January, 2018 University of Groningen. Faculty of Economics and Business. MSc Finance.
_______________________________________________________________
MSc Finance Thesis
Are acquisitions a destruction of shareholder value?
Author:
J. Wichink Kruit - S3026345 -
J.M.Wichink.Kruit@student.rug.nl
Supervisor:
prof. dr. R.E. Wessels
Abstract
In this thesis I empirically test if acquisitions create or destroy shareholder value by examining the effect of acquisition announcements on stock returns of acquiring firms. I perform a so-called event study, on a sample of 748 announcements made between 2011 and 2016 by firms included in the S&P500 index, in which shareholder returns before the announcement of an acquisition are compared with the returns after the announcement. My results show that on average, the shareholders of the acquiring firms earn a return of 0.33% due to acquisition announcements at the day in which the announcement is made. On a yearly basis this represents a return of 83.63%, this implies that acquisitions announcements are creating value for shareholders of the acquiring firms.
Word Count: 7540
1. Introduction
Companies tend to believe that size matters and therefore devote a lot energy (and resources) to what is known as Mergers and Acquisitions (M&A). Even though there are many forms of executing M&A, the goal to merge with or acquire is to create synergies that makes the combined value of the businesses larger than the sum of the individual businesses. The larger a company becomes, the higher its market share, which supposedly will provide the company a competitive advantage due to economies of scale as larger companies usually get better prices from their suppliers as their order volume increases, which then leads to an increase in profits and thus to an increase in shareholder wealth.
Mergers and acquisitions are a prevalent strategy to grow companies by, for example, entering new markets or eliminating a competitor. The number of acquisitions over the last years has increased significantly, from eleven thousand in 1990 to forty-nine thousand in 2016 for a combined deal-value worth of 0.5 trillion USD and 3.6 trillion USD1. Due to the large economic impact of acquisitions, it seems worthwhile to investigate whether M&A activities have a significant impact on shareholder wealth. in particular since the so-called efficient market hypothesis (Fama, 1970) states that all available information is already incorporated in the stock price. If this is the case, then there should be no significant share price changes at the completion of an acquisition, since any revaluations of the concerned companies should already be incorporated into the share price at the acquisition announcement. Thus, if the efficient market hypothesis holds, there should be a steep reaction in share prices when acquisition announcements are made.
In this thesis, I investigate if acquisition announcements are a destruction of value for shareholders of acquiring firms. In order to answer this question empirical research on acquisitions announcements between 2011 and 2016 is conducted to gain knowledge from actual observations. By performing a so-called event study I am able to isolate the effect of acquisition announcements from normal market movements on stock returns. This isolation makes it possible to draw inferences on the effect of acquisition announcement on stock prices and thereby answering if acquisition announcements are a destruction of value for shareholders of acquiring firms.
2. Literature review and hypothesis development
Much research has been conducted on the subject of mergers and acquisitions, covering different aspects of the motivation for firms to engage in M&A activities (Selden and Colvin, 2003; Lynch and Lind, 2002; Hurtt, Kreuze and Langsam, 2000; Harari, 1997); the effect of M&A on firm performance (Cloodt, Hagedoorn and Van Kranenburg, 2006; Dickerson, Gibson and Tsakalotos, 1997; Healy Palepu and Ruback, 1992) and the effect of mergers and acquisitions on the value to shareholders of the companies involved in the transaction .
The effect of mergers and acquisitions to shareholders can be split into the effect on shareholders of the acquiring company and the effects on shareholders of the target company. Many researchers find that targeted companies tend to have significantly higher returns on the announcement day of an acquisition (Asquith, 1983; Chatterjee 1986; Weidenbaum and Vogt, 1987; Bradley, Desai and Kim, 1988; Kaplan and Weisbach, 1992; Jovanovic and Baguinsky, 2004; Liargovis and Repousis, 2011) . The significantly higher returns can be explained by the fact that a takeover premium is paid over the market value of the target company, which is beneficial for the shareholders of the target value (Walsh, 1989).
Firms perform mergers and acquisitions for various reasons. Creating a synergy by combining business activities to decrease costs and increase business performance is a common reason for companies to merge. Another reason is to increase a company’s market share by acquiring one of its competitors. A company can also acquire one of its suppliers to decrease the costs of supplies. It is clear that companies only perform M&A when they expect them to be beneficial. If, as Walter and Barney (1990) state, “M&A are a mechanism for managers to obtain and exploit economies of scale and scope” (p. 82), should an acquisition positively affect shareholders of the acquiring firm? Since acquisitions are mechanisms for managers to obtain and exploit economies of scale and scope it is difficult to elaborate on the contradicting results regarding the effects of acquisitions on shareholders.
In contrast to the aforementioned research, other studies (Dodd and Ruback, 1977; Asquith, 1983; Allen and Sirmans, 1987; Zhu and Malhotra, 2008; Rani, Yadav and Jain, 2014) observe positive returns for shareholders of acquiring firms. As stated before, mergers and acquisitions are mechanisms for managers to exploit economics of scale and scope. Possibilities to exploit economics of scale through acquiring companies will become clear to the market when companies announce their acquisition plans. Therefore, the revaluation of the acquiring company is positive at the announcement date.
A large body of research examines the effects of acquisition announcements on the wealth of the shareholders, with outcomes varying from positive returns to negative returns, and offering different explanations for why such returns occur. Despite extensive research, there is no consensus about the effect of acquisitions on shareholders of the acquiring firm, but it is clear that a market reaction follows from an acquisition announcement. To investigate the relation between acquisition announcements and returns to the shareholders of acquiring firms and to help build a consensus, the following hypotheses were tested.
H0: Announcements for acquisitions do not lead to significant effects on shareholders wealth of acquiring firms.
3. Methodology
The hypothesized relationship between the announcement of an acquisition and effects on the shareholders were tested using the event study methodology developed by MacKinlay (1997). An event study is a statistical method used to measure the impact of a financial event, such as acquisition announcements. Event studies analyse the difference between stock returns that would have been expected in the absence of an event and the returns that are observed when the event occurs. The difference between those returns is called abnormal returns (AR).
Taking the difference between so-called “normal” expected returns and observed returns isolates the effect of an event from normal market movements. Because of this isolation, we can test for significant effects of events in stock returns, which is a commonly used methodology in the field of finance (Asquith, 1983; Lubatkin, 1984; Chatterjee, 1986; Sirmans 1987; Mulherin and Boone, 2000; Fuller, Netter and Stemoller, 2002; Haleblian, Kim and Rajagopalan, 2006; Zhu and Malhotra, 2008; Liargovas and Repousis, 2011; Allen, Rani and Jain, 2014).
To calculate normal returns, we used the market model (MacKinlay, 1997), which is a statistical model that relates the stock returns of securities to the returns of the market using Equation 1:
𝑅𝑖𝑡 = 𝛼𝑖 + 𝛽𝑖𝑅𝑚𝑡+ 𝜀𝑖𝑡 (1)
where 𝑅𝑖𝑡 is the daily stock return for company i at time t calculated as
𝑅𝑖𝑡 = log ( 𝑃𝑖𝑡 𝑃𝑖𝑡−1)
(2)
𝑅𝑚𝑡 is the daily return of the market at time t, calculated the same way as the stock returns using Equation 3:
𝑅𝑚𝑡 = log ( 𝑃𝑚𝑡 𝑃𝑚𝑡−1)
(3)
where 𝑃𝑚𝑡 is the unadjusted closing price at time t. 𝛼𝑖 and 𝛽𝑖 are parameters representing the intercept and slope of the model with a zero mean disturbance term (𝜀𝑖𝑡) and a variance of 𝜎𝜀2𝑖.
Using the market model, the expected daily returns per company were calculated as follows
𝐸𝑅𝑖𝑡 = 𝑎𝑖 + 𝑏𝑖𝑅𝑚𝑡 (4)
In Equation 4, 𝑎𝑖 and 𝑏𝑖 are ordinary least squares (OLS) estimates in the model between observed market returns and stock returns. The OLS estimated values for 𝛼𝑖 and 𝛽𝑖 were used to calculate expected daily returns. Since expected returns in Equation 4 were used to draw inferences about the impact of acquisition announcements on stock returns, it did not make sense to use data affected by the announcement to estimate values for 𝛼𝑖 and 𝛽𝑖. A period before the announcement was used to calculate the OLS estimated values 𝑎𝑖 and 𝑏𝑖. This period is called the estimation window.
Figure 1. Timeline of the event study. The timeline investigated in this event study goes from T0 to T2,
where t=0 represents the acquisition announcement date.
Like McWilliams and Siegel (1997), the estimation window was selected to start 250 trading days before the event. Although arbitrarily chosen, the length of the estimation window should be sufficient large to estimate the values for 𝛼𝑖 and 𝛽𝑖 from Equation 1. Much shorter estimation windows could lead to higher variances in abnormal returns, 𝜎2(𝐴𝑅
𝑖𝑡), due to an additional variance component.
𝜎2(𝐴𝑅𝑖𝑡) = 𝜎𝜀𝑖 2 + 1 𝐸(1 + (𝑅𝑚𝑡− 𝜇̂𝑚)2 𝜎̂𝑚2 ) (5)
In Equation 5, the variance of the abnormal returns consists of two components. The first component, 𝜎𝜀2𝑖, is the disturbance variance as described in Equation 1. The second component, 1
𝐸(1 +
(𝑅𝑚𝑡− 𝜇̂𝑚)2
𝜎̂𝑚2 ), is an additional variance term due to sampling errors in the a and b estimates,
where 𝜇̂𝑚 is the mean market return, 𝜎̂𝑚2 the variance in the market returns and E the estimation period from T0 to T1 (MacKinlay, 1997). When the length of the estimation period increases, the second term in Equation 5 approaches zero. By selecting a 230-day estimation period, we could reasonably assume that the additional variance component in Equation 5 equals zero and could therefore ignore this term.
Day-250 T0
0 +20
T2
Estimation window Event window
Abnormal returns, 𝐴𝑅𝑖𝑡, for company i at time t were calculated as the difference between realized returns calculated in Equation 2 and the expected returns from Equation 4.
𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡− 𝐸𝑅𝑖𝑡 (6)
To infer that the announcement of acquisitions impact companies’ stock prices, abnormal returns were calculated as averages over companies for each time interval. The aggregation of companies was performed via calculating average abnormal returns, 𝐴𝐴𝑅𝑡, in Equation 7:
𝐴𝐴𝑅𝑡 = 1
𝑁∑ 𝐴𝑅𝑖𝑡 𝑁
𝑖=1
(7)
where N is the number of events (i.e., the number of acquisitions in the sample).
As a check on robustness to determine whether the size of the acquisition relative to the size of the acquiring company influences the results, the size weighted average abnormal returns, SWAARi, were calculated using Equation 8:
𝑆𝑊𝐴𝐴𝑅𝑡= ∑ (𝐴𝑅𝑖𝑡∗ (𝐷𝑉𝑖 𝑀𝑉𝑖)) 𝑁 𝑖=1 ∑ (𝑀𝑉𝐷𝑉𝑖 𝑖) 𝑁 𝑖=1 (8)
Average abnormal returns show possible shocks to the stock market at predefined day; however, it is reasonable to assume that that news of a financial event can take time to be reflected in the price, as stated by Vega (2006): “There is good reason to believe that stock markets are efficient because such markets are paradigmatic examples of competition. Nevertheless, rather than adjusting immediately to news surprises, stock prices tend to drift over time in the same direction as the initial surprise. This phenomenon is labelled earnings-momentum or post-earnings announcement drift (PEAD)” (p. 104).
The cumulative average abnormal returns, 𝐶𝐴𝐴𝑅(𝑡1, 𝑡2), was used to draw inferences about event windows from 𝑡1 𝑡𝑜 𝑡2. The 𝐶𝐴𝐴𝑅(𝑡1, 𝑡2) was calculated as the sum of average abnormal returns from 𝑡1 𝑡𝑜 𝑡2, as shown in Equation 9:
𝐶𝐴𝐴𝑅(𝑡1, 𝑡2) = ∑ 𝐴𝐴𝑅𝑡 𝑡2
𝑡=𝑡1
(9)
The reason for choosing the Corrado-rank test for 1-day event windows and the Cowan Generalized Sign Test for longer event windows is best stated by Cowan (1992), who investigated the predictive power of both tests in various event windows: “When both tests are correctly specified, the rank test generally provides more power than the sign test to detect an abnormal return that is present in very short event windows. However, the power of the rank test drops off rapidly as the number of days in the event window increases. The generalized sign test thus is better suited to the investigation of cumulative abnormal returns over event windows of several days.” (p. 356).
To use the Corrado rank test, the abnormal returns per share calculated in Equation 6 had to be transformed into ranks. The returns were ranked from 1, for the lowest AR per security, to 271 (the number of returns from T0 to T2), for the highest AR per security. After ranking the abnormal returns per event, the Corrado rank test statistic, Z1, was calculated for each day from T1 to T2 by dividing the average of the excess rank by the its standard deviation, as shown in Equation 10: 𝑍1 = 1 𝑁∑𝑁𝑖=1(𝐾𝑖,𝑡− 𝐾̅𝑖) S(K) (10)
where 𝐾𝑖,𝑡 represents the rank of the return for security i at time t, 𝐾̅𝑖 is the mean rank for company i, which is by construction equal to the amount of days from T0 to T2 divided by 2 plus one half, and S(K) is the standard deviation term, which is calculated as
As mentioned previously, the Corrado rank test was used for significance in 1-day event windows (i.e., the average abnormal return at time t where T1 ≤ t ≤ T2). The Cowan generalized sign test was used to test for significance of the impact for event windows from t1 to t2 where T1 ≤ t1 < t2 ≤ T2 (i.e., the cumulative average abnormal returns). The Cowan generalized sign test statistic, Z2, was calculated as
𝑍2 = 𝑊 − 𝑁p̂ [𝑁p̂(1 − p̂)]12
(12)
where W is the number of all positive cumulative abnormal returns from t1 to t2, N is the number of events and p̂ is the fraction of positive abnormal returns in the estimation period. p̂ was calculated using Equation 13: 𝑝̂ = 1 𝑁∑ 1 E 𝑁 𝑖=1 ∑ Φ𝑖 𝑡 𝑇1 𝑡=𝑇0 (13.1) Φ𝑖 𝑡 = {1 𝑖𝑓 𝐴𝑅𝑖𝑡 > 0 0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒 (13.2)
In the attempt to ensure that the outcomes of the Corrado rank and Cowan generalized sign test were reliable, an additional parametric t-test was performed as a robustness check. Dividing the (cumulative) average abnormal returns by its standard deviation gave the t-test statistic with n-1 degrees of freedom, as shown in Equations n-14 and n-15, where 𝑡1 in Equation 14 represents the test statistic for average abnormal returns and 𝑡2 in Equation 15 represents the test statistic for cumulative average abnormal returns.
𝑡1 = 𝐴𝐴𝑅𝑡 𝑣𝑎𝑟(𝐴𝐴𝑅𝑡) 1 2 ~ 𝑁(0,1) (14) 𝑡2 = 𝐶𝐴𝐴𝑅(𝑡1, 𝑡2) 𝑣𝑎𝑟(𝐶𝐴𝐴𝑅(𝑡1, 𝑡2)) 1 2 ~ 𝑁(0,1) (15)
The variance for the average abnormal returns at time t in Equation 14 was calculated as:
𝑣𝑎𝑟(𝐴𝐴𝑅𝑡) = 1 𝑁2∑ 𝜎𝜀𝑖 2 𝑁 𝑖=1 (16)
where 𝜎𝜀2𝑖 is the estimate for the zero mean disturbance component in the market model. The variance in cumulative average abnormal returns in Equation 15 was calculated by taking the sum of the variances of the average abnormal returns in the CAAR window:
𝑣𝑎𝑟(𝐶𝐴𝐴𝑅(𝑡1, 𝑡2)) = ∑ 𝑣𝑎𝑟(𝐴𝐴𝑅𝑡) 𝑡2
𝑡=𝑡1
3.2 DATA
To test the hypothesis that the announcement of an acquisition affects the stock price of the acquiring company I collected a sample of daily stock returns of all S&P 500 companies between January 1, 2011 and December 32, 2016. Data needed for the event study were collected from multiple different sources. The main source of data is the ZEPHYR database . ZEPHYR contains data about corporate actions such as mergers, takeovers and initial public offerings (IPO). We extracted all acquisition bids in which a constituent of the S&P500 index is the bidder. This yielded 4210 observations.
We wanted to test whether an announcement of an acquisition affects the stock price, but it is likely that not all announcements are taken seriously by the market, for example, when a company offers a lower value per share than the actual current trading value. To ensure that we only took serious acquisition announcements into account, only completed acquisitions were selected from the 4210 acquisition engagements, which left us with 2688 acquisition deals. Due to solely selecting completed acquisitions a sample selection bias occurs since part of the acquisition announcements are systematically excluded in the sample. The sample bias might affected the abnormal returns in the estimation window resulting in larger positive abnormal returns.
To use the market model to estimate abnormal returns, data on daily stock prices data are needed. Possible proxies to represent the market in the model were the S&P 500 index or a benchmark for the industry in which the acquiring company operates. Daily unadjusted closing prices of the S&P 500 index and all its constituents were extracted from Datastream. All acquiring companies were grouped per industry (i.e., sector) and daily unadjusted closing prices of the S&P 500 sector index were collected from spindices.com. Extracted closing prices were transformed into returns using Equations 2 and 3 for the estimation window (T0 to T1).
Descriptive statistics of the returns in the estimation window are presented in Table 1 which show that company returns tend to have the highest mean returns combined with the highest deviations from its mean. Sector returns and the S&P 500 index returns appear to have lower deviations from their mean, as shown by the variance and standard deviation.
TABLE 1
Descriptive statistics of the stock returns
In this table the descriptive statistics of the stock returns, sector returns and index returns of 748 estimation windows are presented.
returns average returns*
Company sector index Company sector index
Mean (%) 0.0508 0.0440 0.0407 0.0508 0.0440 0.0407 Median (%) 0.0000 0.0760 0.0941 0.0539 0.0453 0.0412 Standard deviation (%) 1.6501 1.0708 1.1818 0.0676 0.0459 0.0514 Sample variance (%) 2.7230 1.1467 1.3967 0.0046 0.0021 0.0026 Kurtosis 14.2467 4.1673 5.1128 -0.0670 0.4135 -0.2269 Skewness -0.1258 -0.3359 -0.4866 -0.0501 0.0681 -0.0548 Range (%) 78.2271 19.2316 15.2761 0.3880 0.2976 0.2992 Minimum (%) -40.1476 -10.5182 -8.9837 -0.1538 -0.0785 -0.1090 Maximum (%) 38.0795 8.7134 6.2924 0.2342 0.2191 0.1902 N 172040 172040 172040 230 230 230
*The descriptive statistics over average returns were calculated after averaging by 𝑅𝑡 = 1
𝑁∑ 𝑅𝑖𝑡 𝑁 𝑖=1 ,
Since sector and index returns are weighted averages from equities in the respective sector and index, those returns should not contain outliers. Company returns can contain outliers, which explains the respectively high standard deviation in Table 1. To control for outliers, a 5% winsorization technique was applied per company from T0 to T1, as shown in Table 2. Using 5% winsorization, all returns above the 97.5th percentile were replaced with the value of the 97.5th percentile and all returns below the 2.5th percentile were replaced with the value of the 2.5th percentile.
TABLE 2
Descriptive statistics of returns after winsorization.
In this table the descriptive statistics of the stock returns of 748 estimation windows are presented after applying 5% winsorization. Company returns Average company returns* Mean (%) 0.0524 0.0524 Median (%) 0.0000 0.0541 Standard deviation (%) 1.4365 0.0584 Sample variance (%) 2.0636 0.0034 Kurtosis 1.8233 -0.1875 Skewness -0.0292 -0.1075 Range (%) 16.9241 0.3252 Minimum (%) -8.1218 -0.1261 Maximum (%) 8.8024 0.1991 N 172040 230
*The descriptive statistics over average returns company were calculated after averaging by 𝑅𝑡= 1
𝑁∑ 𝑅𝑖𝑡 𝑁
Due to winsorizing the data, a bias was introduced to the results. However, not controlling for outliers and leaving the returns as is or deleting outliers would both create a larger bias in the results. Deleting outliers could change outcomes negatively, since possible important information is removed from the data. Leaving outliers as is can create negative effects since extreme high or low values could have an unwanted large impact in calculation, for example, of average abnormal returns in Equation 6.
Comparing Tables 1 and 2, we observed that winsorizing the company returns in the estimation window drastically narrowed the distributing range, from 78% (from -40% to +38%) to 16.9% (from -8.1% to +8%). This is more in line with described statistics of the sector and index, which have distribution ranges from 15.27% to 19.2%. Furthermore, we observed that the kurtosis of the company returns was reduced from 14.2 to 1.8, which leads to the conclusion that Winsorizing company returns effectively dealt with outliers in the data.
TABLE 3
Dispersion of acquisitions in time and sector.
4. Results
In this section, the results are presented and discussed, starting with the selection of the market. The market model as described in Section 3 in Equation 1 relates the returns of stock to the market portfolio. Since the event study is based on abnormal returns which were calculated using the market model, selecting the market portfolio is essential for reliable results
The S&P 500 index, CRSP Value Weighted Index, and the CRSP Equal Weighted Index are frequently used as market portfolios (Mackinlay, 1997). Since all the acquisitions in this research were performed by S&P500 companies, the S&P 500 index can be used as the market portfolio. However, the S&P 500 index is subdivided into 11 sectors that are usable as market portfolios on their own. Table 2 shows a large dispersion of acquisition announcements around industry sectors. Due to this dispersion, it could is a better fit to use sector returns instead of index returns in the market model.
Both the S&P 500 index as a whole and the 11 sectors were used for the ordinary least squares regressions in Equation 1 to determine which is the best fit to represent the market. The R2 of regressions were used to test the goodness of fit of the model, since the R2 indicates how much of the variation of the response data is explained by the model. To determine the best fit, two methods were used to calculate the R2 of the model. The first is by taking average of all R2 from the regressions in Equation 4 as:
where 𝑅𝑚2 represents the R2 of the model, and 𝑅
𝑖2 represents the R2 of the regression in Equation 4 for company i. The 𝑅𝑚2 was calculated using sector returns and index returns as representatives for the market.
The second way to calculate the R2 of the model is by performing a regression between the average company returns per day in the estimation window, 𝑅̅𝑡, and the average market returns per day 𝑅̅𝑚𝑡:
𝑅̅𝑡 = 𝛼𝑖 + 𝛽𝑖𝑅̅𝑚𝑡 (19)
where the average company returns are calculated as 𝑅̅𝑡 = 1
𝑁∑ 𝑅𝑖𝑡 𝑁
𝑖=1 and average market returns
are calculated as 𝑅̅𝑚𝑡= 1
𝑁∑ 𝑅𝑚𝑡 𝑁
𝑖=1 . Again, the market in Equation 19 is represented by either the sectors or the index. In both methods, the R2 of the model using sectors representing the market is a better fit due to the higher R2, as can be seen in Table 4.
TABLE 4
Goodness of fit for the market.
In this table the results of the goodness of fit test is presented showing the R2 of the market model where the
market is represented by either the sectors or the index. Method one uses averages of 748 independent regressions and method two uses the regression of the average returns.
Concluding from Table 4, the sector returns are the most appropriate to use as proxy for the market returns. However, the R2 only explains how much of the variation is explained by the model. It does not say if the model is correct. For the model to be correct, the abnormal returns from T0 to T1 should have a mean of zero. Figures 2 and 3 graphically show the average abnormal returns in the estimation window using different market portfolios. The figures show that in both cases average abnormal returns fluctuate around zero, however returns based on the S&P 500 index yield slightly larger outliers compared to abnormal returns based on sector returns.
Figure 2: average abnormal returns based on index returns. This graph shows the average abnormal
returns, where the market is represented by the S&P 500 index returns, of all firms that announce an acquisition from 250 days to 20 days before the announcement of an acquisition.
Figure 3: average abnormal returns based on sector returns. This graph shows the average abnormal
returns, where the market is represented by different sector returns, of all firms that announce an acquisition from 250 days to 20 days before the announcement of an acquisition.
Descriptive statistics of the abnormal returns in the estimation window are shown in Table 5, which confirms that both, abnormal returns based on sector and index, yield a mean of zero. Using sector returns as market portfolio however appears to have smaller variances compared to the index.
TABLE 5
Descriptive statistics average abnormal returns.
This table shows the descriptive statistics for the average abnormal returns in the estimation window where the market is based on either, the sector or the index.
abnormal returns based on
Sector Index Mean (%) 0.0000 0.0000 Median (%) -0.0007 -0.0037 Standard deviation (%) 0.0366 0.0488 Sample variance (%) 0.0013 0.0024 Kurtosis -0.2768 0.1789 Skewness 0.0770 0.3265 Range (%) 0.1850 0.2928 Minimum (%) -0.1009 -0.1409 Maximum (%) 0.0841 0.1520 N 230 230
Combining the higher R2 and the smaller discrepancies around the zero mean, the market model using sector based market returns is the best fit. Therefore, the sector based abnormal returns was used for the following tests to reach inferences about the relation between acquisition announcements on stock returns.
Figure 4: average abnormal returns in the event window. This graph shows the average abnormal returns
and the size weighted average abnormal returns from 20 days before to 20 days after the announcement of an acquisition.
Figure 4 shows that only significant average abnormal returns are found at the event date itself, since before and after the event day the AAR tend to stay around 0%. Controlling for the size of the acquisition relative to the size of a company, we see that abnormal returns tend to follow the same path as the unweighted calculated average abnormal returns during the event window, except for the announcement day and the day after the announcement day.
At the announcement day, we observe a higher abnormal return in SWAAR (0.60%) than in AAR (0.33%), which could mean that when companies announce large acquisitions, relative to the company size, the market generally interprets this as a positive sign. At the first day after the announcement, however, the relative large announcements appear to have large negative revaluation (-0.42%).
TABLE 6
Average abnormal returns.
In this table an overview is given of the average abnormal returns per day in the event window together with tests statistics of the Corrado rank and t-test.
Test statistics
Day Average abnormal return (%) Corrado Z test statistic t-test statistic
-20 -0.072(*) 1.246 1.939 -19 0.0245*** 4.282 0.659 -18 -0.0779***(**) 3.048 2.100 -17 -0.0848*(**) 1.784 2.284 -16 -0.0129 0.588 0.347 -15 -0.0243 0.244 0.653 -14 -0.0073 1.545 0.198 -13 -0.0636***(*) 8.331 1.714 -12 -0.0023 0.057 0.063 -11 -0.0316 0.035 0.852 -10 0.0578*** 5.586 1.556 -9 -0.0476*** 9.857 1.284 -8 0.0181** 2.427 0.487 -7 -0.0772***(**) 3.916 2.079 -6 -0.0611***(*) 7.055 1.646 -5 -0.0011* 1.861 0.029 -4 0.0121 0.190 0.327 -3 0.0004 0.296 0.011 -2 -0.0259 0.964 0.697 -1 -0.0401* 1.665 1.079 0 0.3319***(***) 38.719 8.942 1 0.0677***(*) 9.943 1.825 2 -0.0744***(**) 7.742 2.004 3 0.0282*** 3.470 0.760 4 -0.0265 0.200 0.715 5 0.0613(*) 0.177 1.652 6 0.0281 0.057 0.756 7 -0.0261 1.142 0.703 8 -0.0727*(*) 1.793 1.960 9 -0.0409* 1.938 1.103 10 -0.0334 0.122 0.899 11 0.0458*** 5.354 1.233 12 -0.092***(**) 9.870 2.479 13 -0.0089 0.491 0.240 14 0.0256 0.816 0.691 15 0.0233 0.047 0.628 16 -0.0227*** 3.116 0.611 17 -0.0762**(**) 1.987 2.054 18 0.0376 0.076 1.012 19 0.0018 0.192 0.049 20 0.0377 0.197 1.017
Note: Asterisks indicate the significance of the results by the Corrado rank test. Asterisks in
Although multiple days in the event window are statistically different from zero at a 1% significance level, they do not appear to be economically significantly different from zero. Daily, statistically significant abnormal returns ranging from -0.092% to 0.067% are hardly labelled as economically significant since such changes are hardly noticeable in one’s portfolio. The average abnormal return of the event day is 0.331%, which could be labelled as economically significant, since shareholders will most likely notice such daily return in their portfolios.
Using the t-test as in Equation 14 as a robustness check gave different results. The robustness check found that only the announcement date itself is statistically significant at a 1% significance level. A few other days are statistically significant at 5% and 10% levels. As a final check, some statistically significant results might be significant by change. In other words, so-called Type 1 errors might have occurred, meaning that we incorrectly rejected the true null hypothesis of being not significantly different from zero.
A 5% significant Bonferroni correction was applied, meaning that the overall probability of incorrectly inferring statistical significance due to a Type 1 error is reduced to 5% (Bonferroni, 1936). Since we conducted 82 tests, the Bonferroni corrected critical p-value becomes 5% / 82 = 6.097 * 10-4 %, corresponding with a critical Z- and t-score of 3.245. With a 5% Bonferroni-correction, therefore, only the announcement date itself is statistically significantly different from zero according to the Corrado rank test and t-test.
As previously stated, it is reasonable to assume that it takes time for the market to react to new information. For that reason, cumulative average abnormal returns were calculated for various windows to test whether there is a significant impact on the returns over a period of days. Multiple periods were tested using the Cowan generalized sign test. Figure 6 shows the cumulative average abnormal return from the 20 days before to the 20 days after an acquisition announcement.
Figure 6: cumulative average abnormal returns in the event window. This graph shows the Cumulative
average abnormal returns in the event window from 20 days before to 20 days after the announcement of an acquisition.
TABLE 7
Cumulative average abnormal returns.
In this table an overview is given of the cumulative average abnormal returns for selected event windows together with test statistics of the Cowan generalised sign test (Z-score) t-test (t-score).
window Cumulative Average
abnormal return (%) Z-soore t-scor t1 t2 -20 -16 -0.2231*(***) 1.9228 2.6883 -15 -11 -0.1291* 1.7766 1.5561 -10 -6 -0.1101 0.3138 1.3268 -5 -2 -0.0144 0.0944 0.1941 -1 1 0.3595***(***) 3.928 5.5931 2 5 -0.0114 0.6795 0.154 6 10 -0.1451(*) 0.3138 1.7481 11 15 -0.0062 0.1676 0.0749 16 20 -0.0218 0.2407 0.2626
Note: Asterisks indicate the significance of the results by the Corrado rank test. Asterisks in parentheses indicate the significance level according to the t-test. Significant results at a 10% level are denoted by *, 5% by ** and 1% by ***.
5. Summary and conclusions
The aim of this master thesis was to find out how the market reacts to acquisition announcements. In the literature, a conclusive answer is given regarding the market reaction for target firms. Shareholders wealth of target firms generally increases when acquisitions are announced, since acquiring companies pay a takeover premium, which results in a higher share price for the target company. No conclusive answer is presented in literature on the wealth of shareholders of the acquiring firm. Therefore, we tried to complement the literature by empirically testing the effects on shareholders of acquiring firms at an announcement of an acquisition by performing an event study.
The event study on acquisition announcement was performed on a data set of acquisitions by companies in the S&P 500 with a deal value larger than 10 million euro between January 1, 2011 and December 32, 2016. This provided a sample of 748 acquisition announcements.
Furthermore, we found evidence that returns in a window of 1 day before to 1 day after the announcement are highly significant, with a cumulative return of 0.36%. However, this is most likely due to the abnormal returns found at the announcement date itself. From the performed event study, we can conclude that announcements for acquisitions lead to significant positive effects on share prices for shareholders of acquiring firms and therefore we reject the null hypothesis and infer that announcements for acquisitions lead so significant effects on shareholders wealth of acquiring firms meaning that that acquisitions announcements are no destruction of value.
REFERENCES
Allen, P. R., & Sirmans, C. F. (1987). An analysis of gains to acquiring firm's shareholders: The special case of REITs. Journal of Financial Economics, 18(1), 175-184.
Asquith, P. (1983). Merger bids, uncertainty, and stockholder returns. Journal of Financial
Economics, 11(1-4), 51-83.
Aw, M. S. B., & Chatterjee, R. A. (2004). The performance of UK firms acquiring large cross-border and domestic takeover targets. Applied Financial Economics, 14(5), 337-349.
Bonferroni, C. E. (1936). Statistical theory of classes and calculation of probabilities. Pub R Ist Superiore Sci
Econ Commerc Firenze, 8, 36-62.
Bradley, M., Desai, A., & Kim, E. H. (1988). Synergistic gains from corporate acquisitions and their division between the stockholders of target and acquiring firms. Journal of Financial Economics, 21(1), 3-40.
Chatterjee, S. (1986). Types of synergy and economic value: The impact of acquisitions on merging and rival firms. Strategic management journal, 7(2), 119-139.
Cloodt, M., Hagedoorn, J., & Van Kranenburg, H. (2006). Mergers and acquisitions: Their effect on the innovative performance of companies in high-tech industries. Research policy, 35(5), 642-654. Corrado, C. J. (1989). A nonparametric test for abnormal security-price performance in event
studies. Journal of financial economics, 23(2), 385-395.
Cowan, A. R. (1992). Nonparametric event study tests. Review of Quantitative Finance and
Datta, D. K., & Puia, G. (1995). Cross-border acquisitions: An examination of the influence of relatedness and cultural fit on shareholder value creation in US acquiring firms. MIR: Management
International Review, 337-359.
DeLong, G. L. (2001). Stockholder gains from focusing versus diversifying bank mergers. Journal of Financial
Economics, 59(2), 221-252.
Dickerson, A. P., Gibson, H. D., and Tsakalotos, E. (1997). The impact of acquisitions on company performance: Evidence from a large panel of UK firms. Oxford Economic Papers, 49(3), 344-361.
Dodd, P. (1980). Merger proposals, management discretion and stockholder wealth. Journal of Financial
Economics, 8(2), 105-137.
Dodd, P., & Ruback, R. (1977). Tender offers and stockholder returns: An empirical analysis. Journal of
Financial Economics, 5(3), 351-373.
Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. The Journal of
Finance, 25(2), 383-417.
Fuller, K., Netter, J., & Stegemoller, M. (2002). What do returns to acquiring firms tell us? Evidence from firms that make many acquisitions. The Journal of Finance, 57(4), 1763-1793.
Haleblian, J. J., Kim, J. Y. J., & Rajagopalan, N. (2006). The influence of acquisition experience and performance on acquisition behavior: Evidence from the US commercial banking industry. Academy of Management Journal, 49(2), 357-370.
Harari, O. (1997). Curing the M&A madness. Management Review, 86(7), 53.
Hurtt, D. N., Kreuze, J. G., & Langsam, S. A. (2000). Can this merger be saved? Journal of Corporate
Accounting & Finance, 11(2), 17-24.
Jensen, M. C. (2001). Value maximization, stakeholder theory, and the corporate objective function. Journal of applied corporate finance, 14(3), 8-21.
Jovanovic B., & Braguinsky S. (2004). Bidder Discounts and Target Premia in Takeovers. American Economic
Review, 94, 46-56.
Kaplan, S. N., & Weisbach, M. S. (1992). The success of acquisitions: Evidence from divestitures. The Journal
of Finance, 47(1), 107-138.
Liargovas, P., & Repousis, S. (2011). The impact of mergers and acquisitions on the performance of the Greek banking sector: An event study approach. International Journal of Economics and
Finance, 3(2), 89.
Lubatkin, M. (1987). Merger strategies and stockholder value. Strategic management journal, 8(1), 39-53. Lynch, J. G., & Lind, B. (2002). Escaping merger and acquisition madness. Strategy & Leadership, 30(2),
5-12.
MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of economic literature, 35(1), 13-39.
McWilliams, A., & Siegel, D. (1997). Event studies in management research: Theoretical and empirical issues. Academy of management journal, 40(3), 626-657.
Mulherin, J. H., & Boone, A. L. (2000). Comparing acquisitions and divestitures. Journal of corporate
Rani, N., Yadav, S. S., & Jain, P. K. (2014). Impact of domestic and cross-border acquisitions on acquirer shareholders’ wealth: empirical evidence from Indian corporate. International Journal of Business
and Management, 9(3), 88.
Selden, L., & Colvin, G. (2003). M&A needn't be a loser's game. Harvard Business Review, 81(6), 70-9.
Vega, C. (2006). Stock price reaction to public and private information. Journal of Financial
Economics, 82(1), 103-133.
Walsh, J. P. (1989). Doing a deal: Merger and acquisition negotiations and their impact upon target company top management turnover. Strategic Management Journal, 10(4), 307-322.
Walter, G. A., & Barney, J. (1990). Management Objectives in Mergers and Acquisitions. Strategic
Management Journal, 11(1), 79-86
Weidenbaum, M., & Vogt, S. (1987). Takeovers and stockholders: Winners and losers. California
Management Review, 29(4), 157-168.
