1
The Managers’ Susceptibility to Overpayment in
Acquisitions Over Time
Faculty of Economics and Business
MSc Business Economics: Finance Track
July 2014
Supervisor Dr. F.S. Peters
Omaima Halima Ez-Zaitouni
2 Abstract
This paper examines the trend in takeover premiumsover the last three decades (1986-2013) and aims to explore the managers’ susceptibility to overpaying for its target. The sample of acquisition is taken from Thomson Financial SDC and includes U.S. completed deals with a transaction value exceeding $100 million dollars between 1986 and 2013. Both targets and acquirers are publicly-listed on NYSE, AMEX or NASDAQ. 1-week and 4-weeks acquisition premiums exhibit a significant downward trend in sample period 1986-2004 and exhibit an overall downward trend when controlling for deal characteristics. An event study is conducted to calculate the cumulative abnormal returns (CARs) and to test whether these coincide with the found trend. This research provides evidence for an upward trend for acquirer CARs, in which the acquisition premium act as an important variable in explaining the acquirer CAR trend. Target shareholders seem to gain significantly from an acquisition announcement through years, a significant upward trend for target CARs is proven. CARs are regressed on indicators to control for other factors determining the outcome of an acquisition announcement.
Keywords: Mergers and Acquisitions, Acquisition premiums, Announcement returns, Winner’s curse, Overpayment
3
Table of contents
Introduction ... 5
II. Related literature ... 8
2.1 Acquisition Motives ... 8
2.1.1 Synergy Hypothesis ... 8
2.1.2 Managerialism Hypothesis ... 8
2.1.3 Hubris Hypothesis ... 9
2.2 Overpayment ... 10
2.3 Announcement returns ... 11
III. Data and Methodology ... 13
3.1 Data ... 13
3.1.1 Sample of acquisitions ... 13
3.1.2 Sample summary statistics ... 14
3.2 Methodology ... 17
3.2.1 Trend estimation and Structural breakpoints ... 17
3.2.2 Event Study ... 18
3.2.3 Regression ... 19
3.2.4 Hypotheses development ... 22
IV. Time Trend ... 25
4.1 Acquisition premiums ... 25
4.1.1 Evidence from 1986-2013 ... 25
V. Results ... 27
5.1 Trend estimations in premiums ... 27
5.2 Explaining the trend in premiums ... 28
VI. Abnormal Announcement Returns ... 32
6.1 Estimating Trend Coefficients: Acquirer CARs ... 32
6.2 Estimating Trend Coefficients: Target firms CARs ... 36
VII. Discussion ... 38
VIII. Concluding remarks ... 39
4
5
Introduction
According to Wallstreet Journal (2013) U.S. companies have been paying a takeover premium averaging 19% above the targets company’s share price one week before the deals were announced. This is considered to be the lowest since at least 1995. Are these lower premiums a sign of companies becoming less subject to the winner’s curse, thereby overpaying less when acquiring potential target firms?
Mergers and acquisitions have been associated with premiums since their existence. These premiums represent the expected value that acquirers can extract from the target firm. The higher the price paid, the more value is expected to be generated from the investment. Yet, high premiums have not always been accompanied by value generation. In fact, premiums have been found to consume the expected synergies as managers tend to pay far more than what the acquisition is worth (Sirower, 1997). As a reaction to this overpayment, stock prices drop upon the announcement of the acquisition which is also known as the negative announcement effect.
An extensive body of research has been on the returns upon announcement of the merger or acquisition. Sirower (1994), finds a negative relationship between premiums and the returns for the shareholders of the acquiring company. The magnitude of the negative effect on the returns even increases with the paid premium. In addition, Fuller, Netter and Stegemoller (2002) find that bidder shareholders lose significantly as stock returns worsen upon announcing their planned acquisitions. Moeller, Schlingemann and Stulz (2005) have documented this negative announcement effect for a large number of U.S mergers and acquisitions. They find that acquisitions from the past decades have been associated with significant drops in stock prices. Economic scholars have explained this phenomenon by irrational behavior of the CEO such as empire building, misalignment of the interests of shareholders and managers and CEO overconfidence (Roll, 1986 and Black, 1989).
A major obstacle in identifying irrational behavior in the bidding process is the absence of a quantitative measure of overpayment due to subjectivity to the winner’s curse. Still, by examining the
6
development of announcement returns and its relation to the trend in acquisition premiums, information on behavioral corporate finance topics such as CEO’s hubris may be revealed.
Using U.S. merger and acquisition data sets this research explores the trend in acquisition premiums and aims to provide an explanation for plummeting acquisition premiums over time. Consistent with economic theory, lower acquisition premiums can be explained through a change in expected synergistic gains or bidding behavior. Acquirers may have improved their ability to value their targets more precisely, thus overbidding less (Roll, 1986). However, lower premiums could also just be a reflection of lower synergies.
Even though there is much consensus on the overpayment theory in explaining acquisition premiums, its development and influence on acquisition premiums through time is almost not to be found in economic literature. By investigating the evolution of acquisition premiums and its effect on announcement returns, the extent to which bidders ignore to adjust for the winner’s curse effect in auctions is examined. Moreover, this paper contributes to the existing literature by giving an insight of how overpayment has evolved over time and its contribution on diminishing acquisition premiums. By looking at the acquirers stock market reactions around the announcement dates, the extent to which the acquirer overvalues its target can be assessed.
Data on all U.S. mergers and acquisitions between 1986-2013 are collected from Thomson Financial SDC. Acquisition premiums in this thesis are defined as the difference between the market price 1 week or 4 weeks prior to the announcement and the actual price paid for the target. Yearly and monthly medians of premiums are constructed and evaluated carefully to decide on their development through time. Abnormal returns upon announcement are estimated over the following event windows [-1,+1], [-5,+5] and [-30,+30] using the standard event study methodology. Trend analyses are conducted to test for a pattern in premiums and acquirer abnormal returns to sketch the evolution of the winner’s curse hypothesis and its relation to the change in acquisition premiums.
7
The medians are plotted against time and show the development of premiums graphically in figure 1. As can be withdrawn from the figures, premiums seem to show a downward trend for at least a part of the sample period. The methodology used to investigate time-series patterns is based on Vogelsang’s (1998) and Brandt, Brav, Graham and Kumar’s (2009) trend estimation methodology and finds evidence for a significant negative coefficient for the trend variable for the period prior to Jan 2005. Furthermore, the break year of 2004 is followed by a significant upward trend in premiums. Controlling for deal characteristics leads to a significant downward trend in premiums for the entire sample period 1986-2013.
In order to emphasize on the relationship between the trend in premiums and abnormal returns, monthly averages of abnormal returns are presented graphically and again statistically tested for a trend. Even though a great part of the sample period was characterized by negative return, acquirer abnormal returns show an upward trend suggesting that acquiring shareholders were increasingly benefiting from the acquisitions. This finding combined with the downward trend in premiums could be interpreted as acquirers overpaying less.
This thesis is organized as follows: Paragraph 2 provides an overview of the existing literature on acquisition motives, acquisition premiums and announcement returns; Paragraph 3 and 4 presents the data sample and discusses the applied research methodologies; The hypotheses to be tested are described in paragraph 5 and are derived from the literature in section 2; Paragraph 6 presents the results of the several tests conducted in this research; Paragraph 7 discusses the results from paragraph 6; Finally paragraph 8 summarizes and concludes.
8
II. Related literature
2.1 Acquisition Motives
2.1.1 Synergy Hypothesis
Firms have several motives to acquire potential target firms. The main motive for acquisitions however are the expected synergistic gains. Acquisitions are valuable to the firm when the value of the combined firm is greater than the sum of the values of the individual firms. These gains are obtained from an increase in operational efficiency, market power, or from a financial gain and are distributed among the shareholders of the acquiring firm and of the target firm (Seth, Song and Pettit, 2000). Premiums paid in transactions should therefore reflect the value of these expected synergies by the acquirer. Yet little evidence has been provided that indicates a strong relationship between synergies and the paid premiums. As Hayward and Hambrick (1997) argue synergies may partially motivate acquisitions but cannot explain large acquisition premiums.
2.1.2 Managerialism Hypothesis
The Managerialism Hypothesis argues that managers may have the incentive to expand and increase the amount of assets through an acquisition or merger in order to maximize their own utility. Since the amount of assets under the manager’s control is a factor in deciding upon managerial compensation, managers are willing to pay a price above the target’s market stock price at the expense of shareholders (Seth et al., 2000). The existence of premiums caused by managers knowingly overpaying for targets is a product of managerial optimism and the misalignment of managers’ interests with those of their shareholders (Black, 1989).
Seth et al. (2000) also argue that diversification activities in an effort to reduce the risk associated with their human capital may provide an explanation to the fact that managers rationally overpay in acquisitions. Grinstein and Hribar (2004) argue that managers may overvalue large targets because they generally provide high private benefits.
9
2.1.3 Hubris Hypothesis
The Hubris Hypothesis first introduced by Roll (1986) states that acquisitions solely occur because managers tend to overvalue and thus make mistakes in evaluating target firms. According to Roll (1986), takeover premiums merely reflect random errors instead of synergistic gains. A possible explanation for the observed tendency of the bidder to overvalue its target lies within the winner’s curse. In a common-value auction setting, the bidder who offers the winning offer bid will find that on average he has overestimated the target’s value. This relates to the fact that in an environment with multiple bidders the probability of a successful takeover increases with the bidder’s offer. As a consequence, the bidder with the highest valuation, has the highest probability of a successful takeover and therefore wins the auction. As competition among prospective bidders for control of the target company increases, the offer price will be even set higher. This is reflected by a higher premium paid by the acquirer. Sudarsanam and Mahate (2006) find transactions in which the acquirer is the sole bidder to outperform multiple bidder contests. This is consistent with the observed decreased bidder returns as a cause of paying too much in a multiple bidder setting. The premium-setting decision is also influenced by the uncertainty regarding future prospects of the combination with the target company (Varaiya and Ferris, 1987). High uncertainty is reflected by a more volatile bidding process and more diverse offers, which in turn leads to higher prices.
More support for the existence of the winner’s curse in acquisitions is provided in the continuation of Varaiya and Ferris’ paper of 1987. Giliberto’s and Varaiya (1989) results indicate that winning bids tend to increase as the number of competitors increases as earlier mentioned. They also find that bidders overpay as a result of pursuing aggressive bidding strategies that fail in adjusting their bidding offer for the winner’s curse. In the most extreme case the Hubris Hypothesis predicts no synergistic gains from takeover bids and the entire premium paid to the target firm is considered to be a transfer from the acquirer to the target’s shareholders. Several studies provide empirical evidence for the negative announcement returns for the acquiring firm that supports this statement and will be discussed in the next paragraphs.
10
2.2 Overpayment
In general, acquisition premiums are defined as the difference between the price paid for the target firm’s stock by the acquirer and the target’s stock price before the announcement is made public. These premiums are justified when they do not exceed the potential synergies that arise from the acquisition or merger. Nevertheless, premiums have been found to consume the expected synergies as managers tend to pay far more than what the acquisition is worth (Sirower, 1997).
A number of explanations for this phenomenon is given by Black (1989) in his paper on bidder overpayment. The first explanation is in line with the winner’s curse emerging in an auction-like setting. Managers tend to increase their probability of winning by making high offers in order to outbid other potential bidders. Hereby ignoring the winner’s curse which is reflected by overpaying for the target firm (Black, 1989).
The second source of overpayment underlines the information asymmetry. Since the intrinsic value of the target is unknown to the potential acquiring firms, the target’s value is estimated based on the information that each of these bidders hold. This leads to an eventual pool of bidders that estimated the target’s value at a high enough value to stay in the game. This is reflected by price offers that exceed the intrinsic value of the target’s stock (Black, 1989). Further, managers that tend to be more optimistic are likely to overvalue the targets and make such high offers that they will be represented more frequently in the pool as well. Malmendier and Tate (2008) found that overconfident CEOs overestimate their ability to generate returns and as a result overpay for target companies and undertake value-destroying mergers. Moreover, this effect was found to be particularly pronounced for firms that have access to internal financing. Since overconfident managers overvalue their company’s stock, they find financing through issuing equity relatively expensive.
The third phenomenon explaining the existence of overpayment in acquisitions are principal-agent conflicts. There is much consensus in academic literature about the divergence and misalignment of the interests of shareholders and managers. While managers are more interested with letting the
11
company grow in order to feed their desire for greater prestige and visibility, shareholders are more concerned about the return of their investment in the company (Black, 1989). Their wish to grow may lead them to overinvest, either by acquisitions or mergers. To justify the high premiums, these managers may exaggerate potential gains (Black, 1989).
2.3 Announcement returns
When the announcement for an acquisition or merger is made public, different pieces of information can be obtained: potential synergies arising from the combination, the value of the acquirer and target firm, and, how the value will be split between the target and the acquirer shareholders. Overpayments and synergies are estimated using movements in the bidder’s and target’s stock prices around an offer announcement.
Since the main motive behind mergers and acquisitions are potential synergistic gains, the stock price should reflect the expected value creation immediately after the announcement. However, much evidence has been provided that contradicts this rationale as bidders’ stock prices drop. Consequently, this is used to accept the overpayment hypothesis and the managers’ susceptibility to the winner’s curse. Bidders tend to lose on average which is evidenced by negative stock announcement returns (Loderer and Martin, 1990; Bhagat and Hirschleifer., 1997).
As discussed, optimism may cause managers to overestimate the target’s value, leading the firm to pay too much for the acquisition. Bearing this in mind, shareholders would thus suffer from such overpayment which in turn is reflected by a fall in the acquirer’s stock price (Black, 1989). On the other hand, the shareholder of the target firm will benefit from the overpayment as they get more than what they should get, which will lead to a price increase. As Andrade, Mitchell and Stafford (2001) discuss in their paper on mergers completed during the period of 1973-1998, the most reliable evidence lies in the movement of stock price before and after the merger or acquisition announcement. The efficiency of the market ensures that target’s and acquirer’s stock prices quickly adjust and includes any expected value changes (Andrade et al., 2001). Moreover Andrade et al. (2001) find overall
12
negative announcement returns for the acquirer. It can be stated that acquisitions and mergers destroy much of the potential synergistic gains as acquirers tend to overvalue the target and thereby destroying shareholders value, resulting in these negative stock returns. On the contrary target firms are considered to be winners in mergers transactions as they benefit from high positive abnormal announcement returns.
Fuller et al. (2002) also observe bidders having significantly negative returns when buying public targets. The opposite holds when bidders buy private or subsidiary targets. In this case, acquiring shareholders benefit from a positive stock return. Faccio, McConnell and Stolin (2006) find that acquirers of listed targets earn an insignificant average abnormal return of -0.38%, while acquirers of unlisted targets earn a significant average abnormal return of 1.48%. Fuller et al. (2002) provide several explanations to this observation of differing market reactions. The most apparent explanation to these differing market reactions is the liquidity-effect. Private firms and subsidiaries are considered to be less liquid for they cannot be bought and sold easily as publicly traded firms (Fuller et al., 2002). The absence of liquidity makes acquiring firms view their targets less attractive and will translate this into a discount when acquiring private firms and subsidiaries. Officer (2007) documents discounts for acquisitions of unlisted targets up to 15% to 30% on average for comparable public transactions. Given the presence of these significant numbers, it can be concluded that liquidity constraints are associated with acquisition discounts.
13
III. Data and Methodology
3.1 Data
3.1.1 Sample of acquisitions
The takeover sample is developed from the complete list of all completed takeover deals announced to the trading floor in the United States. Data on deals are extracted from Thomson Financial SDC and include deals announced and completed between 1986 and 2013. Both targets and acquirers are publicly-listed on U.S. stock exchanges such as NYSE, AMEX, or NASDAQ. Privately held and government-owned companies are excluded from the sample. These firms are considered to be less subject to higher acquisition premiums caused by the winner’s curse. Also, cases in which the acquirer is the parent of the target are left out of the sample. Further variables of interest such as deal attitude, method of payment, relatedness of target and bidder, information on target/bidder stock prices are likewise withdrawn from Thomson One. Acquisition premiums are defined as the difference between the actual price paid per share minus the target’s share price prior to the announcement and are both available in Thomson One. Table 1 shows the median paid premiums per year varying in the period prior to the announcement. As can be deduced from table 1, signs of a pattern in premiums seem to be present.
The negative effect of offer announcements on the bidders’ stock prices will represent the degree of overpayment. In order to capture most of the announcement effect, only deals that have a transaction value of at least $100.0 million dollars will be taken in account. Moreover, the number of days between the announcement and the actual date at which the transaction is exercised is required to be less than
one year. The above described sample selection results in 2695 observations.
A negative stock return for the acquirer indicates overvaluation of the target in shareholders’ perspective. An event study will be conducted in which the effect on stock returns attributable to the announcement of the offer is examined. Information on the announcement and its perceived
14
overpayment are captured by the cumulative abnormal returns (CARs). The event study is carried out with Eventus over several estimation periods.
3.1.2 Sample summary statistics
In this paragraph the sample data are further examined. An overview of several deal details through the sample period is given. Panel A shows deal characteristics for the entire sample period, whereas panel B to D present deal characteristics in 6-year periods.
As can be drawn from panel A, most of the acquisitions are of friendly nature. Averaged acquisitions premiums vary approximately from 36% to 48% while median premiums tend to be significantly lower (28.93-38.76%). Acquisitions paid by cash solely account for 27.76%, whereas stock payments represent approximately 39% of the total sample set.
Table 1: Summary statistics
Table 1 presents a sample of completed acquisitions in the United States for the period 1986-2013. Both targets and acquirers are US public firms listed on US stock exchange and are involved in transaction with a value exceeding $100.0 million dollars. The transaction value excludes fees and expenses and takes only the total value of consideration paid by the acquirer in consideration. Relatedness takes the form of a dummy variable and is equal to 1 when the acquirer and target operate in the same industry. Hostile takeover is a dummy variable indicating whether the deal attitude was hostile. The same holds for friendly takeovers. The method of payment considers cash and stock payments and is again a dummy indicating whether the acquisition was paid in 100% cash or 100% stock. The takeover premium is defined as the run up in target’s stock price from 1 day, 1 week and 4 weeks prior to the announcement to the actual price paid by the acquirer.
Panel A: Summary statistics sample period 1986-2013
All (n=2695) Mean Median
Deal Characteristics
Deal Size (in Million $) 2055.47 523.79
1-day Takeover premium (in %) 36.36 28.93
1-week Takeover premium (in %) 41.18 33.33
4-weeks Takeover premium (in %) 47.68 38.76
100% paid in cash (in % total takeovers) 27.76 .
100% paid in equity (in % total takeovers) 38.63 .
Hostile Takeovers (in % total takeovers) 2.26 .
Friendly Takeovers (in % total takeovers) 97.18 .
Related Takeovers (in % total takeover) 62.97 .
15
Looking at the summary statistics by 6-year periods, a downward trend in hostile takeovers seems to be present. The last period shows no hostile takeovers, this could be due to the fact that this part of the time period consists of only 4 years. Sample period 2010-2013 could therefore be sensitive to aberrant results and should be interpreted carefully. While hostility in takeovers seems to be less observed through years, takeovers of friendly nature have increased since the late eighties from 90.32% of all acquisitions occurring between 1986-1991 to 98.74% during the sample period of 2004-2009.
Panel C shows other deal characteristics determining the outcome of a takeover. According to economic literature on merger and acquisition activities, cash payments and takeovers in the same industry result in higher acquisition premiums. Panel C exhibits an overall increase in related takeovers, suggesting an increase of the acquisition premiums paid for the acquisition. Cash payments show increasing observations starting in 1992, while stock payments are characterized by declining numbers starting in 1992.
Since averages are greatly influenced by outliers, monthly medians are calculated and averaged by year. While deal sizes have been increasing consistently since 1986 (panel C), panel D finds support for the existence of a cyclical trend in acquisition premiums. The data points are also plotted in graph 1, where a downward trend for at least a part of the sample period is observed. Several tests will be conducted to investigate the time-series patterns in acquisition premiums.
16
Table 1: Summary Stat. cont. Panel B: Distribution of
takeovers in 6-year periods
Total
Takeovers Hostile Takeovers
Friendly Takeovers Period N Percentage of total N Percentage of Total Takeovers N Percentage of Total Takeovers 1986-1991 279 10.35% 26 9.32% 252 90.32% 1992-1997 671 24.90% 18 2.68% 652 97.17% 1998-2003 937 34.77% 14 1.49% 916 97.76% 2004-2009 556 20.63% 3 0.54% 549 98.74% 2010-2013 252 9.35% 0 0.00% 250 99.21% Total 269 5 100% 61 2.26% 2619 97.18% Panel C: Deal characteristics of takeovers in 6-year periods Related Takeovers Cash
payments Stock payments
Deal Size Period N Percentage of Total Takeovers N Percentage of Total Takeovers N Percentage of Total
Takeovers Mean Median
1986-1991 86 30.82% 100 35.84% 98 35.13% 752.69 306.73 1992-1997 242 36.07% 108 16.10% 356 53.06% 1029.37 319.44 1998-2003 363 38.74% 185 19.74% 429 45.78% 2357.73 466.38 2004-2009 204 36.69% 229 41.19% 115 20.68% 2888.09 709.80 2010-2013 103 40.87% 126 50.00% 43 17.06% 2346.29 853.14 Total 998 748 1041 Panel D: Acquisition premiums (in %) in 6-year periods
1-day premium 1-week premium
4-weeks premium
Period Median Median Median
1986-1991 40.63 45.52135 53.173
1992-1997 32.46 35.943 41.784
1998-2003 33.15 38.983 45.041
2004-2009 32.33 32.8 37.368
17
3.2 Methodology
3.2.1 Trend estimation and Structural breakpoints
The trend estimation methodology used to investigate time-series patterns and their statistical significance is based on Vogelsang’s paper “Trend Function Hypothesis Testing in the Presence of Serial Correlation” (1998). In order to examine the trend in acquisition premiums, yearly and monthly medians for the 1-week and 4-weeks premiums from January 1986 through December 2013 are computed and plotted in figure 1 (Appendix).
[FIGURE 1 APPROXIMATELY HERE]
Monthly medians of 1-week and 4-weeks premiums are constructed and serve as dependent variables in the equations below. First lags of 1-week and 4-weeks premiums are constructed and are captured by the following time-series model. The variable of interest is the trend variable which counts the months since the beginning of sample period, Jan. 1986.
1) 𝑃𝑅𝐸𝑀1 𝑤𝑒𝑒𝑘𝑡= 𝑏0+ 𝑏1𝑡 + 𝑏2𝑃𝑅𝐸𝑀1 𝑤𝑒𝑒𝑘𝑡−1+ 𝜀𝑡
2) 𝑃𝑅𝐸𝑀4 𝑤𝑒𝑒𝑘𝑠𝑡= 𝑏0+ 𝑏3𝑡 + 𝑏4𝑃𝑅𝐸𝑀4 𝑤𝑒𝑒𝑘𝑠𝑡−1+ 𝜀𝑡
, where the premium is estimated according to the Ordinary Least Squares (OLS) method on its first lag and a time trend. The coefficients of interest are 𝑏1 and 𝑏3, which indicate the presence of a time trend
in case of a significant coefficient.
As there seems to be an apparent breaking point in the downward trend of acquisition premiums, several trend regressions are conducted to determine the break year. Assuming that the sample contains only one break point in the second half of the sample period 1986-2013, the first sample periods to be tested are 1986-2003 and 2004-2013 with break location Jan. 2004. The break year range is derived from the graphical presentation in figure and include 2003 to 2006. Equation 3a. represents the regression for a particular period prior to the break point, whereas equation 3b. represents the
18
regression for the period after the break point. The break location is represented by 𝑏 in the equation stated below, 𝑚 represents the last month of the sample period.
3a) 𝑃𝑅𝐸𝑀1 𝑤𝑒𝑒𝑘
𝑡= 𝑏0+ 𝑏1𝑡 + 𝑏2𝑃𝑅𝐸𝑀1 𝑤𝑒𝑒𝑘𝑡−1+ 𝜀𝑡 , 𝑡 = 1 … 𝑏
3b) 𝑃𝑅𝐸𝑀1 𝑤𝑒𝑒𝑘
𝑡= 𝑏0+ 𝑏1𝑡 + 𝑏2𝑃𝑅𝐸𝑀1 𝑤𝑒𝑒𝑘𝑡−1+ 𝜀𝑡 , 𝑡 = 𝑏 … 𝑚
In addition formal break point tests are conducted to verify the significant change in the trend coefficients after the break year found under the above mentioned methodology. The Chow test allows to test whether this particular location in the sample period causes a change in the regression coefficients. A dummy variable is created that is represented by the number of 1 if the date exceeds the break point. Interaction terms between the dummy variable and lags of the independent and dependent variables are constructed and are used as input for the regression models below.
4) 𝑃𝑟𝑒𝑚𝑖𝑢𝑚1 𝑤𝑒𝑒𝑘 =∝ + 𝑏 1𝑃𝑟𝑒𝑚𝑖𝑢𝑚𝑡−11𝑤𝑒𝑒𝑘+ 𝑏2𝑀𝑜𝑛𝑡ℎ𝑡−1+ 𝐷1𝐵𝑟𝑒𝑎𝑘 + 𝑏3𝑃𝑟𝑒𝑚𝑖𝑢𝑚𝑡−11𝑤𝑒𝑒𝑘∗ 𝐵𝑟𝑒𝑎𝑘 + 𝑏4𝑀𝑜𝑛𝑡ℎ𝑡−1∗ 𝐵𝑟𝑒𝑎𝑘 + 𝑒𝑡 5) 𝑃𝑟𝑒𝑚𝑖𝑢𝑚4 𝑤𝑒𝑒𝑘𝑠 =∝ + 𝑏 5𝑃𝑟𝑒𝑚𝑖𝑢𝑚𝑡−14𝑤𝑒𝑒𝑘𝑠+ 𝑏6𝑀𝑜𝑛𝑡ℎ𝑡−1+ 𝐷1𝐵𝑟𝑒𝑎𝑘 + 𝑏7𝑃𝑟𝑒𝑚𝑖𝑢𝑚𝑡−14𝑤𝑒𝑒𝑘𝑠∗ 𝐵𝑟𝑒𝑎𝑘 + 𝑏8𝑀𝑜𝑛𝑡ℎ𝑡−1∗ 𝐵𝑟𝑒𝑎𝑘 + 𝑒𝑡
3.2.2 Event Study
Following Brown and Warner (1985), similar methodologies in examining the bidder returns are used. In order to examine the effect of the event on the bidder’s stock prices, abnormal returns are estimated with the mentioned standard event study method. The methodology is based on the market efficiency theory, in which the stock price reflects all information related to the firm’s value. Abnormal returns are estimated over the three-day event window (-1,+1), (-5,+5) and (-30,+30) where market model benchmark returns are defined by the CRSP equally weighted index returns. The date at which the acquisition is announced is considered to be the event day. The benchmark return is calculated using the market model.
19
, where 𝑎𝑖 and 𝛽𝑖 are Ordinary Least Squares values from the estimation period. Abnormal returns are
calculated by subtracting the value-weighted CRSP market return from the firm’s return. Equation 3 represents the excess return of stock i on a particular day t.
2 𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡− 𝑅𝑏𝑒𝑛𝑐ℎ𝑚𝑎𝑟𝑘
3 𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡− (𝑎̂𝑖𝑡+ 𝛽̂𝑖𝑡)
To capture the announcement effect over the [-1,+1] event window, the sum of abnormal returns is computed by adding up the daily abnormal returns. The CAR is subsequently calculated for the other time windows namely [-5,+5] and [-30,+30]. To test for the pattern in CARs, a regression will be conducted in which a trend variable is included. If found significant, evidence is found that supports the existence of a trend. The next paragraph discusses the regression testing more in detail.
4 𝐶𝐴𝑅𝑖(−1,1) = ∑−1+1𝐴𝑅𝑖𝑡 5 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝐶𝐴𝑅𝑖 = 1 𝑛∑ 𝐶𝐴𝑅𝑖𝑡 −𝑡 +𝑡
To obtain daily stock returns, CUSIP identification codes were translated to PERMNO codes and were used next to announcement dates as an input to the event study software Eventus.
3.2.3 Regression
Essential to this study is the effect of the development of premiums on the evolution of acquirer and target CARs. In order to capture the development of CARs through years, a trend variable is constructed and regressed against monthly averages of CARs. Again, a significant positive trend coefficient indicates an upward trend. In this case, shareholders were increasingly benefiting from acquisition announcements. To control for other factors determining CARs, multivariate tests on the determinants of acquirers returns are conducted. An OLS regression method is used to determine signs of coefficients of other independent variables determining stock price returns. The dependent variable is the monthly average of CARs from one day (5 and 30 days) prior the event date to one day (5 and
20
30 days) after the event date. Other explanatory variables suggested by economic scholars are also included in the regression.
6) 𝐶𝐴𝑅̅̅̅̅̅̅𝑡 = 𝛼 + 𝛽1∗ 𝑇𝑟𝑒𝑛𝑑 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑡+ 𝛽2∗ 𝑃𝑅𝐸𝑀𝑡+ 𝛽3∗ 𝐷𝑒𝑎𝑙 𝑆𝑖𝑧𝑒𝑡+ 𝛽4∗
𝐶𝑎𝑠ℎ 𝑆𝑒𝑡𝑡𝑙𝑒𝑚𝑒𝑛𝑡𝑡+ 𝛽5∗ 𝑅𝑒𝑙𝑎𝑡𝑒𝑑 𝑇𝑎𝑘𝑒𝑜𝑣𝑒𝑟𝑡+ 𝛽6∗ 𝑆𝑡𝑜𝑐𝑘 𝑆𝑒𝑡𝑡𝑙𝑒𝑚𝑒𝑛𝑡𝑡+ 𝜀𝑡
As Travlos (1987) discusses in his paper, the method of payment has certain implications for the acquirers return upon announcement. Since acquiring managers will use stock as a method of payment when it is overvalued, the shareholders will respond to it negatively causing the price to drop. He finds lower abnormal returns for acquisitions of public firms financed with stock, while cash payments are associated with higher abnormal returns (Travlos, 1987). Cash Settlement represents the percentage of 100% cash payments in a month. Stock Settlement follows the same construction and indicates the percentage of 100% stock payments in a month.
A takeover is considered to be hostile when the target’s management refuses to sell its company to the potential acquirer. Hostile takeovers have also been proven to have a boosting effect on the purchase price, maximizing shareholder’s value (Demidova, 2007). Moreover, Martynova and Renneboog (2011) provide evidence stating that these hostile acquisitions are also marked by lower announcement returns to the acquirer. Deal Attitude will however not be taken in account as the amount of hostile takeovers in a month is negligible small.
In addition, information on the deal size of the transaction is included in the regression. Loderer and Martin (1990) find that larger deals destroy more value than smaller transactions. This leads to acquirers losing more money in these transactions, which is subsequently translated to lower announcement returns for acquiring shareholders. Deal Size represents the total value of consideration paid by the acquirer for the target.
Strategic buyers are considered to gain more from potential synergies, and are likely to pay higher prices. Morck, Schleifer and Vishny (1990) argue that when the target does not operate in the same industry, the acquisition announcement is followed by negative returns to the acquiring shareholders.
21
Related Takeovers assesses the resemblance of business nature between these strategic buyers and
their financial competitors.
Since this paper investigates the trend in the overbidding tendency by acquirers, a trend variable is added to control for the months to be examined. A significant negative coefficient shows evidence for a downward trend while a positive coefficient indicates an upward trend. The table below provides an overview of the independent variables and a brief description.
Table 2: Overview independent variables
Variables Description
Deal Size
Deal Size is defined as the median of the values of consideration paid by acquirers, excluding fees and expenses for a certain month
PREM
Premiums are equal to the offer price to the target’s stock price 1-week prior to the announcement
Related Takeover
Related acquisitions are distinguished by equal SIC code for acquirer and target and represent the monthly percentage of acquisitions meeting this description
Cash Settlement Stock Settlement
Monthly percentage of acquisitions in which the payment occurred in 100% cash
Monthly percentage of acquisitions in which the payment occurred in 100% stock
Trend Variable
This trend variable is included as a proxy for change in observed abnormal returns, and counts the months since Jan. 1986
22
3.2.4 Hypotheses development
Building upon the theories provided in section 2, this paragraph discusses the hypotheses to be tested. The key objective of this research is examining the effect of the development of acquisition premiums on the susceptibility to the winner’s curse of acquirers through years. First, the development of acquisition premiums and announcement returns for both target and acquiring firms from 1986-2013 are explored through trend analyses. Next, the cumulative abnormal returns will be analyzed to provide evidence on the acquirers ability to value their target correctly and their relation to the found trend in premiums. Finally, specific deal characteristics discussed in the previous paragraph and their influences on the development of abnormal returns are examined.
Following this analogy the hypotheses are stated as follows:
Acquisition premiums: Trend analysis
𝐻1: Acquisition premiums show a downward trend from 1986-2013
The key objective of this research is to investigate whether the downward trend is a result of companies being less subjective to the winner’s curse. To capture the movement of acquisition premiums paid in acquisitions occurring in the period 1986-2013, median premiums are computed and plotted against a time variable. To test whether the premiums show a downward trend as the Wallstreet Journal article (2013) indicates, a trend variable is constructed and tested for its significance and sign. A significant negative coefficient corresponds with the statement in the news article and shows evidence for a downward trend.
Abnormal announcement returns: Trend analysis
𝐻2: Abnormal Announcement returns for acquirers have become less negative over the period 1986-2013
23
In order to examine whether a downward trend in acquisition premiums is accompanied with less susceptibility to the winner’s curse, the following hypotheses with regard to the abnormal announcement returns are tested. The development of the effect of the announcement on the bidder’s and acquirer’s stock returns are examined. As explained by the economic theories stated in paragraph 2, targets tend to benefit from a takeover while acquiring shareholders experience a negative abnormal return upon announcement. The positive announcement returns for target shareholders represent the value created as acquirers tend to overpay for an acquisition. The negative announcement returns for acquirers are also a cause of the price paid for an acquisition, however are represented by a negative sign as the acquisition is seen as value destruction rather than creating value for the acquiring shareholders. The higher the premium paid, the higher the magnitude of the effect on stock returns, and vice versa. The calculated CARs are regressed on a time variable to determine the trend in announcement returns of acquirers and targets. A positive significant coefficient indicates increased announcement returns over time, while a negative significant coefficient shows a sign of a downward. Monthly CAR averages will also be regressed on deal characteristics mentioned in the previous section. The development of the hypotheses regarding deal characteristics is provided in the next paragraph.
Deal characteristics
𝐻4: The sample period 1986-2013 is characterized by a negative relation between acquisition premiums and acquirer’s abnormal returns and explain a significant part of the trend in CARs
Higher premiums are often associated with managers overpaying for targets. As a consequence, stock prices fall as soon as the announcement is public. This finding leads to the hypothesis of the trend in premiums being greatly responsible for the evolution in CARs. A negative relationship between these two variables is expected.
24
𝐻5: The sample period 1986-2013 is characterized by a positive relation between cash settlements and acquirer’s abnormal returns and explain part of the trend in CARs
The method of payment is one of the significant factors determining the announcement return. Cash payments are considered to generate higher shareholder value for acquiring companies and are therefore associated with less negative or even positive returns for acquirers. It can be hypothesized that the development of the acquirer’s abnormal returns holds a positive relationship with cash bids.
𝐻6: Greater deal sizes cause stock prices to react negatively resulting in negative abnormal returns of the acquirer and explain part of the trend in CARs
The waste caused by overpayment tends to increase as the deal value increases. Greater deal sizes increase the risk of perceived overpayment and thus negative abnormal returns for acquiring shareholders. Higher deal sizes may therefore cause stock prices to drop more, and vice versa.
𝐻7: Acquiring targets in the same industry causes stock prices to react negatively resulting in negative
abnormal returns the acquirer and explain part of the trend
Strategic buyers are considered to acquire targets at higher prices. As a result, stock prices tend to react negatively upon the announcement of the acquisition of a potential target operating in the same industry. As a result, an upward trend in acquirer CARs may be associated with less related takeovers over time.
25
IV. Time Trend
4.1 Acquisition premiums
4.1.1 Evidence from 1986-2013
Figure 1 represents time-series plots of aggregate 1-week and 4-weeks acquisition premiums and include the monthly moving average of the monthly median of the acquisition premiums. As can be withdrawn from the graph, monthly median do show signs of a downward trend for at least a part of the sample period. The second part of the sample period seems to be characterized by increasing premiums pointing to a break year. This downward trend is even stronger when yearly medians are computed and plotted. According to figure 1, paid premiums were more volatile post 2006. Furthermore, the downward trend seemed to be followed by an increasing trend from approximately 2006 and on.
As can be drawn from the results from figure 1 and table 1 acquisitions premiums show peaks during economic recessions. The sample period 1986-2013 include three economic recessions in the following years: 1990-1991, 2000-2001 and 2008-2009. These recessions are characterized by an average monthly median of the 1-week acquisition premiums of 55.63%, 41.60% and 45.22%, respectively. The same peaks are observed for the 4-week premiums. Comparing these premiums to the average premiums paid 3 years prior to these economic busts, acquirers pay 7.0% more in these recessions than what they would be willing in previous years. The tendency to pay higher premiums during these economic meltdowns can be explained by the fact that targets are more hesitant in accepting deal offers and thus bargain harder for higher prices. This is because target firms believe that current prices are temporarily depressed and thus do not reflect their true value (Baker, Pan and Wurgler, 2009).
Following the recession year of 1991, the 1991-1997 period alone shows a clear downward trend for nearly 7 years as the economic climate ameliorates. The same pattern of declining premiums is found after the recession year of 2000 and reach an averaged median of 32.41% in the subsequent years prior to the next recession. This points to a further decrease since the recession of 1990-1991, where
26
the averaged median in the subsequent years to the next recession was equal to 36.97%. The downward trend seems to be interrupted by a deterioration of the American economic situation, which could also indicate a cyclical pattern. The Great Recession of 2007-2009 exhibits a more volatile development of premiums than previous recessions. During the latter recession firms paid a maximum 1-week premium of 79.9% in December 2008 while six months earlier they were paying a median premium of 13.9%. Although historic evidence provided in table 1 concise with high premiums being an episodic phenomenon instead of a trend, trend tests are conducted to provide evidence for a trend during at least a part of the sample period.
27
V. Results
5.1 Trend estimations in premiums
Table 2 presents the results of the 1-week and 4-week regressions estimating the trend and lagged premium coefficients. The main interest is in the trend variable, which captures the movement of premiums over one unit of time (months). Looking at the 1-week trend estimates for the entire sample period, no statistical evidence can be extracted that supports the existence of a significant downward trend. Unlike 1-week premiums, 4-week premiums show a significant downward trend, indicating that premiums have been declining by 0.023% per month since 1986.
[TABLE 2 APPROXIMATELY HERE]
The regression is also performed for the sample period prior to and after several breaking time locations. In order to identify the strongest breaking point, the fit of the models is maximized by adding up the coefficients of determination (R²) of the model before and after the break year. The following years are found to be significant break locations: 2003, 2004 and 2005. All trend estimations prior to these break points are represented by negative coefficients and are highly statistically significant. The downward trend is followed by positive significant trend coefficients and are besides of greater magnitude, indicating a steeper slope from the breaking point and on. The strongest break point is identified as January 2005, where the trend coefficients for the 1- week and 4-weeks premiums prior 2005 are estimated at -0.040 and -0.049, respectively. This shows that premiums were declining by 0.040% and 0.049% per month from 1986 till 2004, followed by a monthly increase of 0.147% and 0.120% from 2005 and on.
[TABLE 3 APPROXIMATELY HERE]
The Chow test concludes and provides statistical evidence to reject the null hypothesis. F-statistics obtained from the Chow test for the 1-week and 4-weeks premiums are equal to 5.03 and 3.40, respectively. The found F-values are larger than the critical value of F, which implies that the null
28
hypothesis of structural stability can be rejected. Hereby, it can be assumed that the coefficients change significantly after this particular point in time.
Table 3: Chow stability test F-statistics A.
Structural break tests for Jan 2005
1-wk PREM
4-wks PREM
Basic Chow Test
3.40 (p<0.05)
5.03 (p<0.01) Notes: F-crit at a 5% significance=2.421, the chow test was conducted according to equation F(3, 327). The break year of 2005 for 1-wk premiums and 4-wks premiums were denoted statistical significance at the 5%
and 1% levels, respectively.
The visual evidence in figure 1 is supported by the results in table 3 and corresponds with the hypothesis that premiums have been declining for at least a part of the sample period. In order to provide an economic rationale that explains the trend and the break point in 2004, factors that determine premiums will be analyzed. The break point is also taken in account when analyzing these determinants and leads to a breakdown of the entire sample period into two samples with sample period 1986-2004 and 2005-2013.
5.2 Explaining the trend in premiums
To seek understanding in the behavior of premiums and its drivers, a regression is performed that aims to capture a more accurate development of premiums over time. Many factors may exist that influence the paid premiums and therefore the movement in premiums. By including some of these determinants (regression 7), a more accurate movement of premiums is captured. Besides the trend may also be partly explained by these factors, which will also be shown in the regression results. Deal characteristics such as deal attitude, method of payment, relatedness, number of bidders and target size have been found to influence acquisition premiums significantly. It should be noted that information on the number of bidders is very limited for the given sample set and will therefore be disregarded. The variables discussed under section 3.2.3 are extended with two additional variables found to explain premiums as well.
In order to assess a more accurate development of premiums through time, the GDP growth is included to correct for business cycles. A high GDP growth can be associated with increased confidence leading
29
to an overestimation to extract synergies from the acquisition and thus higher premiums. Moreover, a high GDP growth may be associated with a higher level of competition in the bidding process. As discussed in section 2, a higher level of competition is considered to push up bidding offers, leading to higher purchase prices. Favorable economic conditions may consequently have a positive effect on the presence of multiple bidders and therefore increase the competitive nature of the auction. This leads to a higher price paid by the winner of the auction.
The size of the target firms is also found to have a negative effect on acquisition premiums. Acquirers tend to pay significantly lower premiums when high values are at stake. As a result managers are more hesitant to pay higher premiums in acquisitions. The complexity of integrating large firms may also affect the expected synergies negatively as it leads to more uncertainty (Hayward, 2002). The following regression captures the above mentioned variables and is carried out for the sample period 1986-2013:
7) 𝑃𝑟𝑒𝑚𝑖𝑢𝑚𝑖𝑡 = 𝛼𝑖𝑡+ 𝛽1∗ 𝑇𝑟𝑒𝑛𝑑𝑡+ 𝛽2∗ 𝐺𝐷𝑃 𝐺𝑟𝑜𝑤𝑡ℎ (%)𝑡+ 𝛾1∗
𝑅𝑒𝑙𝑎𝑡𝑒𝑑𝑛𝑒𝑠𝑠𝑖+ 𝛾2∗ 𝐶𝑎𝑠ℎ 𝑃𝑎𝑦𝑚𝑒𝑛𝑡𝑖𝑡+ 𝛾3∗ 𝑆𝑡𝑜𝑐𝑘 𝑃𝑎𝑦𝑚𝑒𝑛𝑡𝑖𝑡+
𝛾4∗ 𝐻𝑜𝑠𝑡𝑖𝑙𝑒 𝑇𝑎𝑘𝑒𝑜𝑣𝑒𝑟𝑖𝑡+ 𝛽5∗ 𝑇𝑎𝑟𝑔𝑒𝑡 𝑆𝑖𝑧𝑒𝑖𝑡+ 𝜀𝑖𝑡
Table 4 shows the revised trend coefficients for the 1-week and 4-week premiums corrected for the earlier mentioned factors. A significant downward trend in 1-week and 4-weeks premiums for the entire sample set is proven after controlling for other variables. 1-week premiums were significantly decreasing with 0.045% while 4-week premiums were decreasing at an equal rate of 0.045% per month.
[TABLE 4 APPROXIMATELY HERE]
In accordance with prior research a negative coefficient for target size is presented in table 4. This could be translated to reticence to pay high premiums for large targets. A dollar increase in the target’s stock price reduces the paid premium by 0.288% and 0.373% for the 1-week and 4-week premiums, respectively. However, the trend coefficient does not seem to get less significant when adding the
30
variable to the regression. Therefore it can be concluded that the trend in premiums is not explained by target size.
Hostile takeovers are found to increase 1-week and 4-week premiums by 12.62% and 9.57% respectively. An explanation for the downward trend in premiums is provided by the attitude of the deal. By including the variable indicating hostile attitude in a takeover, the trend coefficient’s slope and explanatory power weakens. This corresponds to table 1. Panel B. where a downward in hostile takeovers is detected. The observed lower premiums were indeed associated with less hostility and also explain part of the downward trend in acquisition premiums.
Transactions that are settled in 100% cash end up with 3.55% above the 4-weeks premiums. This corresponds to shareholders demanding higher premiums in situations that will force them to pay immediate taxes on their gains. The effect of GDP growth rate shows a positive effect on the 4-week premiums. Again, the significant relation between cash payments and premiums raises the question whether the downward trend could be explained by less cash payments through time. Given the summary statistics provided in section 3.1.2., no inducements are provided to believe that cash payments might have caused declining premiums. The results from table 4 corresponds with the summary statics and provide no evidence that shows that declining premiums were partly explained by declining cash payments.
An aberrant result is shown by the variable indicating the related nature of business of the target and acquirer. Instead of the assumed positive relationship, table 4 shows negative signs for the coefficients and are significantly different from zero at a 10% level. Lower premiums were paid for targets in the same industry, which contradicts the belief that strategic buyers are willing to pay higher premiums for targets. Table 1 shows that an increasing share of the acquisitions took place in the same industry. Given this observation and its significant negative relation with premiums, it is expected to see its effect on the trend in premiums. Table 4 concludes and shows the trend’s significance weaken as the
31
variable is included in the regression, which indicates that the type of takeovers partly explains the trend in premiums.
However, as the trend coefficient stays significantly negative one can conclude that these deal characteristics do not explain the whole trend. Also, the trend break seems to disappear when including deal characteristics so that the downward trend applies to the entire sample period instead of the earlier assumed period of Jan 1986-Jan 2005.
In order to state whether the downward trend was accompanied by higher cumulative abnormal returns to the acquirer, an event study is conducted in which returns upon announcement are calculated. The tendency to overpay for an acquisition is examined through abnormal returns over time. Acquisitions in which the bidding process is less influenced by the winner’s curse will be characterized by higher cumulative abnormal returns for acquiring shareholders. In contrast to acquiring shareholders, CARs from target firms are considered to hold a positive relationship with the trend in acquisition premiums in case of less overpayment. The next paragraph discusses the abnormal performance of both acquiring firms and target firms. Monthly averages of acquirer and target CARs are calculated and presented graphically. Results from CAR trend estimations are linked to hypotheses 2 and 3, regarding the abnormal announcement returns.
32
VI. Abnormal Announcement Returns
6.1 Estimating Trend Coefficients: Acquirer CARs
Central to this thesis is assessing the development of the tendency of acquirers to overpay for target firms by examining the trend in CARs and comparing this with the movement of premiums. The regression results in the previous section show a consistent downward trend in acquisition premiums. This paragraph explores the behavior of announcement returns through years and aims to explain the relation between the found trend and declining premiums.
As discussed in the previous paragraphs negative abnormal returns for the acquirer are often explained by overpayment. In this case the bidder pays a price above the target’s intrinsic value and transfers its shareholders’ wealth to the target firm. Andrade et al. (2002) discuss in their paper that the most reliable evidence for overpayment lies in the movement of stock prices before and after the announcement. Given the above stated, it can be hypothesized that a decline in acquisition premiums is accompanied with less overpayment. Less overpayment would be reflected by less negative abnormal returns to the bidding firm and less positive abnormal returns to the firm to be taken over. Given the downward trend in premiums for the sample period 1986-2013, acquirer abnormal returns would show an upward trend in case of less overpayment. The following event windows are used to calculate acquirer’s abnormal returns by year: [-1,+1], [-5,+5] and [-30,+30]. Since no outliers are observed, yearly and monthly averages are calculated and plotted in figure 2.
[FIGURE 2 APPROXIMATELY HERE]
Looking at the yearly acquirer CAR averages derived from the several estimated event windows, an upward trend seems to be existent especially during the second half of the sample period. For the sample period 1986-2004 where a downward trend in acquisition premiums has been proven, [-30,+30] and [-5,+5] CARs show an increase only from 1986 to approximately 1995. The upward trend is followed by a drop to then rise again, showing signs of a cyclical pattern instead of an upward trend. Noteworthy is the consistent downward trend for the [-1,+1] CARS from 1986 to approximately 2001,
33
suggesting that lower acquisition premiums were not accompanied by higher stock returns for acquiring shareholders as has been hypothesized. The downward trend in premiums, show a positive relationship with the [-1,+1] monthly CARs which could find some of its explanation in the Synergy Hypothesis. According to the hypothesis premiums reflect the value of potential synergies. The lower premiums could therefore merely be an indication of decreased (perceptions of) synergistic gains but may still result in lower stock returns as the overpayment share in premiums increases. Moreover, the abnormal returns tend to react more negatively as the days prior to announcement decreases. This could be a result of more accurate deal information coming out.
The yearly CAR averages from all estimation windows exhibit a striking result from 2001. An upward movement in CARs is observed, while premiums were proven to decline. A possible explanation corresponds with the hypothesis regarding the abnormal returns; acquirers have been offering prices closer to the target’s intrinsic value. By acquiring the target at a price which lies closer to its true value, the acquirer is considered to destroy less value which is translated in higher acquirer abnormal returns. However, it may be clear from the graphical presentations in figure 2 that lower premiums are not always associated with higher acquiring stock returns. This is because stock returns hold numerous pieces of information of which overpayment is just a fraction of the total information set contained in stock prices.
Turning to the existing evidence where negative abnormal announcement returns for a large number of U.S. mergers and acquisitions is documented, the graphs in figure show more encouraging results for U.S. acquirers. As the graphs exhibit, the negative returns upon announcement become less negative and even turn positive for the [-1,+1] and [-5,+5] estimation windows by 2010 and 2011, respectively. Monthly CAR averages are analyzed as well but show less strong movements. Looking at the sample period 1986-2013, no clear upward trend in CAR averages can be derived from figure 2.
Statistical evidence for the presence of a trend is provided when monthly CAR averages are regressed on a trend variable (equation 8). Table 5A. contradicts the visual display of CARs where no clear upward
34
trend is observed. Moreover, monthly averages seem to behave consistently through time for [-1,+1] and [-5,+5] event windows, becoming less negative as time develops. Interpreting the regression results in table 5A., abnormal returns upon announcement were increasing with 0.0049% and 0.0057% per month, respectively.
8) 𝐶𝐴𝑅̅̅̅̅̅̅̅ = 𝛼 + 𝛽𝑡 1∗ 𝑇𝑟𝑒𝑛𝑑 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒 + 𝜀𝑖
[TABLE 5A. APPROXIMATELY HERE]
Since stock returns also hold deal specific information of which its significant influence on stock prices has been proven, this trend may overestimate the acquirer’s improved ability to value its target more accurately. In order to capture a more accurate trend of the CARs, deal characteristics are included in the regression. By adding these variable one at a time, the drivers of the trend can be examined. The regression stated below aims to explain the movement in CARs and captures the trend corrected by these independent variables.
9) 𝐶𝐴𝑅̅̅̅̅̅̅̅ = 𝛼 + 𝛽𝑡 1∗ 𝑇𝑟𝑒𝑛𝑑 𝑉𝑎𝑟𝑖𝑎𝑏𝑙𝑒𝑡+ 𝛽2∗ 𝑃𝑅𝐸𝑀𝑡+ 𝛽3∗
𝐷𝑒𝑎𝑙 𝑆𝑖𝑧𝑒𝑡+ 𝛽4∗ 𝐶𝑎𝑠ℎ 𝑆𝑒𝑡𝑡𝑙𝑒𝑚𝑒𝑛𝑡𝑡+ 𝛽5∗ 𝑅𝑒𝑙𝑎𝑡𝑒𝑑 𝑇𝑎𝑘𝑒𝑜𝑣𝑒𝑟𝑡+ 𝛽6∗ 𝑆𝑡𝑜𝑐𝑘 𝑆𝑒𝑡𝑡𝑙𝑒𝑚𝑒𝑛𝑡𝑡+ 𝜀𝑖
[TABLE 6 APPROXIMATELY HERE]
There is much consensus on the tendency of acquirers to overpay for targets. These overpayments are characterized by high premiums leading to an assumed negative relationship between premiums and acquirer stock returns upon announcement. Looking at table 6 where the results from the regressions are provided, 1-wk premiums are marked by their expected negative signs and are for all estimation windows significantly different from zero at a confidence level of 5%. The negative relation between premiums and acquirer abnormal returns lead to lower stock returns. This confirms the hypothesis that premiums are perceived as overpayment instead of a reflection of expected synergies only. Furthermore, by including the premium variable the explanatory power of the trend variable seems to
35
weaken, indicating that a significant fraction of the trend is captured and explained by the movement in premiums.
Consistent with the literature, deal size has a significant negative effect on the CARs. However, no effect on the observed trend in CARs is detected. Adding the variable deal size to the regression merely increases the trend coefficient and its significance in explaining the monthly averages of the CARs. The same reasoning leads to the belief that cash settlements and the related nature of targets also have their part in explaining the development in the CARs estimated within the [-5,+5] time frame. The estimated trend coefficients in column 6 of table 6A. 6B. and 6C, all show significant positive signs after controlling for the control variables. The abnormal returns [-1,+1], [-5,+5] and [-30,+30] increase by 0.0044%, 0.0055% and 0.0138% per month, respectively. Given these results, acquiring shareholders were increasingly benefiting from acquisitions through the years. The perception that acquiring shareholders lose in general is not necessarily contradicted as CARs remain negative for the greatest part of the sample period.
Even though, the trend estimates in column 6 are considered to be a purer estimation of the real trend, they should still be handled with caution when using them as a measure for overpayment. This is because acquirer CARs may have other unobserved determinants explaining the trend that are not captured in the regression. Therefore the development in CARs may only serve as an estimate of irrational behavior instead of a measure for assessing the evolution of overpayment. The next paragraph analyzes the trend in target abnormal returns and compares it to the found downward trend in premiums.
36
6.2 Estimating Trend Coefficients: Target firms CARs
In contrast to acquirer shareholders, declining acquisition premiums should reflect less value creation for the target shareholders since they receive less premium. Figure 3 showing abnormal returns upon announcement for target firms does seem to show support for the hypothesis regarding the downward trend in target abnormal returns only for the sample period until approximately 1996. Declining premiums during this period were accompanied by lower abnormal returns for the target shareholders as hypothesized. However, this positive relationship does not seem to exist throughout history. The second half of the sample period is characterized by increasing abnormal returns, indicating value creation to the shareholders after announcing the acquisition or merger as economic literature discusses. Based on the plotted averages in figure 3, a clear relationship between the development of acquisition premiums and CAR averages is however not found.
[FIGURE 3 APPROXIMATELY HERE]
A regression is conducted that captures the trend in target CARs during this period of declining premiums. The regression results seem to support the rejection of the hypothesized downward slope of target CARs. The significant positive trend coefficients for the [-30,+30] and [-5,+5] presented in table 5B. prove the consistent increasing profits for target shareholders, as has been discussed in extensive research papers. Shareholders in target firms seem to gain significantly at the announcement of takeovers through time. Even though, the upward trend in target CARs does not reveal the nature of these paid premiums it bears an interesting outcome. Several researchers have documented that target firms tend to benefit from acquisitions as they receive a premium above the value of the share. With declining premiums one would expect an equal downward trend in CARs instead of the observed increasing target abnormal returns.
[TABLE 5B. APPROXIMATELY HERE]
By including the same determinants as in the previous section the influence of these deal characteristics on the trend is examined. According to table 7 the percentage of cash payments holds
37
a positive relationship with the monthly target CARs. This finding is consistent with the tax explanation. Shareholders demand higher premiums in situation that will force them to pay immediate taxes on their gains (Huang and Walkling, 1987). The higher premiums lead to a jump in the target’s stock price and thus positive abnormal returns. The percentage of cash payments also seems to explain the trend in monthly target CAR averages. By adding the cash variable to the regression, the trend coefficient’s significance declines, thereby losing some of its explanatory power.
On the other hand, stock settlements seem to have the opposite effect on target stock returns. This is in line with the Method of Payment Hypothesis which states that managers prefer to settle the deal in stock when their own stock is overvalued or the target firm’s stock is undervalued. Looking at its effect on the trend, stock settlements also seem to explain some of the trend in CARs. The other control variables that have been proven to have a significant effect on target CARs do not explain the upward trend in CARs. Column 6 in table shows significant positive trend coefficients even after controlling for deal characteristics. Monthly increases for [-30,+30] and [-5,+5] CARs are equal to 0.0367% and 0.0276%. This contradicts the hypothesis stating a downward trend in CARs, suggesting that target CARs were increasingly benefiting from acquisition announcements.
