Acquisition timing and long-run stock performance
University of Groningen Faculty of Economics and Business
Department of Finance
MSc Business Administration, Finance April, 2012
Luuk Bakker
Luuk_bakker@hotmail.com Student number: 1551779
2
Acquisition timing and long-run stock performance
Abstract
This thesis investigates whether timing of acquisitions has an impact on long-run stock performance. I look at three different types of timing: market timing, firm-specific timing and timing in an acquisition wave. Consistent with the findings of Rhodes-Kropf and Viswanathan (2004), Bouwman, Fuller and Nain (2009) and Croci, Petmezas and Vagenas-Nados (2010), I find that acquisitions announced at moments of high market value underperform compared to those announced at low market value. I find that timing a share's financed acquisition when firm stock value is high leads to less long-run return. This confirms the
findings of Rhodes-Kropf, Robinson and Viswanathan (2005). They find that firms exploit temporary stock overvaluation to finance acquisitions. In the long-run, the stock price is corrected by the market, leading to a lower return. I also find that acquisitions made at the end of an acquisition wave outperform
those made at the beginning of an acquisition wave. These findings contradict the managerial herding theory of Bouwman, Fuller and Nain (2009).
JEL: G34
3
1. Introduction
Most of the recent studies on mergers and acquisitions (M&A) find that mergers and acquisitions do not provide positive abnormal returns, but negative abnormal return for the acquirers. A review of acquisition performance studies by Tuch and O’Sullivan (2007) shows that the overwhelming majority of long-run acquirer performance is negative.
Various studies have been performed to find out what causes this underperformance and what determinants affect the success of M&A. In their review study, Agrawal and Jaffe (2000) try to explain the commonly found long-run negative performance of acquisitions. They find that managers tend to use equity to finance the acquisition when their equity is overvalued and issue debt or finance out of retained earnings when shares are undervalued.
The differences between acquisitions of private firms and public firms have been researched by Capron and Shen (2007). They find evidence that points to the role of private information on acquirer performance. The availability of information differs between private and public firms. More information is available on public firms because they generally have higher regulatory disclosure requirements, greater coverage by analysts and better ties to investment banks (Capron and Shen, 2007).
Moeller, Slingemann and Stulz (2004) find that the announcement return for large acquirers is two percent lower compared to small acquirers, irrespective of the form of financing or whether the firm is public of private. They find evidence that large firms make acquisitions with negative dollar synergy gains and offer a larger acquisition premium than small firms.
Megginson, Morgan and Nail (2004) find that mergers that decrease corporate focus result in significant shareholder losses in the three-year period after mergers. They define focus as the opposite of diversification and they use the Herfindahl index to measure the level of corporate diversification. They also find that mergers that increase or preserve focus result in higher long-run performance.
4
Does the timing of acquisitions influence their success?
The studies of Soegiharto (2008) and Croci, Petmezas and Vagenas-Nanos (2010) provide a theory of what this effect might be. They find that more M&As are announced when the market is overvalued. The fact that most acquisitions occur procyclically in waves implies that more acquisitions are announced when equity values are higher and financing with equity is cheaper. Managers might be inclined to take advantage of the ease with which they can get equity, thereby making acquisitions which are less profitable than when equity financing is more expensive. This might ultimately lead to a price correction if investors think the potential values of these acquisitions are overestimated.
In this thesis I will measure success by the long-run stock return. Stock return is an appropriate measure of success because it reflects the changes in the wealth of shareholders, who are the ultimate owners of a firm’s control rights. Therefore, they must be the centre of any discussion concerning firm performance (Jensen, 1984). Several earlier studies on firm performance following mergers and acquisitions do not focus long-run returns, but on the announcement returns or short-run returns. Their belief in market efficiency leads many writers to presume that the market immediately makes the full adjustments to account for the new information on the firm. The possibility exists, however, that the market does not always accurately predict the future performance of mergers and acquisitions (Conn, Cosh, Guest and Hudges, 2005). Many studies found post-merger long-run negative results that cannot be ignored, and which can be assumed to contradict the theory of efficient markets (Agrawal and Jaffe, 2000). Further, Dutta and Jog (2009) have found in their Canadian acquisition investigation that the stock market corrects for an initial overreaction to an acquisition announcement.
When explaining different long-run returns that are due to differences in timing, the literature on initial public offerings is of interest. The market timing theory, which is extensively used in the literature on initial public offerings, states that firms prefer external equity when the cost of equity is low and that firms prefer debt otherwise (Huang and Ritter, 2004). According to this market timing theory of capital structure, or windows of opportunity theory, corporate management can perceive their risky securities as misvalued by the market. When the relative cost of equity (debt) is low and the firm has financing needs, they will issue equity (debt). This theory is also applicable to acquisitions. When the cost of external equity is low, executives are inclined to make use of this to finance acquisitions as well.
5 relatively cheap equity that is available. To determine if this is the case, they studied whether managers time an acquisition around a peak of the valuation of their company, but they find no evidence of this.
Bouwman et al. (2009) have also investigated another type of timing, the timing in a merger and acquisition wave. They find evidence for the managerial herding theory, which suggests that firms that acquire later in a merger wave are likely to perform poorly relative to firms that acquire early in a wave.
In this thesis, timing will be investigated in three different ways. First, I will look at the timing in relation to the economic cycle, which will be referred to from here on as market timing. Next, I will investigate whether or not managers time acquisitions to correspond with the time when their stock value is at or near its peak value; this will be referred to as firm-specific timing. Finally, I will test the theory of managerial herding, for which Bouwman et al. (2009) find evidence.
This research looks at acquisitions performed by firms in the European Union from 1999 to 2011. Existing theories on the relationship between acquisition timing and long-run acquirer returns are examined and tested. Using the Calendar Time Abnormal Return (CTAR) method for calculating long-run abnormal returns, I test for the three relationships between timing and performance mentioned above.
This thesis distinguishes itself from previous studies by focusing specifically on the effects of timing on acquisition performance. I confirm the finding that high-valuation market acquisitions underperform compared to low-valuation market acquisitions. I also find that firms that use their high stock value to finance their acquisition underperform in the long run. Finally, I find contradictory evidence on the managerial herding theory, advocated by Bouwman et al. (2009). This takes the idea of managerial herding leading to underperformance into question. The outcomes of this thesis can be used by investors as it gives indications for the long-run return effects of acquisitions based on variations in timing factors.
6
2. Theoretical review
This section elaborates on the theoretical background of the relation between the acquisition timing and its success. First, I will discuss previous findings on the causes of M&A activity waves and the relation between these waves and the success of acquisitions. Then I move to the theory that applies to the relation between timing and long-run performance. Finally, based on previous findings and theories in M&A literature, I present my hypotheses.
2.1 M&A waves
In any research on mergers and/or acquisitions, it is important to mention merger waves, for the gross of all acquisitions occur in waves (Rhodes-Kropf and Viswanathan, 2004). In his review study on studies of M&A performance, Soegiharto (2008) separates the literature on merger waves in two parts: The earlier studies that advocate the neoclassical view and the more recent studies that support the behavioural view of merger waves.
The neoclassical view, which was first documented by Gort (1969), observes that interindustry variation in takeover activity is in line with an economic disturbance model. In line with these observations, Jensen (1993) found that technological and supply shocks cause merger activity. Mitchell and Mulherin (1996) test whether shocks in industries cause more takeover activity. They used sales data as a proxy for industry shocks. Then they showed that shocks, combined with the cause of the shock, such as deregulation, financing innovation and foreign competition, caused takeover activity in the 1980s. The standard neo-classical theory is, however, unable to explain why waves are significantly more concentrated than economic shocks encourage them to be (Soegiharto, 2008). In 2005, Harford posits that shocks only lead to a merger when the economic conditions generate enough capital liquidity to accommodate the necessary restructurings. Neoclassical theory has been able to describe most of the characteristics of merger waves, but it is unable to explain whether cash or stock is the medium used for payment (Soegiharto, 2008).
2.2 M&A and overvaluation
7 Misvaluation and its relation to mergers and acquisitions has also been investigated by Rhodes-Kropf and Viswanathan (2004). They argue that stock bids are more likely to be accepted by a target in high-valuation markets. In their model, management of acquiring firms have private information about the value of their firms and the potential value of merging with the target. Target management only has private information about their own stand-alone value. The valuations of companies vary over time and move above and below the true value of the companies. When companies’ valuations do not reflect the true value, there is misvaluation. These misvaluations contain a firm-specific component and a market-wide component. When the target firm is misvalued, the targets’ management knows this, but they cannot determine what part of this misvaluation is caused by a market-wide effect and what part is caused by a firm-specific effect.
The bidders’ stock bids are positively related to the market-wide misvaluation component. A target attempts to filter out this market-wide effect before deciding whether to accept or reject the bid. Again, the target sees that his own firm is overvalued, but does not know how much of this is caused by a market-wide component and how much is due to the firm-specific component. The correction of the bidder firm value is done correctly on average, but the more the market is overvalued, the larger the target’s expectation of his firm-specific misvaluation is, because he cannot tell which of the two components is causing his firm’s misvaluation. As the target filters too little market-wide effect out of the bid when the market is overvalued, it overvalues the bidder’s stock and overvalues the bidder’s stock bid. Thus, stock offers look more valuable to targets in high-valuation markets due to market overvaluation, leading to a higher likelihood of acceptations of bids (Rhodes-Kropf and Viswanathan, 2004).
Because these acquisitions are based on misvaluation, a stock price correction for these firms can be expected. This leads to a lower long-run abnormal return of acquisitions announced in an overvalued stock market compared to acquisitions announced in an undervalued market. This theory is tested in hypothesis 1a. To fully examine the implications of the theory of Rhodes-Kropf and Viswanathan (2004) on acquisition performance, share-financed acquisitions are compared to cash-financed acquisitions. If targets are more likely to accept a shares offer from the bidder in a high market state and if that results in a stock price correction in the long run, then, in a high market state, acquisitions that are financed with shares should underperform compared to acquisitions financed with cash. This is because target firms are not more inclined to accept a cash offer based on this overvaluation. This leads to hypothesis 1b.
8
H1b: Acquisitions financed with shares in high-valuation markets underperform more than acquisitions financed with cash.
In a follow up article by Rhodes-Kropf, Robinson and Viswanathan (2005), a model is developed to test the theory that misvaluation affects merger activity. They separate the market-to-book ratio into three components and investigate each component separately. Using these components, their evidence supports the misvaluation theory of Rhodes-Kropf and Viswanathan (2004) and the theory of Shleifer and Vishny (2003). Rhodes-Kropf et al. (2005) find that acquirers are priced higher than targets and 60% of this difference is attributable to firm-specific error. They also find that firms with a temporary overvaluation compared to their sector acquire more.
The correlation between high-valuation markets and increased M&A activity is also studied by Bouwman et al. (2009). They investigate if acquisitions undertaken in booming markets provide less return than those undertaken in depressed markets and they look for the cause of this. Their expectations are confirmed about the long-run underperformance of high-market acquisitions compared to low-market acquisitions. One of the causes they investigate is if acquirers announce their stock acquisitions when the stock value of the acquirer is close to an annual high. The stock-price correction that will follow in the long run might explain the poor performance of acquirers that time the market in this manner. However, Bouwman et al. (2009) find no evidence of this type of firm-specific timing leading to underperformance. Two hypotheses are constructed to test these relationships. Hypothesis 2a tests if, in general, acquisitions with firm-specific timing lead to more underperformance than those without firm-specific timing. Because this theory focusses on whether or not share-financed acquisitions are timed, the acquisitions financed with shares must be looked into. This is done with hypothesis 2b. It is expected that acquisitions financed with shares that are announced at a moment of high acquirer stock value perform worse in the long run than share-financed acquisitions that have not been announced at a moment of high acquirer stock value.
H2a: Acquisitions made when acquirer stock value is high perform worse than acquisitions made when acquirer stock value is not high.
H2b: Acquisitions financed with shares perform worse when they are timed at a point when acquirer stock value is high.
9 the only way firms can find out what the quality of innovations is, is by looking at the experience of the early adopters. Early adopters are the first to try an innovation, which in this case is an acquisition. The waves grow when the experiences of early adopters are positive, leading to even more adoptions. Thus, managers demonstrate herding behaviour by copying the early adopters. The wave ends when information on recent adopters’ experience is negative, as this causes other firms to refrain from making the same adoption. The wave stops when the number of adopters drops to zero; thus, at the end of a wave the least profitable adoptions are completed. Bouwman et al. (2009) use this theory as their main explanation for the ending of merger waves. They find that mergers that are undertaken at the end of a wave are less profitable than those initiated at the beginning of a wave. This finding will be tested in hypothesis three.
H3: Acquisitions made at the end of a merger wave perform worse than acquisitions made in the beginning of a wave.
While the first two hypotheses consist of two parts, the third consists of only one part. The reason for this is that the method of payment is irrelevant in the herding theory. This thesis will focus on testing the five hypotheses stated above.
3.
Data
The acquisition firm data for this thesis is extracted from the Zephyr database published by Bureau van Dijk (BvD) and has the following requirements:
1. The acquisitions are announced during the period of January 1st, 1997 to January 1st, 2011. 2. The acquirer firm must be based in a country that is a member of the European Union from the
start of the sample period.
3. The acquirer firm must be listed on a major European stock exchange. 4. The minimum value of the acquisition is EUR 40 million.
5. The acquirer stake in the target increases by at least 50 percent. 6. The acquisition must be completed.
7. To take in all effects of an acquisition on acquirer return, the month in which the first rumours of the acquisition occur equals the month that the acquisition is announced.
10 The first six requirements are fulfilled by selecting these properties in the Zephyr database menu. The seventh is fulfilled by checking whether the rumour month is equal to the announcement month. All acquisitions not fulfilling this requirement are removed from the dataset.
The acquirer firm’s monthly total returns, the S&P 350 index total returns, all listed European security returns and the monthly Euribor German bond rate are extracted from the Thomson Reuters Datastream database. The total return is used so the return data are unaffected by dividend payments. Monthly returns are extracted because this thesis analyses monthly abnormal returns.
The dataset consists of 982 acquisitions from European Union-based acquirers from 1997 to 2011. The median deal value is 146 million euros. Of these acquisitions, 167 are financed with shares, 623 with cash and 192 with a cash/share combination or with another financing method. Table 1 shows the number of acquisitions per country. The overwhelming percentage of the acquisitions come from the UK. Table 2 shows the number of acquisitions per year.
Table 1. Acquisitions per country
Country Austria Belgium Denmark Finland France
# of acquisitions 11 22 15 44 126
Country Germany Greece Ireland Italy Luxembourg
# of acquisitions 80 10 27 57 8
Country Netherlands Portugal Spain Sweden UK
# of acquisitions 54 8 67 55 430
Table 2. Acquisitions per year
Year 1997 1999 2001 2003 2005 2007 2009
# of acquisitions 25 62 72 47 88 109 30
Year 1998 2000 2002 2004 2006 2008 2010
11
4.
Methodology
In this section I will elaborate on the methodology of this thesis. First, two commonly used methods for measuring long-run performance and their up- and downsides will be explained, followed by my reasoning for choosing one of these two. Finally, the method for classifying the economic state will be explained.
4.1 Measures of long-run performance
There has been much debate on which method is best for measuring long-run abnormal stock performance. The two most advanced methods for calculating long-term abnormal performance are the Buy and Hold Abnormal Return (BHAR) approach and the Calendar-Time Abnormal Return (CTAR) approach. The BHAR method measures abnormal performance by comparing the return in a virtual investment in a portfolio of firms with acquisitions to the returns from an investment of a reference portfolio of firms with the same characteristics but without acquisitions. The difference is measured over a certain period, which results in the abnormal return (Mitchell and Stafford, 2000). The CTAR approach measures the long-term performance by computing monthly portfolio returns of firms that were involved in an acquisition within previous T months for the entire sample period. These returns are then corrected with the Fama and French three factor model to calculate abnormal returns. While the BHAR method is the more popular of the two methods, it has been criticized by Ritter (1991) and Fama (1998) for having several flaws. Moreover, Mitchell and Stafford (2000) argue that BHAR assumes the independence of individual events, while these events actually do show dependence. Therefore, Mitchell and Stafford (2000) and Fama (1998) consider the CTAR method to be superior to the BHAR method in calculating long-run abnormal returns. In this thesis, I follow the recommendations of Fama (1998) and Mitchell and Stafford (2000) and use the CTAR method. Fama (1998) also recommends a monthly approach, because monthly expected returns are more predictable than those of longer periods. Therefore, I will focus on monthly returns.
4.2 CTAR in detail
12 When a firm makes a second acquisition within three years of the first, the time the firm stays in the portfolio is extended to three years after the last acquisition. Multiple acquisitions by a firm will only be included in a portfolio if these consecutive acquisitions have the same payment method and timing classification. Therefore, if a firm acquires a second time with a different payment method or with a different timing classification, it is not used for the portfolio construction.
Following Mitchell and Stafford (2000), Bouwman et al. (2009) and Lougran and Ritter (2000), I use the three factor model of Fama and French (1993) to regress the portfolio excess returns. In their study they identify 3 common risk factors for stock market returns. The portfolio return is controlled for size, book-to-market ratios and excess market returns. The formula of the Fama and French factors is shown in equation 1:
(1) ( )
where is the return on the event portfolio, is the risk-free rate, for which the rate on one-month German bond rate1 will be used, is the portfolio excess return, ( ) is the market excess return, and are the returns on two zero-investment portfolios that are explained below, and
is the error term.
Following Fama and French (1993), the expected return is calculated by using their three-factor model, mimicking portfolios for common risk factors in returns. Next to the market excess return, the common risk factors influencing stock returns are firm size and book-to-market ratio. To account for both of these influences, the following method is used to include them in the regression: For each calendar year six portfolios are formed by ranking all securities traded on stock exchanges of European Union countries (the same countries as where the data on acquisitions came from) on two variables. The first variable is book-to-market ratio. The firms stocks are split into categories based on if they are “high”, “medium” or “low” book-to market stocks. The “high” (from here on known as H) book-to-market stocks are the 30% with the highest book-to-market ratio over the previous year. The “low” book-to-market stocks (from here on known as L) are the 30% with the lowest book-to-market ratio over the previous year and the remaining 40% in the middle are the “Medium” book-to-market stocks (from here on known as M).
13 The firms are then also ranked by size (price times the number of shares). The largest 50% of firms form the “big” portfolio (from here on known as B) and the smaller 50% of firms form the “small” portfolio (from here on known as S). The reason firms are only split up in two portfolios on size compared to three on book-to-market ratio lies in the significance of the influence. The book-to-market ratio has a stronger influence on average stock return than size (Fama and French, 1993).
At this point, six portfolios can be formed (S/L, S/M, S/H, B/L, B/M and B/H) where S/L contains firms in the lower half of the size ranking and in the lowest 30% of the book-to-market ranking. In order to include the size effect and book-to-market equity effect in the test, two portfolios are formed: Small Minus Big (SMB) and High Minus Low (HML). SMB is calculated by deducting the simple average returns of the small portfolios (S/L, S/M and S/H) from the simple average returns of the large portfolios (B/L, B/M and B/H). HML is formed by deducting the simple average returns of the low market-to-book equity firms (S/L and B/L) from the high market-to-book equity firms (S/H and B/H). Following Mitchell and Stafford (2000), Bouwman et al. (2009) and Lougran and Ritter (2000), these portfolios are rebalanced every year.
The CTAR method has several shortcomings that must be dealt with. The model assumes the number of firms exiting and entering the portfolio is constant. Because these events, such as mergers, are clustered over time and the portfolio is updated each month, the number of firms entering and exiting varies significantly (Fama and French, 1998). There are periods where most acquisitions come from one particular industry. Therefore, the portfolio is likely to be heavily influenced by a certain industry at each point in time. At longer intervals these industries’ influences change, which could lead to biased estimates (Mitchell and Stafford, 2000).
Loughran and Ritter (2000) argue that a weighted least squares regression must be used to solve this problem. Mitchell and Stafford (2000), however, find evidence that weighted least squares regression does not solve the problem and argue for another solution. They recommend requiring a minimum of ten firms in the event portfolio at each measured month to mitigate the problem. They also find no evidence that ordinary least squares (OLS) performs poorly in detecting abnormal return performance and thereby argue that a minimum of 10 firms in the portfolio in each month is sufficient to solve the problem of portfolio entering and exiting activity. I follow the recommendations of Mitchell and Stafford (2000) and require a minimum of ten firms for the portfolio.
14 the portfolio changes over time, the residual variance may also be changing. This may cause the OLS estimator to be inefficient. Mitchell and Stafford’s (2000) solution is to add dummy variables to make distinctions in periods of high, low and average event activity. In this study, White’s corrections will be used to control for heteroscedasticity.
4.3 Classification of the market state
Classifying the market state is of critical importance in this study. The Substantial Up and Down method (SUD) is developed by Fabozzi and Francis (1977). Here, months are classified only if they make a substantial movement where the total market value exceeds the average market value plus or minus one half of the standard deviation of the market value of the entire sample period. Thus, a substantial up (down) market occurs if the market value of a month is above (below) the average market value plus (minus) one half of the standard deviation of the market value. All months not fulfilling that criterion (average months) are ignored. The months are classified based on market trends, meaning that when a market is up for only one month in a row, the month is classified as an average month. So a minimum of two consecutive months with the same classification is required to classify a month as up or down.
This definition is still being applied by many other researchers in classifying the market condition, for example, Woodward and Anderson (2009), Iorio and Faff (2000) and Lubatkin and Chatterjee (1991). However, they each use different numbers of consecutive months to classify a trend and they vary in the amount of standard deviations by which month’s return must exceed the average return.
15 Figure 1. Market state classification and detrended S&P 350 Europe index
The detrended Speu-350 index and the market state classification over the sample period. The detrended Speu-350 index is calculated by subtracting the sample’s best straight-line fit from the monthly index value data. Market state classification is where a value of 1 equals high market, 0 medium market and -1 low market.
In Table 3, the number of acquisitions that have been announced in each market state, subdivided by payment method, are presented. It shows more than twice as many acquisitions are announced in the high market state compared to the low market state. Further, the number of acquisitions paid with shares divided by the number of acquisitions paid with cash is shown in the table. This ratio indicates that more shares acquisitions are announced in high state markets.
Table 3. Number of acquisitions per market state and payment method
In this table, rows show the number of acquisitions announced in high-, medium-, and low market state. The first four columns show the number of acquisitions in total and then separated into payment methods, where ‘Financed with other payment method’ includes acquisitions paid with a combination of cash and shares and acquisitions paid with methods not classified as cash or shares. The final column shows the number of share acquisitions divided by the number of cash acquisitions.
Number of acquisitions
Shares to cash Ratio Market State Total
Financed with shares Financed with cash Financed with other payment method High market 464 82 275 107 0,30 Medium market 319 48 209 62 0,23 Low market 199 37 139 23 0,27 Total 982 167 623 192 1-1-1997 28-9-1999 24-6-2002 20-3-2005 15-12-2007 10-9-2010 -1,5 -1 -0,5 0 0,5 1 1,5 200 400 600 800 1000 1200 1400 1600 1800 2000 1-1997 1-12 -19 9 7 1- 11-199 8 1- 10-199 9 1- 9-2000 1- 8-2001 1- 7-2002 1- 6-2003 1- 5-2004 1- 4-2005 1- 3-2006 1- 2-2007 1-2008 1- 12-200 8 1- 11-200 9 1- 10-201 0
16 4.4 Classification of firm-specific timing
Firm-specific timing acquisitions are those acquisitions in which the acquirer announces the acquisition at a moment the stock value of the firm is at a high point. A takeover is classified as one with firm-specific timing if the stock value of the acquiring firm at the moment of acquisition announcement is at least 85% of the maximum stock value for that firm over the previous 24 months. Table 4 shows the number of acquisitions for these classifications.
Table 4. Number of acquisitions per firm stock timing classification and payment method In this table, rows show the number of acquisitions with firm-specific timing and without firm-specific timing. The first four columns show the number of acquisitions in total and then separated into payment methods, where ‘Financed with other payment method’ includes acquisitions paid with a combination of cash and shares and acquisitions paid with methods not classified as cash or shares. The final column shows the number of share acquisitions divided by the number of cash acquisitions.
Number of acquisitions
Shares to cash Ratio Market State Total
Financed with shares Financed with cash Financed with other payment method Firm-specific timing 479 82 291 106 0,28 No firm-specific timing 503 85 332 86 0,25 Total 982 167 623 192 4.5 Classification of herding
For testing the theory of managerial herding, acquisitions at different points in a merger wave are compared. Following Bouwman et al. (2009), the high market portfolio from paragraph 4.3 will be split into early- and late movers, where early (late) movers are classified as the first (last) 20% that announced an acquisition in a merger wave. As in Bouwman et al. (2009), a merger wave starts when the first acquisition is announced in high market classification. The end of a merger wave is when the first acquisition is announced in the average market state after a period of high market.2 Table 5 shows the number of acquisitions of early movers and late movers and the financing methods.
2 Other measures of merger waves (based purely on the number of acquisitions) might be preferable here, but considering the
17 Table 5. Number of acquisitions for early and late movers
In this table, the rows show the number of acquisitions from early movers and late movers. The first four columns show the number of acquisitions in total and then separated into payment methods, where ‘Financed with other payment method’ includes acquisitions paid with a combination of cash and shares and acquisitions paid with methods not classified as cash or shares. The final column shows the number of share acquisitions divided by the number of cash acquisitions.
Number of acquisitions
Shares to cash Ratio Market State Total
Financed with shares Financed with cash Financed with other payment method Early mover 93 16 50 27 0,32 Late mover 93 12 57 24 0,21 Total 186 28 107 51
4.6 Portfolio return descriptive statistics
18 Table 6. Descriptive statistics portfolio returns
In this table, the descriptive statistics of all portfolios used for the regression are given. These statistics are based on portfolio returns before the corrections for abnormal return are made with the CTAR method.
Portfolio Average Median Maximum Minimum
Standard Deviation All acquisitions 0,056 0,207 5,046 -5,864 1,624 High market -0,176 0,014 9,940 -9,437 3,137 Medium market 0,061 0,185 2,424 -4,182 1,113 Low market 0,606 0,387 4,468 -3,526 1,360
High market (shares payment) -0,032 0,179 10,964 -9,651 2,857
High market (cash payment) -0,652 -0,497 15,672 -15,228 5,163
Firm-specific timing 0,007 0,153 5,578 -5,913 1,573
No firm-specific timing 0,108 0,198 4,984 -6,981 1,862
Firm-specific timing (shares payment) -0,394 -0,064 11,392 -9,801 4,249
No firm-specific timing (shares payment) -0,003 0,146 7,256 -8,175 2,344
First movers -0,602 -0,476 10,119 -11,494 3,875
Last movers 0,140 0,180 9,378 -9,233 3,636
4.7 Regressions to test differences between portfolios
To obtain the abnormal returns of each classification with CTAR, separate portfolios are constructed, where the abnormal return is estimated as in equation one for each subsample. Subsamples will also be drawn based on the method of payment. In each of the three control variables (market excess return,
high minus low and small minus big) the average is subtracted from each observation, thereby
setting the averages to zero.
To test the coefficient and the significance of the difference between the two portfolios, a pooled sample with a dummy variable is used. Here both portfolios (from here on known as groups) are put together into one pooled portfolio, where the dummy is zero for the first group and one for the second group.
This results in the following equation:
(2) ( )
19 Similar to the tests of equation one, the averages of control variables are subtracted from the data. The averages of the control variables ((
),
,
)
are calculated separately for both groups of each pooled sample. The average of the first group of each control variable is subtracted from the first group’s observation and the average of the second group is subtracted from each second group’s observation. The average of the dummy variable is, however, not calculated and subtracted separately for each group, but for the entire sample.5.
Results
In this section, the regression results of all portfolios’ abnormal returns using the Calendar-Time Abnormal Return methodology are discussed. First, the result for the portfolio of all acquisitions is presented. Section 5.1 shows whether acquisitions made in high-valuation markets underperform relative to acquisitions made in low-valuation markets (hypothesis 1a). This is followed by whether acquisitions financed with shares in high-valuation markets underperform more than acquisitions financed with cash in high-valuation markets (hypothesis 1b). In section 5.2, the regressions of hypothesis 2a and 2b, which concern the influence of firm-specific timing on long-run performance, are examined. Finally, section 5.3 shows the results of hypothesis 3 concerning the connection between timing in an acquisition wave and long-run performance.
I apply Ordinary Least Squares (OLS) regression to estimate the abnormal returns in the event portfolios. Several sub-portfolios have been constructed to test each hypothesis. For example, the high-market valuation portfolio is subdivided into a high market portfolio financed with cash and a high market portfolio financed with shares. Regarding these results, the reader must bear in mind that the CTAR method for calculating abnormal return leads to more extreme coefficients of the returns compared to other abnormal return calculation methods such as Buy and Hold Abnormal Returns or Cumulative Abnormal Returns, as was found by Bouwman et al. (2009).
20 5.1 Market timing
This subsection discusses the differences between high- and low market acquisitions, followed by a closer look at the differences between the types of financing. In Table 7.1 the high market acquisitions show a significantly negative long-run abnormal return, where the low market acquisitions have a significantly positive long-run abnormal return. The significance of this difference is confirmed by the regression of the pooled high market and low market sample. Here, the coefficient of the dummy variable (significant at the 1% level) indicates acquisitions announced at a moment of high market value underperform compared to those announced at a moment of low market value by -0.825% per month. This confirms hypothesis 1a, that low market acquisitions perform better than high market acquisitions. This is consistent with the findings of Schleifer and Vishny (2003), Rhodes-Kropf and Viswanathan (2004) and Bouwman et al. (2009), which show that overvaluation has a negative impact on acquisition performance.
In Table 7.2, hypothesis 1b is tested. The high market acquisitions paid with shares are compared to the high market acquisitions paid with cash. The separate samples show that acquisitions financed with shares underperform more than acquisitions financed with cash. The pooled portfolio regression confirms this, for the coefficient of the dummy is -0.637% (significant at the 5% level). This means the high market cash payment acquisitions underperform less than the high market acquisitions financed with shares. This finding is consistent with the findings of Rhodes-Kropf and Viswanathan (2004), showing that target firms are more inclined to accept share offers in high-valuation markets. This ultimately leads to a price correction in the long run.
5.2 Firm-specific timing
In this subsection, the long-run abnormal returns of firms that time their acquisitions at moments of high firm stock value to exploit their temporary overvaluation are investigated. Table 8 shows the results of the regressions.
In Table 8.1, the intercept of the firm-specific timing portfolio is more negative than that for the portfolio without firm-specific timing. However, no proof of a significant difference is found in the pooled portfolio. The coefficient for the dummy variable (ζ) is -0.101%, but it is not significant. Therefore, in general, with all payment methods included in the sample, acquirers that time their takeover at a moment of high firm stock value do not underperform relative to acquirers that do not time their takeover when firm stock value is high. Hypothesis 2a is therefore not confirmed.
21 acquisitions without firm-specific timing and one of share-financed acquisitions with firm-specific timing. The intercepts of the firm-specific timing acquisitions of both portfolios are significantly negative, with coefficients of -0.239% and -0.491% respectively. The result of the pooled sample shows that share acquisitions timed at a moment of high firm stock value underperform compared to acquisitions financed with shares that are not timed at high firm stock value by a stock return of 0.296% per month (significant at the 10% level). This reveals that acquisitions financed with shares that are timed around a peak of stock value underperform in the long run compared to acquisitions financed with shares that are not timed around the peak stock value. This finding confirms hypothesis 1b.
5.3 Early movers versus late movers in an acquisition wave
22 Table 7. Results Ordinary Least Squares regressions
This table presents the results of OLS regressions, which estimate monthly abnormal long–term performance for three years after acquisition announcement for equally weighted acquisition portfolios in percentages. The acquisitions have been performed by firms in countries that are member of the European Union as of 1997. The sample period is from January 1999 through December 2010. In 7.1, the first regression includes all acquisitions from the dataset. The second and third regressions are from subsamples. They include, respectively, all acquisitions announced in a low market period and all acquisitions announced in a high market period. In 7.2, the first regression includes the High Market sample and the following two include the High Market (Cash Payment) acquisitions and High Market (Shares Payment) acquisitions. These are all estimated with the following equation:
( )
where is the monthly event portfolio return minus the risk free rate. For the risk-free rate, I use the one month German
bond rate. is the estimate of abnormal return. These are regressed against the Fama-French three factor model, where (
) equals the market excess return, is the return difference between firms with a high market-to-book value and firms with a
low market-to-book value, and is the return difference between big and small firms. In the pooled portfolios the portfolio returns are put together with an added dummy variable that estimates the difference in returns. The pooled portfolios are estimated with the following equation:
( )
In 7.1 the dummy variable D is zero for the Low Market acquisition portfolio and one for the High Market acquisition portfolio. In 7.2 the dummy variable D is zero for the High Market (Cash Payment) portfolio and one for the High Market (Shares Payment) portfolio. captures the difference between the two portfolios. ***, ** and * indicate a significance level of, respectively, 1, 5 and 10%. 7.1 Estimate in percentages, (t-value) Acquisition type β h s ζ Adjusted R² # of observations All acquisitions -0.181 21.717 -3,158 0.471 0.769 144 (-2.690)*** (19.768)*** (-2.887)*** (0.373) Low market 0.403 13.500 1.324 -3.467 0.241 103 (3.426)*** (4.913)*** (0.480) (-1.418) High market -0.422 40.018 -7.225 2.365 0.801 110 (-3.124)*** (19.555)*** (-3.577)*** (0.961)
Pooled: High Market & Low Market -0.023 32.554 -4.726 1.066 -0.825 0.640 213 (-0.228) (17.754)*** (-2.598)*** (0.538) (-4.011)*** 7.2 Estimate in percentages, (t-value) Acquisition type β h s ζ Adjusted R² # of observations
High Market (Cash Payment) -0.279 35.874 -1.317 -4.071 0.801 108 (-2.244)** (19.161)*** (-0.699) (-1.807)*
High Market (Shares Payment) -0.915 53.699 -34.427 17.869 0.804 86 (-3.690)*** (15.588)*** (-9.934)*** (4.182)***
Pooled: High Market (Cash Payment) & High Market (Shares Payment)
-0.561 44.214 -17.209 6.210 -0.637 0.691 194
(-3.446)*** (18.738)*** (-7.244)*** (2.154)** (-1.942)**
23 Table 8. Results Ordinary Least Squares regressions
This table presents the results of OLS regressions, which estimate monthly abnormal long–term performance for three years after acquisition announcement for equally weighted acquisition portfolios. The estimates are in percentages. The acquisitions have been performed by firms in countries that are member of the European Union as of 1997. The sample period is from January 1999 through December 2010. In 8.1, the first two regressions include, respectively, all acquisitions announced without firm-specific timing and with specific timing. Firm-specific timing occurs when a acquisition is announced when firm stock value at least 85% of the maximum value in the past 2 years. In 8.2, the first two regressions include the sample of No Firm-specific Timing with Share Payment and Firm-specific Timing with Share Payment. These are all estimated with the following equation:
( )
where is the monthly event portfolio return minus the risk free rate. For the risk-free rate, I use the one month German
bond rate. is the estimate of abnormal return. These are regressed against the Fama-French three factor model, where (
) equals the market excess return, is the return difference between firms with a high market-to-book value and firms with a
low market-to-book value, and is the return difference between big and small firms. In the pooled portfolios the portfolio returns
are put together with an added dummy variable that estimates the difference in returns. The pooled portfolios are estimated with the following equation:
( )
In 8.1 the dummy variable D is zero for the No Firm-specific Timing acquisition portfolio and one for the Firm-specific Timing acquisition portfolio. In 8.2 the dummy variable D is zero for the No Firm-specific Timing (Share Payment) portfolio and one for the Firm-specific Timing (Share Payment) portfolio. captures the difference between the two portfolios. ***, ** and * indicate a significance level of, respectively, 1, 5 and 10%.
8.1
Estimate in percentages, (t-value)
Acquisition type β h s ζ Adjusted R²
# of observations No firm-specific timing -0.129 23.900 -5.412 2.747 0.711 144 (-1.518) (17.246)*** (-3.923)*** (1.727)* Firm-specific timing -0.229 20.634 -1.061 -1.740 0.747 144 (-3.398)*** (18.709)*** (-0.967) (-1.374)
Pooled: Firm-specific timing &
No firm-specific timing (-3.247)*** -0.179 (24.710)*** 22.295 (-3.608)*** -3.210 (0.049) 0.476 (-0.912) -0.101 0.716 288 8.2 Estimate in percentages, (t-value) Acquisition type β h s ζ Adjusted R² # of observations No Firm-specific Timing (Shares Payment) (-2.435)** -0.239 (19.240)*** 30.586 (-8.225)*** -13.417 (4.582)*** 8.515 0.782 129 Firm-specific Timing (Shares
Payment) (-2.790)*** -0.491 (11.539)*** 29.909 (-7.469)*** -19.010 (2.621)** 8.195 0.672 107 Pooled: Firm-specific Timing
& No Firm-specific Timing (Both Shares Payment)
-0.373 30.355 -16.343 8.355 -0.296 0.716 236
(-3.882)*** (24.405)*** (-10.954)*** (4.727)*** (-1.735)*
24 Table 9. Results Ordinary Least Squares regressions
This table presents the results of OLS regressions, which estimate monthly abnormal long–term performance for three years after acquisition announcement for equally weighted acquisition portfolios. The estimates are in percentages. The acquisitions have been performed by firms in countries that are member of the European Union as of 1997. The sample period is from January 1999 through December 2010. The first regression includes the first 20% of high market acquisitions (First Movers). The second regression includes the last 20% of high market acquisitions (Last Movers). These are all estimated with the following equation:
( )
where is the monthly event portfolio return minus the risk free rate. For the risk-free rate, I use the one month German
bond rate. is the estimate of abnormal return. These are regressed against the Fama-French three factor model, where (
) equals the market excess return, is the return difference between firms with a high market-to-book value and firms with a
low market-to-book value, and is the return difference between big and small firms. In the pooled portfolio, the portfolio returns are put together with an added dummy variable that estimates the difference in returns. The pooled portfolio is estimated with the following equation:
( )
The dummy variable D is zero for the First Movers acquisition portfolio and one for the Last Movers acquisition portfolio. captures the difference between the two portfolios. ***, ** and * indicate a significance level of, respectively, 1, 5 and 10%.
Estimate in percentages, (t-value) Acquisition type β h s ζ Adjusted R² # of observations First Movers -0.898 43.867 -9.633 -1.128 0.812 84 (-4.861)*** (17.129)*** (-3.863)*** (-0.391) Last Movers -0.070 41.157 -4.238 -2.263 0.820 76 (-0.392) (17.313)*** (-1.644) (-0.746)
Pooled: First Movers & Last Movers
-0.505 42.523 -7.266 -2.028 0.828 0.816 160
25
6.
Conclusion and discussion
In this thesis, I investigate the effects of acquisition timing on long-run stock performance. I study the long-run abnormal monthly returns of 982 European acquisitions during the period 1999 to 2011 using the Calendar-Time Abnormal Return (CTAR) method. By constructing subsamples, different theories of timing effects are tested: Market timing, firm-specific timing and herding theory. The market timing theory states that target firms accept more bids with share financing when the market is positively misvalued, resulting in a price correction for the acquirers on the long-run. Firm-specific timing theory investigates a very similar effect, only that acquirers time acquisitions based on their own stock value instead of the market value. Acquirers might take advantage of its temporary high stock value and time an acquisition financed with shares when this is the case. Again, the eventual stock price correction results in a negative abnormal return. Finally, I investigate the managerial herding theory in which acquisitions announced in the beginning of a merger wave are assumed to perform better than acquisitions announced at the end of a merger wave. Readers must bear in mind that the CTAR method results in more extreme return coefficients than other methods of calculating long-run abnormal return.
I find that acquiring firms in general underperform at -0.181% per month for the three years after acquisition announcement. Consistent with the findings of Rhodes-Kropf and Viswanathan (2004), Bouwman et al. (2009) and Croci, Petmezas and Vagenas-Nados (2010), I find that high market acquisitions underperform compared to low market acquisitions. High market acquisitions long-run abnormal returns are -0.422% per month and that low market acquisitions have a 0.403% monthly abnormal return. The pooled portfolio shows that the difference is significant; thus, acquisitions announced when the stock market is at a point of high value perform worse by 0.825% per month compared to acquisitions announced when the stock market is at a point of low valuation. This difference is equal to 10.362% on a yearly basis.
Share acquisitions announced in high market states underperform relative to cash acquisitions announced in high market states with -0.637% per month (significant at the 5% level). This is a confirmation of the theory of Rhodes-Kropf and Viswanathan (2004), who state that acquisition targets are more inclined to accept a firm’s stock bid when market valuations are high.
26 financed with shares announced at high levels of acquirer stock value provide 0.296% (significant at the 10% level) less monthly return in the long run compared to share-financed acquisitions announced when the acquirer’s stock value is not at a high level. This is consistent with the findings of Rhodes-Kropf, Robinson and Viswanathan (2005) that suggest that firms exploit this high stock price level to finance acquisitions with shares. In the long run, a stock price correction causes these firms to perform worse than firms not exploiting this overvaluation.
Contradictory to the findings of Bouwman et al. (2009), I find no evidence that managerial herding leads to long-run underperformance. Bouwman et al. (2009) found that first movers in a merger and acquisition wave perform better than late movers. This finding is explained by managerial herding behaviour, where initiators of innovations or new trends are more successful than imitators at the end of a trend. My evidence points in the opposite direction. I find that first movers in an acquisition wave have 0.828% less monthly returns compared to last movers.
The difference between these findings and the findings of Bouwman et al. (2009) can be partly explained by the difference in sample geography. I test acquisitions in the European Union, whereas Bouwman et al. (2009) test acquisitions from the United States. Moreover, the measure of abnormal performance used to test for this theory differs. I use the CTAR method where Bouwman et al. (2009) use BHAR and three-day Cumulative Abnormal Returns (CAR). While they do use the CTAR method for measuring underperformance of high market acquisitions, they do not use it in tests for their herding theory.
No matter what the cause is, the fact that I find significant contrary evidence on early and late movers in a wave puts the herding theory into question and points in a different direction. I find that acquisitions that are announced at the end of a merger wave are more profitable in the long run. An explanation can be that acquirers learn from the experiences of previous acquirers, which results in better performance.
28
7.
References
André, Paul, Maher Kooli, and Jean-François L’Her, 2004, The long-run Performance of Mergers and Acquisitions: Evidence from the Canadian Stock Market, Financial Management 33, 27-43.
Agrawal, Anup, and Jeffrey F. Jaffe, 2000, The post-merger performance puzzle, Advances in Mergers
and Acquisitions 1, 7-41.
Bouwman, Christa H. S., Kathleen Fuller, and Amrita S. Nain, 2009, Market valuation and acquisition quality: Empirical evidence, The Review of Financial Studies 22, 633-679.
Capron, Laurence, and Jung-Chin Shen, 2007, Acquisitions of private vs. public firms: private information, target selection, and acquirer returns, Strategic Management Journal 28, 891-911.
Conn, Robert Charlie L., Andy Cosh, Paul M. Guest, and Alan Hughes, 2005, The impact on UK acquirers of domestic, cross-border, public and private acquisitions, Journal of Business Finance &
Accounting 32, 815-870.
Croci, Ettore, Dimitris Petmezas, and Evangelos Vagenas-Nanos, 2010, Managerial overconfidence in high and low valuation markets and gains to acquisitions, International Review of Financial Analysis 19, 368-378.
Dutta, Shantanu, and Vijay Jog, 2009, The long-term performance of acquiring firms: A re-examination of an anomaly, Journal of Banking & Finance 33, 1400-1412.
Fabozzi, Frank J., and Jack Clark Francis, 1977, Stability tests for alphas and betas over bull and bear market conditions, The Journal of Finance 32, 1093-1099.
Fama, Eugene F., and Kenneth R. French, 1993, Common risk factors in the returns on stocks and bonds,
Journal of Financial Economics 33, 3-56.
Fama, Eugene F., 1998, Market efficiency, long-term returns and behavioral finance, Journal of Financial
Economics 49, 283-306.
Gort, Michael, 1969. An economic disturbance theory of mergers. Quarterly Journal of Economics 83, 624-642.
Harford, Jarrad, 2005, What drives merger waves, Journal of Financial Economics 77, 529-560.
Huang, Rongbing, and Jay R. Ritter, 2004, Testing the market timing theory of capital structure, working paper, University of Florida.
Iorio, Amalia Di, and Robert Faff, 2000, An analysis of asymmetry in foreign currency exposure of the Australian equities market, Journal of Multinational Financial Management 10, 133-159.
29 Jensen, Michael C., 1993. The modern industrial revolution, exit, and control systems. Journal of Finance 48, 831- 880.
Loughran, Tim, and Jay R. Ritter, 2000, Uniformly least powerful test of market efficiency, Journal of
Financial Economics 55, 361-389.
Lubatkin, Michael, and Sayan Chatterjee, 1991, The Strategy-Shareholder Value Relationship: Testing Temporal Stability Across Market Cycles, Strategic Management Journal 12, 251-270.
Megginson, William L., Angela Morgan, Lance Nail, 2004, Journal of Banking & Finance 28, 523-552. Mitchell, Mark L., and Harold Mulherin, J., 1996. The impact of industry shocks on takeover and restructuring activity, Journal of Financial Economics 41, 193-229.
Mitchell, Mark L, and Erik Stafford, 2000, Managerial decisions and long-term stock price performance,
The Journal of Business 73, 287-329.
Moeller, Sara B., Frederik P. Schlingemann, René M. Stulz, 2004, Firm size and the gains from acquisitions, Journal of Financial Economics 73, 201-228.
Persons, John C., and Vincent A. Warther, 1997, Boom and bust patterns in the adoption of financial innovations, Review of Financial Studies 10, 939-967.
Rhodes-Kropf, Matthew, David T. Robinson, S. Viswanathan, 2005, Valuation waves and merger activity: The empirical evidence, Journal of Financial Economics 77, 561-603.
Rhodes-Kropf, Matthew, S. Viswanathan, 2004, Market valuation and merger waves, The Journal of
Finance 59, 2685-2718.
Ritter, Jay R., 1991, The long-run performance of initial public offerings, The Journal of Finance 46, 3– 27.
Schleifer, Andrei, and Robert W. Vishny, 2003, Stock market driven acquisitions, Journal of Financial
Economics 70, 295-311.
Soegiharto, 2008, Drivers of merger waves, Gadjah Mada International Journal of Business 10, 1-23. Tuch, Christian, and Noel O’Sullivan, 2007, The impact of acquisitions on firm performance: A review of the evidence, International Journal of Management Reviews 9, 141-170.
Woodward, George, and Heather M. Anderson, 2009, Does beta react to market conditions? Estimates of ‘bull’ and ‘bear’ betas using a nonlinear market model with an endogenous threshold parameter,
