005* -1,919 CASH 0,080** 2,384 0,085** 2,520 0,078** 2,343 STRONGINVPROT 0,046 1,097 0,040 0,945 0,047 1,130 COMPLPERIOD 0,001*** 5,097 0,001*** 5,445 0,001*** 5,164 Adj. R2 0,164 0,142 0,169
Jarque-Bera test (p-value) 0,00% 0,00% 0,00%
Ramsey RESET test (p-value) 15,54% 62,69% 28,65%
TABLE VII
Cross-sectional analysis - Equations (3), (4), and (5) - reduced sample
This table presents the results of the cross-sectional analysis as defined by equations (3), (4), and (5), where acquisition premium (AP) is the dependent variable. The independent variables are SDC - the relative difference in market capitalization of acquiring and target company, SDE –the relative difference in number of employees of acquiring and target company; STRICTREL - a dummy for relatedness according to
strict measure; LOOSEREL - a dummy for relatedness according to loose measure; CROSSBOARD - a dummy for crossboarder deals; M&A EXPERIENCE - the number of acquisitions conducted by acquirer in the period from 01/01/2000 until 31/12/2007 (and fulfilling relevant criteria); CASH is a dummy for deals in which cash was used as a method of payment; STRONGINVPROT - a dummy for deals where target is based in the country with legislation in favour of M&A activity; COMPLPERIOD - the number of days between acquisition announcement and its completion. Jarque-Bera test, Ramsey RESET test and White test are presented in CHI2 p-values. Respectively, ***, **, and * denote
statistical significance at the 1 %, 5 %, and 10 % level.
Equation (3) Equation (4) Equation (5)
Coefficient t-stat Coefficient t-stat Coefficient t-stat
α1 0,119*** 2,671 0,132*** 2,933 0,145*** 2,814 SDC 0,051** 2,034 x x 0,050** 1,985 SDE x x 0,023 0,921 x x STRICTREL -0,009 -0,214 -0,008 -0,196 x x LOOSEREL x x x x -0,044 -0,971 CROSSBOARD -0,035 -0,783 -0,027 -0,587 -0,027 0,045 M&A EXPERIENCE -0,004 -1,476 -0,003 -1,041 -0,004 -1,498 CASH 0,082** 2,006 0,084** 2,043 0,080* 1,961 STRONGINVPROT 0,059 1,163 0,052 1,004 0,058 1,135 COMPLPERIOD 0,001*** 4,140 0,001*** 4,349 0,001*** 4,157 Adj. R2 0,141 0,122 0,146
Jarque-Bera test (p-value) 0,00% 0,00% 0,00%
Ramsey RESET test (p-value) 34,22% 90,64% 57,42%
where shareholders demand higher premium in situations that will force them to pay immediate taxes on their gains35 (Huang and Walking (1987)).
The model also shows that acquirers with higher experience in M&A activity tend to pay lower premium. This complies with findings of previous studies which show that more frequent acquirers have higher acquisition returns and are able to strike deals on more favorable conditions, thus also pay lower premium. However, the explanatory variable, M&A experience, is significant only for a full sample (at a 10 % confidence level) and statistically insignificant when applied to a reduced data sample.
Estimation of the model described in equation (4) reveals that the relative difference of number of employees between acquirer and target tends to have no explanatory power over the acquisition premium. It is intuitive that the market value of involved companies determines more the specifics of the deal than the number of employees. On the other hand, it remains surprising that there is such a low correlation between these two measures of size. Completion period remains the variable with highest explanatory power also in this model. Its coefficient is small, positive, and highly significant. A dummy for acquisitions paid in cash is also significant with a 5 % confidence level. No other variables seem to have explanatory power in the model where number of employees is used as a measure for relative size difference. It is therefore natural that the value of adjusted R2 is lower than in the case of the previous model, as described by equation (3).
The dummy for relatedness according to strict measure appears to have no explanatory power over acquisition premium; the value of its t-statistics is extremely low. Dummies for cross boarder deals and strong investor protection in the target’s country of origin have slightly higher value of t-statistics that the one for relatedness, but remain clearly statistically insignificant.
The dummy for relatedness according to strict measure is strongly insignificant in all the above described models. I substitute this variable with relatedness by loose measure and so estimate equation (5) in order to see whether its explanatory power increases. As shown in Tables VI and VII, that indeed proves to be the case. Even though the variable remains still statistically insignificant at a 10 % confidence level, the value of its t-statistics increases substantially and the explanatory power of the whole model, expressed by adjusted R2, also slightly increases. This suggests that the loose measure is a more appropriate proxy for
35 The tax regulations in most countries included in the data sample oblige investors to pay tax on income
relatedness for use in this model. On the other hand, it is important to mention that the coefficient of this dummy is negative (the same as in the case of strict measure) which would mean that related deals are associated with lower premiums. Such finding would be opposite to Hypothesis 2.
The Ramsey RESET test shows that applied functional form is appropriate and there is no need for use of higher power of independent variables. Furthermore, the White test does not indicate presence of heterescedasticity among residuals. A very low p-value of Jarque-Bera test implies that residuals are not normally distributed, however, there is no simple and effective way how to deal with this phenomenon. According to the information provided above, I consider OLS to be an appropriate method for the estimation of the models presented in equations (3), (4), and (5).
As suggested in previous section, I drop all insignificant variables from the model (except the SDC and relatedness which are the primary focus of this thesis) in order to see whether its properties improve. The results from the reduced form of the models, where dummies for investor protection and cross boarder deals for both samples and M&A experience only for reduced sample are excluded, are reported in Tables XI and XII (in the appendices).
Most variables do not change their explanatory characteristics in the reduced form of the models. Acquirer’s M&A experience has slightly larger negative coefficient and is significant at a 5 % confidence level which suggests that its explanatory power increases after dropping insignificant variables. The opposite is the truth for the relative size difference for the reduced data sample. This variable remains statistically significant for the full sample, however when applied to a reduced data sample, the value of t-statistics drops substantially and the variable on the whole tends to be insignificant. Such a result is worrying to some extent and difficult to explain, since there is no multicollinearity among explanatory variables which could explain the cause for this phenomenon. The explanatory power of SDE as an alternative measure of size difference increases, but remains insignificant.
TABLE VIII Chow`s breakpoint test
This table presents results of Chow`s breakpoint test. Two subgroups examined are the related and unrelated acquisitions, sorted by either loose or strict measure. The results are reported in form of F-statistics p-values which indicate on what confidence level it is possible to reject H0 that parameters are stable for both subgroups. Results are reported
for equations (6) and (7), and for both, full and reduced sample (excludes observations with negative premium)
Full sample Reduced sample
(6) (7) (6) (7) Loose measure (p-value) 24,50% 59,08% 25,04% 35,40%
Strict measure (p-value) 19,48% 48,85% 60,17% 67,41%
TABLE IX Tests of equality
This table provides an overview of results from testing the equality of acquisition premiums for two subgroups, related and unrelated acquisitions. Acquisition is regarded as related according to strict measure in case that acquirer and target have assigned the same US SIC code (all four digits), according to loose measure they do not differ in first two digits of the US SIC code. Results from ANOVA analysis (Wilcoxon signed-rank test) are reported in CHI2 p-values, which show on what confidence level are means (medians) of two subgroups statistically different from each other. Results are reported in two data samples, full
sample includes all deals, reduced sample reports only deals with positive acquisition premium (AP).
Full Sample Reduced Sample
Related Unrelated All Related Unrelated All
Strict measure
No.of observations 118 104 222 91 72 163
Mean 16,57% 14,76% 15,72% 23,10% 23,32% 23,20%
Std. Deviation 25,50% 27,71% 26,51% 25,60% 29,47% 27,29%
ANOVA F-test (p-value) 61,35% 96,00%
Median 7,45% 5,38% 6,76% 14,91% 12,75% 13,18%
Wilcoxon signed-rank test (p-value) 23,30% 51,14%
Loose measure
No.of observations 158 64 222 116 47 163
Mean 14,54% 18,65% 15,72% 21,62% 27,11% 23,20%
Std. Deviation 23,95% 31,99% 26,51% 24,31% 33,52% 27,29%
ANOVA F-test (p-value) 29,57% 24,55%
Median 6,74% 7,72% 6,76% 12,75% 13,38% 13,18%
V. Discussion
A. Impact of relative size difference
The results clearly show that higher relative difference in market capitalization of involved companies is associated with higher acquisition premium, which contradicts the Hypothesis 1 formulated in this thesis. Such a finding could indicate the presence of a so called size effect, as specified in Moeller et al. (2004), which states that large acquirers tend to overpay. However, when connected with the fact that the average acquisition premium weighted by target’s market capitalization is substantially smaller than equally weighted average36, it seems, rather, that acquisitions of smaller targets tend to receive higher premium. How can such a relationship be explained?
The intuition behind it could be that smaller companies provide high potential for value creation when they are combined with a significantly higher acquirer. Examples of such potential could be access to new markets, core competencies, and/or new technologies. An acquirer may be eager to gain such advantages at any costs, and the increased acquisition price caused by high premium could be perceived more negligible in case of relatively smaller targets. On the other hand, it is very difficult to explain why the potential for value creation should not be higher for relatively larger targets, which offer, in addition to the above mentioned opportunities also much a more salient increase in market power, or economies of scale. For example Seth (1990) concludes that the possibility of value creation by combining two companies increases as their relative size difference decreases. And higher value creation opportunities should be associated with higher acquisition premium. A positive relationship between acquisition premium and relative size difference accordingly implies that the influence of bargaining power asymmetry on the value of acquisition premium, as described in the Section I.A., is practically non-existent. This is simply not intuitive, as long as I assume rational and value maximizing behavior of managers representing both acquirer and target.
According to this assumption, managers of an acquiring firm try to negotiate lowest possible price, and target’s managers try to maximize wealth of existing shareholders by bargaining the highest price the acquirer will be willing to pay. One could argue that managers of relatively larger targets simply do not do enough to extract rents for their existing shareholders in the acquisition, but I am not aware of any plausible explanation why this kind of behavior would be more salient in case of larger targets. Accordingly, it is very difficult to
36 As Table III reveals, premium weighted by target’s market capitalization is 7,36 % and equally weighted
