Information Asymmetry and the
Revaluation Effect after Failed Mergers
and Acquisitions
July 1, 2017
Abstract
Through the use of exogenously withdrawn mergers and acquisitions, this is the first paper that investigates the effect of information asymmetry on failed mergers and acquisi-tions. Based on the findings of Malmendier et al. (2016) and Liu (2017), a difference in revaluation between opaque and transparent targets is expected. Two of the three measures of information asymmetry used by this paper confirm this variation. While the difference in revaluation between cash and stock bids is still present, the results show a lower revalua-tion for opaque cash targets compared to transparent cash targets, and a higher revaluarevalua-tion for opaque stock targets compared to transparent stock targets.
Master Thesis
Name Thijs Dijkman
Student nr. 10359524
Supervisor prof. dr. T. Caskurlu Specialisation Corporate Finance
University University of Amsterdam, Amsterdam Business School Faculty Faculty of Economic and Business
Statement of Originality
This document is written by Thijs Dijkman who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Contents
1
Introduction
3
2
Literature review
5
2.1
Drivers of mergers and acquisitions and method of payment . .
5
2.2
Information asymmetry in mergers and acquisitions . . . .
7
2.3
The revaluation effect . . . .
9
2.4
Hypotheses development . . . .
10
3
Data and Methodology
11
3.1
Data . . . .
11
3.2
Methodology . . . .
13
3.2.1
Model description
. . . .
13
3.2.2
Exogenously withdrawn sample . . . .
15
4
Descriptive statistics
17
4.1
Sample description . . . .
18
4.2
Revaluation cash versus stock
. . . .
22
5
Results
23
5.1
Information asymmetry on the choice for cash or stock . . . .
24
5.2
Opaque versus transparent targets
. . . .
27
5.3
Information asymmetry on target revaluation . . . .
29
6
Conclusion
34
1
Introduction
Mergers and acquisitions is one of the most impacting events during the lifetime of a company. The value implications of mergers and acquisitions are under high scrutiny. While many studies focus on the firms that successfully engage in a takeover transaction little is known about the ones that fail in these endeavours. Still, a lot can be learned by investigating these types of events. When a bid fails to succeed, there is a revaluation of the target’s stock price, meaning the stock price before the announcement of the bid is different compared to the stock price after the failure. Malmendier et al. (2016) investigate this revaluation and find a significant difference in revaluation between cash and stock bids. They show that this revaluation effect is larger for cash offers than for stock offers, implying that the market revalues a company at a higher price when a bidder makes a cash offer compared to when a bidder makes a stock offer. However, other characteristics of the deal could also affect this revaluation. In this paper, the effect of information asymmetry on the revaluation effect is investigated.
One explanation for the difference in cash and stock revaluation relates to the valuation of the acquirer and the target. Shleifer & Vishny (2003) find that firms are incentivised to get their equity overvalued so that they can make “cheaper” acquisitions with stock. This is in line with the market-timing theory, which predicts that a company will use stock instead of cash when the managers believe their company is overvalued (Baker & Wurgler, 2002). For stock bids this implies that investors receive a signal of general overvaluation (Rhodes-Kropf et al., 2005). For cash bids this implies that investors receive a signal of general undervaluation (Dong et al., 2006). Another related finding is that Liu (2017) shows that private equity is able to spot undervalued firms and “cherry pick” their leveraged buyouts. He also finds that firms that are considered to have higher information asymmetry have higher revaluation compared to transparent firms.
These previous findings result in the question of whether or not information asymmetry of the target firm affects the revaluation in failed cash and stock acquisitions. First, the choice between using cash and stock itself could be affected by information asymmetry issues. Stock can be used to align the target’s interest with the interest of the acquirer and yield the potential synergies to both sides (Eckbo et al., 1990). If there is a high level of uncertainty about the amount of synergies that can be obtained in the takeover, a stock acquisition can be used to address the risk of overpayment. To further enhance this research, this paper first investigates if information asymmetry drives the choice for cash or stock. Using a probit regression, I
first investigate whether opaque targets are more likely to receive a cash bid or more likely to receive a stock bid. Then, I investigate the effect of information asymmetry on the revaluation of the targets. To minimize endogeneity issues, an exogenously withdrawn sample is created by extensive hand-collected data research with the use of Lexis-Nexis. Using this exogenously withdrawn sample, I reconstruct the revaluation effect discovered by Malmendier et al. (2016). Furthermore, the difference in opaqueness within both cash and stock are investigated. Finally, a regression controlling for other factors is used to discover differences in revaluation between cash and stock, and their level of information asymmetry.
The first part of my research shows that two of the three measures of information asym-metry used for this research do affect the choice in method of payment. The results show that opaque targets are more likely to be acquired with stock acquisitions. These results confirms the hypotheses of Hansen (1987); Fishman (1989) and Eckbo et al. (1990).
The results provided by the research of the relationship between information asymmetry and the revaluation of failed cash and stock acquisitions are harder to assess. While stock targets still generally have a lower revaluation than cash targets, two of my three information asymmetry measures imply a higher revaluation for opaque stock targets compared to transpar-ent stock targets, and a higher revaluation for transpartranspar-ent cash targets compared to opaque cash targets. This results in a converging difference in revaluation between cash and stock targets when information asymmetry is high.
This paper contributes to the existing literature by elaborating on the revaluation effect after failed mergers and acquisitions. It elaborates on the previously mentioned paper by Mal-mendier et al. (2016) by incorporating the additional element of information asymmetry on the revaluation effect. It also brings notice to the theory of misvaluation driving the mergers and acquisition market, researched by Shleifer & Vishny (2003) and Rhodes-Kropf et al. (2005). Furthermore, this paper brings notice to the concept of the choice for cash and stock being dependent on the level of information asymmetry, and if stock is used to minimize cost of information asymmetry (Hansen, 1987).
The paper is organised in the following manner. In Chapter 2, the related literature is reviewed. I first review previous empirical studies which presented different underlying reasons as to why mergers and acquisitions occur. The papers of Rhodes-Kropf et al. (2005); Dong et al. (2006); Gorbenko & Malenko (2014) that cover the relationship between stock market valuation and takeover activity are discussed. Next, literature about information asymmetry in mergers
and acquisitions is reviewed (Dionne et al., 2010; Andrade et al., 2001; Chang & Suk, 1998). Subsequently, recent literature about the revaluation effect is examined (Malmendier et al., 2016; Liu, 2017). Furthermore, the hypotheses of this research are formulated. In Chapter 3, the data collection and the empirical strategy of the paper is presented. In order to compensate for endogeneity issues, this paper makes use of an “exogenous” sample collection based on similar empirical strategies used in related literature (Liu, 2017; Malmendier et al., 2016; Chang & Suk, 1998). Chapter 4 provides the data description, and Chapter 5 defines the results. Finally in Chapter 6, the limitations and final conclusion of the research are discussed.
2
Literature review
In this chapter the existing literature on this topic is reviewed. To understand where the revalu-ation effect comes from the drivers of the mergers and acquisitions market are discussed. After this examination, theories of the effect of information asymmetry on mergers and acquisitions are discussed. Subsequently, studies investigating the revaluation effect are reviewed. Finally, using the discussed theories the hypotheses for this empirical study are stated.
2.1
Drivers of mergers and acquisitions and method of payment
First of all, the neoclassical economical explanation of a merger or acquisition is that “the whole has to be greater than the sum of the parts”. In other words, the reason for a takeover are the potential synergies that can be obtained. In real life, the underlying reason to engage in a takeover differs among the different situations. The company’s strategy, empire building or financial opportunities are all reasons to do an acquisition (Humphery-Jenner, 2012; Gorbenko & Malenko, 2014). The one thing that is certain, is that the acquiring company has to pay, and most often a premium to the current share price. However, the value of a company can be estimated in many different ways. In the case of public companies millions of investors set the price of a company for you. If a company becomes relatively cheap it could become a more attractive takeover target. This raises the question if the investor misvaluation in public companies drives the takeover market. Different studies show evidence that this is indeed the case (Rhodes-Kropf et al., 2005; Dong et al., 2006; Gorbenko & Malenko, 2014).
In the merger and acquisition market we observe observe so-called merger waves. Shleifer & Vishny (2003) argue that the clustering in merger activity comes from the state of current
stock market valuation. They developed a model which explains why firms choose to pay either in cash or in stock in their merger and acquisition transactions. Their model predicts that in periods of overvaluation, bidders with overvalued stock use stock-for-stock acquisitions to buy undervalued companies. The underlying reason behind this theory is that firms buy (undervalued) real assets with their overvalued stock (Shleifer & Vishny, 2003). Therefore, firms with overvalued equity are more likely to become an acquirer in takeovers and firms with undervalued equity are more likely to become a target in takeovers.
Rhodes-Kropf & Viswanathan (2004) elaborate on the theory of Shleifer & Vishny (2003). They argue that the simple explanation for overvalued bidders preferring to use stock is incom-plete because targets can simply reject the offer. However, Rhodes-Kropf et al. (2005) do also find that misvaluation, both under and over valuation, drives the takeover market. In their paper they confirm that stock deals are more likely to occur when the acquirer is overvalued. They also find that cash deals are more likely to be made in times of general undervaluation of the market. Stock deals however are more likely to occur in times of general market overvaluation. In general, this results in cash targets being more undervalued than stock targets. This is in sup-port of the evidence of the paper by Malmendier et al. (2016), which shows the difference in revaluation between stock and cash mergers. If the market is aware of these empirical findings, the event that an acquirer uses cash to buy a company should result in a signal for the investor that the target is undervalued.
In fact, other empirical evidence confirms there are some differences in types of acquisitions which affect the stock market. Bhagat et al. (2005) investigated the stock returns for bidders and targets in case of mergers and acquisitions. They find that combined bidder-target stock returns are higher for hostile offers, lower for equity offers and lower for diversifying offers. They argue that these effects reflect revelation about bidder stand-alone value, not differences in gains from the potential synergies itself. They also find that bidders with low Tobin’s q on average have more negative announcement period returns. Whereas, target announcement period returns are negatively correlated to the targets’ Tobin’s q. Finally, the acquisition of a smaller target by a large bidder on average creates a smaller value improvement measured as a fraction of combined value, than combinations of similar-size firms. However, they do argue that there could be a revelation bias in their research (Bhagat et al., 2005). This bias shows that if the bid reveals favourable news to the market about stand-alone bidder value, the cause of the combined bidder-target equity return will wrongly be attributed to the expected
value improvement. In this research the exogenously withdrawn sample is created to resolve the revelation bias.
The key takeaway is that mergers and acquisitions can be driven by misvaluation in the stock market. However, in previous literature the potential synergies always affected the ex-pected returns of the mergers and acquisitions. In this research, withdrawn deals are used which eliminates the noise of potential synergies. This revaluation effect is also investigated by Malmendier et al. (2016). If firms pick undervalued companies we should observe a pos-itive revaluation effect of the target. However, the choice for cash and stock may be affected by asymmetric information about the target. Investors base their valuation on the information provided. This effect on asymmetric information in mergers and acquisitions is explained in the next section.
2.2
Information asymmetry in mergers and acquisitions
Asymmetric information is a well-known phenomenon that is present in almost every problem in the world. In the world of finance most often asymmetric information affects the behaviour of managers and investors, the price of an asset and premiums that are paid in mergers and acquisitions. Hansen (1987) and Fishman (1989) studied the role of asymmetric information in the choice for cash or stock in mergers and acquisitions. They assume two-sided information asymmetry, where the acquirer and the target have their own private information about their value. They show, that the division of the takeover gains becomes a function of the size of the bid and the medium of exchange. The bidder that offers a high bid has relatively high expected “overpayment cost”. On the other hand a low bid reduces the probability that the bid will be successful. Unsuccessful bids have “lost synergy gain” costs. Eckbo et al. (1990) concludes that in a cash offer, the acquirer bears the risk of overpayment. Whereas a stock offer transfers some of the mispricing risk to the target. Therefore, a stock offer might be a safer opportunity when there is a high level of information asymmetry.
Officer et al. (2009) investigate the relationship between target information asymmetry and the acquirer returns in mergers and acquisitions. They document a significant and substantial higher announcement return to stock acquirers when the target is difficult to value. They in-terpret their results as an reward for using stock as payment method when target’s assets and operations are difficult to value. The reward is higher for firms where operations are hard to value, e.g. high R&D expenses or a lot of intangible assets. The reward may come from
acquir-ers who might be better to value the target’s assets. Opacity could first result in a target being more misvalued and naturally attract acquirers. Subsequently, this misvaluation might trigger the assumption that the target was undervalued in the first place and should recieve a positive revaluation.
So do acquirers benefit from having better information than investors? In a study by Dionne et al. (2010) mergers and acquisitions are compared to auctions. When an auction starts, the more people that bid the more becomes clear about the value of the asset. Dionne et al. (2010) find that informed buyers, buyers with at least 5% of the shares before the announcement, pay a significantly lower premium compared to buyers that do not have privileged information. This indicates that better informed buyers have the ability to better value their target, or at least have better bargaining power.
Another explanation of information asymmetry affecting the choice of payment in mergers and acquisitions could be related to the pecking-order theory and the market-timing theory. Managers will try to “time the market” by issuing equity when the share price is overvalued (Baker & Wurgler, 2002; Dong et al., 2006). However, due to information asymmetry, the pecking order theory argues that an equity issue can be considered as a signal that the issuing firm is overvalued (Myers & Majluf, 1984). This could explain the negative announcement returns for acquirers that initiated a stock acquisitions (Andrade et al., 2001; Malmendier et al., 2016). However, it is hard to implement the effect of the target revaluation using these theories. Receiving a bid from an overvalued acquirer should not affect the targets revaluation share price if the bid fails (Malmendier et al., 2016).
Andrade et al. (2001) provide extensive research on acquirer and target announcement re-turns in mergers and acquisitions for the period in 1973 to 1998. They investigate announce-ment returns for acquirer and targets and difference in stock and cash announceannounce-ment returns. They find a short window abnormal return for the target of 13% in stock acquisitions and 20.1% announcement return in cash acquisitions. For acquirers the announcement returns are -1.5% and 0.4%. In the long run acquirers financing with stock face a -9.0%*** in an equal-weighted portfolio and a -4.3% in a value-weighted portfolio long run return and in cash acquisition a not significant -1.4% compared to an equal-weighted portfolio, and 3.6% compared to a value-weighted portfolio. This indicates a better performance (or price paid) by acquisitions financed with cash. It also indicates that financing a merger or acquisition with stock should give a negative signal of the value of the acquirer since the long performance will be negative.
These findings support the hypothesis that when mergers fail, bidding firms that offered stock should experience a positive abnormal return. In contrast, firms that offered cash should experience a negative abnormal return when the merger is withdrawn. Chang & Suk (1998) find evidence for this theory. They observed the acquirer stock return upon the announcement of withdrawn takeovers. On average they find a positive abnormal return when the acquirer offered stock, and a negative abnormal return when the acquirer offered cash. Chang & Suk (1998) argue that the causal relationship of this observation lies in the asymmetric information hypothesis. The decision to not issue common stock conveys favourable information to the market. If the management thinks their stock is undervalued they won’t go through with the stock merger. The target firms suffer from a significant decline in stock price. Obviously, this effect is due to the premium that shareholders would have earned if the acquisition would go through. However, if we compare pre-announcement price versus the price after the failed merger we observe a revaluation. Which implies that the market observes a signal about the underlying value of the firm with the mergers and acquisition activity.
2.3
The revaluation effect
When the termination of a merger or acquisition is announced the market again perceives a signal. If there are no rumours of a subsequent bid, the target will observe a negative shock in its share price. Since no actual synergies can be obtained the expectation would be that the price will return to its original pre-announcement value (Bradley et al., 1983). However, prices heavily fluctuate after the withdrawal. Resulting in a far more negative returns relative to the pre-announcement price but also far more positive returns relative to the pre-announcement price. This “revaluation effect” has different underlying factors that might explain the direction of the price adjustment.
Malmendier et al. (2016) investigate this revaluation effect. They observe a significant difference between failed cash and stock bids. A failed cash offer has a revaluation effect of 15% while the stock offer returns to the pre-announcement value. So the signal of either a cash or a stock offer are perceived as different. As is explained in the literature review, one of the reasons could be that the choice for cash or stock might be affected by the valuation of acquirer and target. A cash acquisition might be a signal for general undervaluation, resulting in a positive revaluation. A stock acquisition might be a signal for acquirer overvaluation (Shleifer & Vishny, 2003), which has nothing to do with the targets price. Malmendier et al. (2016) do
not find that this revaluation effect is caused by future takeovers nor by changes in operational performances.
Ben-David et al. (2015) find that firms that are considered to be overvalued are more likely to use stock-for-stock mergers. This could drive the underperformance of stock acquirers and the over performance of cash acquirers. However, keep in mind that these results are exam-ined by mergers that go through. Implying that this over- or underperformance is explaexam-ined by the valuation of the bidder instead of the target. Another view is considered by Gorbenko & Malenko (2014). In their research they identify two types of bidders, strategic bidders and financial bidders. They show a higher valuation of the target by strategic bidders compared to financial bidders. Naturally, one could argue that this occurs because of synergies of strategic bidders. However, in a large part of their subsample (22.4%), the financial bidder offers a higher premium. This can be explained because financial bidders have access to debt at a lower cost than strategic bidders (Gorbenko & Malenko, 2014). Implying that in worse economic condi-tions financial bidders profit from the low cost of debt and low valuation of the companies. Liu (2017) researches the revaluation effect of withdrawn LBO’s by private equity. The research shows the same revaluation effect as the cash offer out of Malmendier et al. (2016). This con-tributes to the theory that Private Equity can ‘cherry-pick’ their targets. Another interesting finding is that the revaluation effect is stronger for firms with more information asymmetry. This implies that due to information asymmetry a company can become more undervalued. Or that the signal of receiving a takeover bid is stronger for firms with higher information asym-metry. This also raises the question whether opaque targets are more likely to be taken over with cash or stock.
2.4
Hypotheses development
So far the research to the revaluation effect has let to the discovery of the different revaluation between cash and stock acquisitions. The effect of information asymmetry has yet to be investi-gated. The first part of this empirical study investigates if the choice in cash or stock is affected by the level of information asymmetry. Based on the theory of Hansen (1987), I expect that if a firm has a lot of information asymmetry it will use stock for the acquisition because this will reduce the “overpayment cost”. Therefore I state the following hypothesis:
Hypothesis 1 Firms with more information asymmetry have a higher chance to receive a stock offer than a cash offer.
I also expect a negative relation between acquirer market-to-book ratio and cash as method of payment. When an acquirer is overvalued he will be more tempted to do a stock-for-stock acquisition in favour of the current shareholders (Shleifer & Vishny, 2003; Rhodes-Kropf et al., 2005; Martin, 1996).
Subsequently, as is stated by Liu (2017), the revaluation effect for opaque targets is higher than for transparent targets in leveraged buyouts. Opaque targets are naturally more vulnera-ble to misvaluation and therefore have a higher chance to become undervalued or overvalued. This could also be because of the information asymmetry the weight of the signal of being a potential take-over target is stronger. I expect that this rationale will hold for other mergers and acquisitions. So given that there could be more misvaluation due to the information symmetry, I expect that the revaluation effect is stronger. This results in the following hypothesis:
Hypothesis 2 The revaluation effect of opaque targets is stronger than for transparent targets. On the other hand, opaqueness could lead to uncertainty among investors. The best refer-ence point to investors could be the price before the announcement. Therefore, it could also be the case that the revaluation effect is weaker for opaque targets.
3
Data and Methodology
The aim of this paper is to investigate if the revaluation effect is larger for targets that have a higher level of information asymmetry. First, this paper follows the paper of Malmendier et al. (2016) by identifying the revaluation effect of failed cash and stock acquisitions. Secondly, this paper investigates if the choice in cash or stock depends on the level of information asymmetry of the target firm. Subsequently, the differences in revaluation between opaque and transparent targets are investigated. Finally, while controlling for other variables, the effect of information asymmetry on the revaluation effect is shown. In this section the data collection and empirical strategy are discussed.
3.1
Data
I start the data collection by gathering information about successful and unsuccessful deals. In-formation about mergers and acquisition bids is retrieved from the Thomson One SDC database. I download all completed and withdrawn bids data from 1980 to 2015, and excluded bids with
a deal value smaller than $1 million. From the CRSP/Compustat merged database I retrieved information about the fundamentals of the companies. Subsequently, the I/B/E/S database is used to obtain information about forecasted earnings per share, actual earnings per share, fore-cast dispersion and analyst coverage. For the withdrawn deals stock returns the CRSP event study database is used. I calculate the abnormal returns (CAR) as
CAR = t X
j=1
(rij − rmj) (1)
where rij and rmj denote firms i’s equity return and the CRSP value-weighted market return at time j, respectively. The same methodology is used by Liu (2017) and Malmendier et al. (2016). For each event the time frame consist of 25 days pre-announcement to 25 days after the withdrawal. This timeframe comes from the run-up theory of Schwert (1996).
The databases use different company identifiers. To make the files ready to merge I retrieved all CUSIP, NCUSIP, PERMCO and PERMNO numbers from the CRSP daily database. From the sample of Thomson One I merged the Thomson One 6-digit CUSIP on the NCUSIP and assign the PERMCO’s to the companies. From there the PERMCO and PERMNO numbers are used to merge with the datasets from Compustat, CRSP and I/B/E/S. One of the problems is that the databases do not cover the same firms. While the Thomson One SDC database covers deals world wide, CRSP/Compustat only covers firms listed in the US and Canada. Especially the I/B/E/S database misses a lot of information about the firms in the deals. This limits the database to a small amount of deals as is shown in Chapter 4. Finally, for the full sample selection I filter the dataset with the following rules. The deal value should be bigger than 1 million dollars, and percentage sought bigger than 50%. Furthermore, I require information from I/B/E/S, CRSP and Compustat. I also remove targets that received subsequent bids that occur in the next year after the withdrawal. Also, since the findings of Liu (2017) show a certain revaluation effect of leveraged buyouts, I remove leveraged buyouts to exclude the effect of private equity.
3.2
Methodology
3.2.1 Model description
First, this paper examines the choice in method of payment for a M&A deal. Liu (2017) find that private equity “cherry pick” their targets and that a higher revaluation occurs when information asymmetry is high. This raises the question of other companies do also select opaque targets for their takeovers. Another point of view could be that because the level of information asymmetry is high, a firm will use stock to minimize the expected “overpayment cost” (Eckbo et al., 1990). It is important to know whether the choice of cash and stock is affected by the information asymmetry. If in fact the level of information asymmetry matters, the findings of Malmendier et al. (2016) could suffer from omitted variable bias. This because leaving the information asymmetry of the target out gives a bias to the magnitude of the method of payment. For example, assuming that the revaluation is lower for opaque targets, if stock is used as method of payment to acquire opaque targets the effect of using stock is biased. For this research the following probit regression is used:
P (Cash) = β0 + β1Inf ormationAsymmetry + β2Of f erpremium
+ β3Hostile + β4qtarget + β5qacquirer (2)
The probit includes the three measures of information asymmetry and control variables. As proxy’s for information asymmetry, I use the absolute difference between the forecasted earnings per share and actual earnings per share divided by the share price (Forecast error), the standard deviation of analyst forecasted earnings per share divided by share price (Analyst dispersion) and the number of analyst that cover the target firm divided by total assets (Analyst coverage). These measures are also used by Liu (2017); Duchin et al. (2010); Krishnaswami & Subramaniam (1999) and He & Tian (2013). Note that the studies use different variables to normalise the variables. For example, Liu (2017) does not normalise the number of analyst, while Duchin et al. (2010) normalized the number of analyst by the book value of assets because the size is correlated with their dependent variable. In my thesis, I choose to normalize the difference in actual and forecasted EPS and analyst forecast dispersion by the share price, and to normalise the number of analyst covering the firm by the book value of assets as is used in Duchin et al. (2010). The proxy’s are measured one quarter before the announcement of the deal. With these measures the opaque targets versus the transparent targets are divided into
two sub-samples. divided by stock price I included several control variables. The first one is Offer Premium. The offer premium is on average higher for cash than stock deals as is shown in Table III and therefore should be included as control variable. The second and third control variables are q target and q acquirer, measured as market-to-book ratios. Rhodes-Kropf et al. (2005); Dong et al. (2006) find that acquirers with a high market-to-book value are more likely to use stock in the take-over. Firms that have lower valuation are more likely to receive cash bids. Furthermore, the hostile deal dummy is included to see whether hostile deals occur more often with cash or stock acquisitions.
Secondly, differences in revaluation between opaque firms and transparent firms are inves-tigated. The suggested approach comes from a combination of Liu (2017) and Malmendier et al. (2016). As revaluation measure, I use the cumulative abnormal return between 25 days pre-announcement and 25 days after withdrawal, as is shown in Equation 1. This is used by Liu (2017) and Malmendier et al. (2016) and comes from the run-up theory of Schwert (1996). Since each number of days between announcement date and withdrawn date is different, I normalised the time between these dates to a relative time-window, namely the time to com-pletion in percentage. To construct this relative time-window I use a linear approximation. I divide the days passed by the total of days in the window per company. I construct this relative time window in percentages and round them of to 2 decimals. To make approximations of the return between these relative time windows I make use of linear interpolation between the actual trading days, i.e.,
CARi(tg) = (1 − wi,tr)CARi(btRTic) + w(i,tr)CARi(btRTic + 1) (3)
where bxc refers to the floor function and w(i,tr) = tRTi−btRTic . The cumulative abnormal
return (CAR) over time of stock and cash can be found in Figure I. Subsequently, I divide the sample between cash and stock and search whether within cash and stock offers the level of information asymmetry has an effect on the revaluation. Targets are considered to be opaque if they are in the above the median of the |Consensus forecasted EPS − actual EPS| / Stock price. For the Analyst dispersion targets are considered to be opaque if they are above the sample median. For the Analyst coverage the targets are considered to be opaque if they are below the sample median.
The method of Table V obviously suffers from omitted variable bias. As Rhodes-Kropf et al. (2005); Dong et al. (2006); Malmendier et al. (2016) show the revaluation effect is affected
by the market-to-book ratios. In Equation 4, I control for these characteristics. I also control for size and offer premium by using the Transaction Size and Offer premium.
Revaluation =β0+ β1Inf ormationAsymmetry + β2Of f erpremium + β3qtarget
+ β4qacquirerβ5T ransactionSize (4)
Here the sample is divided in a pure cash and stock sample. The results are shown in Table VI. Finally, in Table VII the results are combined and added with a Cash dummy to see what the difference between cash and stock opaque targets are. Here the following regression is used:
Revaluation =β0+ β1Inf ormationAsymmetry ∗ Cash + β2Inf ormationAsymmetry + β3Of f erpremium + β4qtarget + β5qacquirer + β6T ransactionSize
(5) 3.2.2 Exogenously withdrawn sample
The main issue by examining failed deals is that the reason for the withdrawal might be en-dogenous. If, for example, information about the target comes out which affects the price of the target the revaluation effect would be lower while it is not caused by the difference in cash and stock. Nor is the amount of information asymmetry of importance.
To address this problem, exogenous reasons for withdrawal are used. This paper makes use of exogenous withdrawal reasons which are used in previous papers (Chang & Suk, 1998; Malmendier et al., 2016; Liu, 2017). The failure reasons are based on detailed news search in LexisNexis and the deal synopsis of the Thomson One SDC database. In Table I the main categories are displayed. The reasons for withdrawal are identified as “High Endogenous”, “Semi Endogenous” and “Low Endogenous”.
The “High Endogenous” sample contains failed deals with reasons where a lot of infor-mation about the target affects the price. “Target News” refers to failed deals associated with public news about the target, for example when Cobra Electronics Corporation withdrew its offer for Lowrance Electronics Inc..1 “Market Problems” refers to deals associated with
lems in the industry. The downfall in the industry affects the price of both companies and is therefore considered endogenous. “Outbid” refers to deals that are withdrawn because the target is taken-over by someone else. “Withdrawn reason not specified” refers to deals where there is no reliable information about the deal.
The “Semi Endogenous” reasons consist of reasons where endogeneity is less, but still of an issue. “Alliance” refers to deals where the acquirer and target made a pact. For example when Avalon Corporation withdrew its offer on Maynard Oil Company.2 “Shareholder rejection” indicates rejection by shareholders by for example, an insufficient number of shares being tendered or shareholders that vote against the offer in proxy filings.3 “Management rejection” refers to deals where the board insists shareholders to block the deal or engage in anti-takeover provisions.4 “Price too low” is when both parties could not agree on the price, or when I was unable to identify if the targets shareholder or management rejected the offer.
The “Low Endogenous” sample consist of deals where endogeneity is considered as low or non-existent. “Bidder Problems” refers to deals where the acquirer withdrew the deal because of problems internally. “Management Terms” refers to negotiations where the board of the acquirer and target could not come to an agreement about the management.5 “Lack of Finance” refers to deals that couldn’t be completed because of the lack of financing. For example when the Wickes Companies tried to acquire Lear Siegler and couldn’t obtain the required financing from the bank,6 “Regulator” refers to deals where the government or regulators intervention prevented the deal from going through. For example in a deal between Compuware and Viasoft7 and between Omnicare Inc. and PharMerica8.
The number of cases and distribution looks familiar to Malmendier et al. (2016). Although, I left out the category of Target News (private) since I was not able to find that information. This explains why my “Withdrawn reason not specified” is somewhat bigger (39% instead of 36% in Malmendier et al. (2016). My subsample consist of 103 “Exogenous” withdrawn firms. In comparison the sample of Malmendier et al. (2016) consist of 81.
2http://academic.lexisnexis.eu/??lni=3S8G-8XJ0-0007-J1GB&csi=237924&oc=00240&perma=true 3http://academic.lexisnexis.eu/??lni=7W7T-N4X1-2PMW-W00Y&csi=237924&oc=00240&perma=true 4http://academic.lexisnexis.eu/??lni=4KWY-BJS0-TXJ2-N2XM&csi=237924&oc=00240&perma=true 5http://academic.lexisnexis.eu/?lni=3RH0-B6P0-0001-13K9&csi=237924&oc=00240&perma=true 6http://academic.lexisnexis.eu/??lni=3S8G-8F30-0007-H2DD&csi=237924&oc=00240&perma=true 7http://academic.lexisnexis.eu/??lni=45G0-VCX0-014S-Y3KJ&csi=237924&oc=00240&perma=true 8http://academic.lexisnexis.eu/?lni=555Y-5TG1-DY7P-R1B7&csi=237924&oc=00240&perma=true
Table I
Reason for Withdrawal
Table I presents the reason for the withdrawal of the deal. Extensive news research has been done via Lexis-Nexis. The search includes news articles around the withdrawal date of the deal. The reasons specified in bold and italic are considered to be semi endogenous and low endogenous. Together they represent the ’exogenous’ withdrawn sample. The methodology is consistent with the papers Chang & Suk (1998); Liu (2017) and Malmendier et al. (2016) who use identical reasons for similar identification strategies.
Failure reason Cases Percentage
High Endogenous
Withdrawn reason not specified 115 39%
Target News 18 6% Market Problems 8 3% Outbid 53 18% Semi Endogenous Alliance 2 1% Shareholder rejection 17 6% Management rejection 20 7%
Price too low 4 1%
Low Endogenous
Bidder Problems 2 1%
Management Terms 7 2%
Lack of Finance 7 2%
Regulator 25 8%
Other than bid price, target news, or target performance 19 6%
Bids in ”Exogenous” withdrawn sample 103 35%
Bids considered as Endogenous 194 65%
4
Descriptive statistics
In this chapter the data is reviewed. Following the data gathering instructions, which can be found in Chapter 3, the successful dataset consist of 2164 cash or stock deals. The unsuccessful sample consist of 297 pure cash and stock deals. Of the withdrawn deals 222 are cash deals and 75 are stock deals. One problem is that not all databases cover the same firms. Especially the information of the I/B/E/S database is limited. Together with the criteria of the exogenous deal selection this limits the data significantly. Krishnaswami & Subramaniam (1999) have had the same problem. In comparison, Malmendier et al. (2016) pure cash and stock sample
consist of 236 deals, retrieved in the time frame of 1980 to 2008. Whereas the data set used in this paper consist of 297 deals in the time frame 1980 to 2015. The exogenous withdrawn sample in Malmendier et al. (2016) consist of 81 deals and the data set here consist of 103 deals. However, not all deals have information on the I/B/E/S database. This limits the number of deals to 79 in Table V, and to 66 in Table VII.
4.1
Sample description
In Table II the summary statistics of the main dataset are displayed. The successful deals are filtered to select deals with information out of all databases. The amount of observations for the unsuccessful deals vary by each variable. This depends on which database is used to retrieve the data. In the right column the p-value of the difference is given. The first three variables contain the measures of information asymmetry. Related literature uses the same measures of information asymmetry (Duchin et al., 2010; He & Tian, 2013; Krishnaswami & Subramaniam, 1999). Furthermore, the percentage of cash and stock deals are displayed. There are more stock acquisitions than cash acquisitions in the successful deal sample (30% and 60% respectively) . However, in the withdrawn sample there are more cash offers (25% to 75%). Also the percentage of hostile deals in the withdrawn sample is significantly higher than the percentage of the successful deal sample. Malmendier et al. (2016) observe the same and this could due the fact that hostile deals have a lower rate of completion than other bids (Ngo & Susnjara, 2016). Another interesting fact, discovered both by this paper and Malmendier et al. (2016) is that the offer premiums of successful deals are not necessarily higher than those of unsuccessful deals. The mean and median of Transaction size and Target size are skewed towards smaller enterprises in both successful deals as in unsuccessful deals. Also due to the presence of large outliers, the mean of both q of target and q of acquirer are higher for successful deals than for unsuccessful deals. However, the medians are comparable. To control for the oultiers I winsorised from zero to the 95th percentile before running the regressions. The median of q of the target is higher compared to the acquirer. This seems to support the theory that low market to book acquirers acquire high market to book targets, as is shown by Rhodes-Kropf et al. (2005).
The dataset of the unsuccessful deals is unbalanced. The observations of the measures of information asymmetry are about two-thirds of the total observations. This significantly drops the amount of deals which can be researched. This causes some problems with further analysis
since the low amount of observations drops significance and reliability of the data in Table V, Table VI and Table VII.
In Table III the differences between cash and stock are displayed. The Offer premium for cash offers is higher than for stock offers. The q of the target in cash offers is a little bit higher than for stock offers. The q for acquirers for stock offers is bigger than cash offers. However, these differences do not seem to be significant in the exogenous withdrawn sample. This is due to the low amount of observations. For the larger sample the differences are significant.
Table II
Summary statistics: Successful Deals and Unsuccesfull deals
Successful Deals Unsuccessful Deals
Variable Mean Median Std.dev Min Max Obs. Mean Median Std.dev Min Max Obs. p-value
Forecast error 0.03 0.00 0.17 0.00 3.92 2164 0.14 0.01 1.06 0.00 15.00 215 .00 Analyst dispersion 0.00 0.00 0.00 0.00 0.16 2164 0.00 0.00 0.01 0.00 0.06 209 .091 Analyst coverage 5.81 4.00 5.78 1.00 40.00 2164 6.44 4.00 6.29 1.00 39.00 216 .130 Cash 0.40 0.00 0.49 0.00 1.00 2164 0.75 1.00 0.43 0.00 1.00 297 .000 Stock 0.60 1.00 0.49 0.00 1.00 2164 0.25 0.00 0.43 0.00 1.00 297 .000 Time to failure 133.50 113.50 90.38 11.00 906.00 2164 127 92 125 6 774 297 .238 Offer premium 0.41 0.33 0.33 0.00 2.05 2164 0.38 0.27 0.42 0.00 3.27 295 .835 Hostile 0.01 0.00 0.11 0.00 1.00 2164 0.34 0.00 0.47 0.00 1.00 297 .000 Transaction size 3.10 0.31 15.35 0.00 520.39 2164 2.80 0.14 23.36 0.00 385.95 280 .777 Target size 2.29 0.23 10.60 0.00 336.13 2164 2.57 0.11 23.80 0.00 405.54 295 .730 q of target 3.49 1.23 6.04 0.07 36.43 2164 1.56 0.99 1.80 0.19 9.78 279 .000 q of acquirer 1.30 0.89 1.89 0.04 14.16 2164 1.64 0.89 2.65 0.15 13.58 246 .000
Note: Table II summarizes the statistics for the successful deals and unsuccessful deals. The sample consist of 2164 successful pure cash and stock deals and 297 withdrawn stock deals reported by the Thomson One SDC database in the time-frame of 1985 to 2015. The variables Cash, Stock, Offer premium and Hostile are measured as percentages. Transaction size is measured in $billion and Time to failure is in days. CRSP/Compustat is used to compute the Target size (measured in $billion), the q of targetand the q of acquirer which are market-to-book ratios for the target and the acquirer respectively. The I/B/E/S database is used to obtain the measures for information asymmetry. Both Forecast error and Analyst Dispersion are normalized by the share price. Note that the Analyst Coverage measured as the number of analyst covering the firm is not yet divided by the book value of assets. This is required and done for the further data analysis. In the last column p-values are given for the differences between successful and unsuccessful deal statistics.
Table III
Summary statistics: Exogenously withdrawn Cash and Stock bids
Cash Offers Stock Offers
Variable Mean Median Std.dev Min Max Obs. Mean Median Std.dev Min Max Obs.
p-value Forecast error 0.32 0.00 2.12 0.00 15.00 50 0.16 0.00 0.65 0.00 3.69 32 .669 Forecast Dispersion 0.00 0.00 0.01 0.00 0.05 49 0.00 0.00 0.01 0.00 0.06 31 .622 Analyst Coverage 7.44 6.50 5.73 1.00 24.00 50 6.78 5.00 5.53 1.00 21.00 32 .608 Time to failure 118.79 69.00 127.61 6.00 737.00 63 187.78 147.00 147.16 13.00 737.00 40 .013 Offer premium 0.41 0.35 0.27 0.00 1.57 63 0.33 0.22 0.33 0.01 1.58 40 .185 Hostile 0.68 1.00 0.47 0.00 1.00 63 0.63 1.00 0.49 0.00 1.00 40 .552 Transaction size 1.94 0.16 3.93 0.00 23.08 63 3.03 0.21 8.02 0.00 44.30 38 .362 Target size 1.41 0.11 2.78 0.00 14.66 63 2.75 0.16 7.09 0.00 39.02 40 .183 q of target 2.04 1.14 2.26 0.29 9.58 62 1.52 1.02 1.73 0.23 9.78 38 .234 q of acquirer 1.54 0.91 2.23 0.20 13.41 61 1.65 0.86 2.74 0.40 13.58 35 .836
Note: Table III summarizes the statistics for the Exogenous withdrawn sample. The sample consist of 63 withdrawn cash deals and 40 withdrawn stock deals reported by the Thomson One SDC database in the time-frame of 1985 to 2015. The variables Cash, Stock, Offer premium and Hostile are measured as percentages. Transaction sizeis measured in $billion and Time to failure is in days. CRSP/Compustat is used to compute the Target size (measured in $billion), the q of target and the q of acquirer which are market-to-book ratios for the target and the acquirer respectively. The I/B/E/S database is used to obtain the measures for information asymmetry. Both Forecast error and Analyst Dispersion are normalized by the share price. Note that the Analyst Coverage measured as the number of analyst covering the firm is not divided by the book value of assets. This is required and done for the further data analysis. In the last column p-values are given for the differences between cash and stock bids are displayed.
4.2
Revaluation cash versus stock
The effect of the announcement, negotiation and withdrawal of the pure cash and stock bids are displayed in Figure I. The figure shows the cumulative abnormal return starting from 25 days before the announcement of the bid to 25 days after the withdrawal of the bid. The cash targets receive a higher announcement return than stock targets. This is mainly caused by a higher offer premium. After the withdrawal the target has a revaluation of around 10%.
The stock targets have a lower announcement premium. This is mainly caused by a lower premium offered by the acquirer. Between the announcement and withdrawal of the bid stock targets already face a revaluation back to the original price. After the withdrawal the stock targets face a negative return of around -4% The cash and stock acquirers face negative abnor-mal returns during the whole period. This results are in line with Malmendier et al. (2016). Although, the stock acquirers observed by Malmendier et al. (2016) have a revaluation of -20% and in this paper around they observe a revaluation of -10%.
The figure shows that cash targets have a higher revaluation effect than stock targets. The casual explanation lies in Rhodes-Kropf et al. (2005); Shleifer & Vishny (2003) and Dong et al. (2006). Firms that have overvalued equity try to do acquisitions with stock. Therefore the reason for the acquisition does not have to lie in an undervaluation of the target, but more in the acquirer making an advantage of their overvalued stock (Rhodes-Kropf et al., 2005). Dong et al. (2006) also argue that cash acquisitions are done when the market is undervalued. Therefore a cash bid can be perceived as a signal that the target is undervalued, which could cause the higher revaluation effect.
Note that in a non-controlled environment there could be a chance of future take-over that could be reflected in the revaluation price. However, Malmendier et al. (2016) controlled for this factor and showed that this is not the case. In this paper the sample selection consist of withdrawn deals where the target did not received a takeover bid for the next two years. This paper further investigates where the difference between cash and stock revaluation comes from. It also studies if the revaluation whitin a certain method of payment differs by the level of information asymmetry.
Figure I CAR over time
Figure I displays the cumulative abnormal return for the announcement of the bid and failure of the deal. The figure runs from 25 days before announcement of the bid (B) to 25 days after the failure of the bid (F). Between the bid and failure are the relative amount of trading days calculated as in Equation 3. The sample consist of the exogenous withdrawn pure-cash and pure-stock deals.
5
Results
In this section the results of the research are displayed. First the choice for stock or cash is examined. In Table IV the probit on the choice for cash and stock is displayed. After the examination of the effect of information asymmetry on the method of payment, the difference in revaluation between opaque and transparent targets are displayed, which can be found in Table V. Finally, to control for other variables that may affect the revaluation, a regression including the effect of information asymmetry on the revaluation is shown in Table VI and Table VII.
5.1
Information asymmetry on the choice for cash or stock
First, the effect of information asymmetry on the choice for cash and stock is discussed. For this analysis the probit regression, found in Equation 2, is used. The result is shown in Table IV. In regression one to six the full sample, consisting of both successful as unsuccessful bids, is used to examine the choice for cash and stock. In regression 7 to 12 the withdrawn sample is used.
The full sample shows a negative significant effect for Analyst dispersion on the choice for cash, implying a choice for stock when information asymmetry is high. This is in favour of hypothesis 1. The Analyst coverage has a positive and significant effect on the choice for cash, which implies a choice for cash when information asymmetry is low. This is also in favour of the hypothesis of choosing stock over cash when the target information asymmetry is high. Forecast error is not significant but by controlling for other factors in regression 4, a correlation between Forecast error and the choice for cash that is in line with the hypothesis is shown. Regarding the control variables, a higher Tobin’s q for the acquirer has a significant negative effect on the probability to receive a cash bid. This is in line with the findings of Martin (1996) which show that acquirers with a high market-to-book value are more likely to use stock in the take-over.
Also the Offer premium is higher for cash offers. Therefore, for the revaluation effect there should be controlled for the offer premium. Furthermore, q of the target is not significant. Meaning that the market-to-book ratio of the target, a proxy for level of growth opportunities, does not seem to affect the choice in cash or stock. Also, hostile deals are more likely to be cash deals, consistent with findings of Malmendier et al. (2016) and Chang & Suk (1998). The withdrawn sample shows that transparent targets are more likely to be taken over with cash. However, only one of the three measures is significant. Which is tend to be too weak to confirm hypothesis 1. Note that the amount of observations drops significantly in the withdrawn sample and could be the cause of the insignificant results. Therefore the findings of the withdrawn sample are less reliable than for the full sample. A causal relationship may be explained by the failure of the bid. The failure might be caused by the firms not being able to overcome the information asymmetry issues. However, this research does not provide clear evidence for this matter.
Examining these results the evidence tends to confirm hypothesis 1. Two of the three proxy’s for information asymmetry in the full sample show a significant effect of
informa-tion asymmetry affecting the choice for method of payment. Therefore it can be confirmed that the effect of information asymmetry on choice of payment could be caused by the level of information asymmetry which is in line with theories of Hansen (1987); Fishman (1989) and Eckbo et al. (1990). Economically, the reason for these findings can be explained because due to the higher level of information asymmetry it is harder to measure how the synergies can be obtained. By using stock the acquirer reduces the risk of overpayment (Hansen, 1987).
Table IV
Information asymmetry and choice of payment
Dependent Variable = Revaluation CAR
Full sample Withdrawn sample
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Forecast error 0.01 -0.03 -0.09 -0.05 (0.052) (0.056) (0.184) (0.269) Analyst dispersion -0.21*** -0.16*** -0.09 -0.11 (0.052) (0.055) (0.184) (0.252) Analyst coverage 0.28*** 0.41*** 0.54*** 0.65** (0.052) (0.058) (0.192) (0.319) Offer premium 0.21** 0.22*** 0.12 1.14** 1.11** 0.96** (0.083) (0.084) (0.085) (0.506) (0.495) (0.487) Hostile 0.62*** 0.56*** 0.67*** 0.15 0.15 0.05 (0.140) (0.141) (0.139) (0.270) (0.270) (0.284) q target 0.00 0.00 0.00 0.07 0.06 0.01 (0.001) (0.001) (0.002) (0.054) (0.054) (0.057) q acquirer -0.02*** -0.02*** -0.02*** -0.07** -0.07** -0.08*** (0.004) (0.004) (0.005) (0.030) (0.030) (0.031) Transaction Size -0.01 0.00 0.01 -0.12* -0.12* -0.03 (0.013) (0.013) (0.013) (0.069) (0.065) (0.072) Intercept -0.17*** -0.06* -0.31*** -0.25*** -0.18*** -0.39*** 0.68*** 0.59*** 0.40*** -0.04 0.01 -0.06 (0.037) (0.036) (0.037) (0.050) (0.053) (0.049) (0.131) (0.131) (0.126) (0.303) (0.316) (0.286) Observations 2379 2372 2373 2277 2277 2277 217 217 210 118 118 118 Pseudo R-squared 0.0000 0.0052 0.0093 0.0220 0.0249 0.0384 0.0010 0.0017 0.0342 0.0961 0.0960 0.1247
Note: The dependent variable is the Revaluation CAR. Forecast error is dummy, and proxy for information asymmetry, that is 1 if the firm is above sample median. Analyst dispersionis a dummy, and proxy for information asymmetry that equals 1 if the firm is above the sample median. Analyst coverage is also a dummy, and proxy for information asymmetry and equals 1 if the firm is above the sample median. A firm is considered opaque if EPS forecast-actual equals 1, if Analyst dispersion equals 1. and if Analyst coverage equals 0. The other variables are control variables. *, ** and *** indicate significance at 10%, 5% and 1%, respectively.
5.2
Opaque versus transparent targets
In Figure I it is shown that cash acquisition targets have a higher revaluation than stock acqui-sition targets. This is in line with Malmendier et al. (2016) and confirms theories of Shleifer & Vishny (2003); Rhodes-Kropf et al. (2005); Bhagat et al. (2005). However, the effect of information asymmetry is yet to be investigated. As is stated in the hypothesis development, opaque targets are expected to have a higher revaluation. In Table V the differences between opaque and transparent targets are displayed.
As is explained in Chapter 3, the sample is divided into two subsamples by each measure of information asymmetry. Then, the difference in cumulative abnormal return during the an-nouncement and withdrawal of the bid between opaque and transparent targets is investigated. The results show that cash targets have a higher revaluation effect than stock targets. The cash targets’ revaluation is significant from zero. In two of the three measures transparent cash tar-gets seem to have a higher revaluation in regard to opaque tartar-gets. However, none of these results is significant and therefore the results are too weak to state a difference in revaluation between opaque cash and transparent cash targets.
The stock targets revaluation is not significant from zero. In contrast to the cash bids, the three measures of information asymmetry show a higher revaluation for opaque stock targets but only one of the measures is significant. The results of the analyst dispersion show a signif-icantly higher return for opaque stock targets than for transparent stock targets. This confirms hypothesis 2, a higher level of information asymmetry results in a higher revaluation of the stock.
However, overall it is hard to conclude any significant outcome. While one of the measures for information asymmetry does show a significant higher revaluation effect for opaque stock targets, the other ones do not show any significant difference. Therefore, it can be concluded that the results do not confirm hypothesis 2. Keep in mind that there are no control variables and this results could suffer from omitted variable bias. Also, the number of observations becomes very low, 48 and 31 cases for cash and stock bids respectively. This might affect the significance level of the outcomes.
Table V
Information asymmetry and target firms’ stock returns
Opaque targets Transparent targets Difference Panel 5A: Analyst forecast error
Analyst forecast error > sample median
Analyst forecast <= sample median Cash bids (48 cases)
Announcement CAR .348*** .256*** .092
Between CAR -.132*** -.042** -.090**
Withdrawn CAR -.074*** -.039*** -.035
Revaluation CAR .069* .153*** -.084
Stock bids (31 cases)
Announcement CAR .207*** .107*** .101
Between CAR -.085* -.076* -.009
Withdrawn CAR -.051* -.072** .021
Revaluation CAR -.017 -.059 .076
Panel 5B: Analyst forecast dispersion
Analyst dispersion > sample median
Analyst dispersion <= sample median Cash bids (48 cases)
Announcement CAR .332*** .271*** .061
Between CAR -.072*** -.101*** .029
Withdrawn CAR -.056*** -.057*** .001
Revaluation CAR .155*** .068** .087
Stock bids (31 cases)
Announcement CAR 0.136** .18*** -.044
Between CAR -.02 -.137*** .118
Withdrawn CAR -.011 -.108*** .097*
Revaluation CAR .077* -.111** .189**
Table V–Continued Panel 5C: Analyst coverage
Analyst coverage < sample median
Analyst coverage >= sample median Cash bids (48 cases)
Announcement CAR .279*** .334*** -.055
Between CAR -.084** -.097*** .013
Withdrawn CAR -.05*** -.064*** .014
Revaluation CAR .107*** .112*** -.005
Stock bids (31 cases)
Announcement CAR .169*** .111** .059
Between CAR -.097** -.048 -.049
Withdrawn CAR -.004 -.067** .028
Revaluation CAR .004 -.032 .036
Note: Table V reports the relationship between the level of information symmetry of the target and the abnor-mal stock returns running from 25 days before the announcement of the bid until 25 days after the withdrawal of the bid. The sample consist of 79 ”exogenously withdrawn” pure cash or stock bids. The table reports three measures for information asymmetry, obtained from I/B/E/S database. In Panel 5A, the |consensus forecasted EPS− actual EPS| / Stock price is used as measure of information asymmetry. In Panel 5B, the Standard deviation of analyst forecasted EPS/ Stock price is used. And in Panel 5C the Number of analyst covering the target firms/ Firm assets is used to divide the sample in opaque and transparent firms. All three information asymmetry measures are measured one quarter before the announcement of the bid. Announcement CAR runs from 25 days prior to the announcement until one day after. Between CAR runs from one day after the bid until one day before the withdrawal. Withdrawn CAR runs from one day prior to the withdrawal until 25 days after the withdrawal. Revaluation CAR runs from 25 days prior to the announcement until 25 days after the withdrawal. The last column shows the difference in CAR between opaque and transparent firms. Statistical significance at the 1%, 5%, and 10% level is indicated by ***, **, and *, respectively.
5.3
Information asymmetry on target revaluation
So far, only the difference in revaluation between opaque and transparent targets within cash or stock deals is investigated. However, other factors could affect the revaluation. Therefore a regression output is done to control for certain variables. In Table VI the regression output is displayed.
The results for cash offers show no significant outcome for the variables of interest. Mean-ing that the revaluation effect does not depend on whether the target is opaque or transparent. It is also interesting to see that the offer premium does not have a significant effect on the revaluation. The market-to-book ratio of the target has a negative effect on the revaluation. Meaning that if the growth opportunities where low, the revaluation effect is lower. However, if the acquirer has a high market-to-book value the revaluation effect is higher.
The results for stock offers are shown in regression 4, 5 and 6 in Table VI. The Forecast erroris positively significant at a 10% level. Implying that opaque stock targets have a higher revaluation effect than transparent targets. The Analyst dispersion is positive and significant. Again, this implies a higher revaluation for opaque stock targets. Analyst coverage is negatively correlated however not significant at all.
These results confirm hypothesis 2 for the stock bids. However, due to the low amount of observations the results have to be interpreted carefully. The significance and correlation of control variables seem to be in line with previous research. In regression 5 the offer premium becomes significant. This seems logical since a higher premium could, despite more synergies between the acquirer and target, imply a bigger undervaluation pre-announcement. The q of the target has a negative impact on the revaluation. Economically, this could be explained by that the target was already valued at a high price and therefore had no need of revaluation. The q of the acquirer has a positive impact on the revaluation. Implying that if the acquirer has a high market-to-book ratio the target observes a higher revaluation. This could be explained by that a high market-to-book ratio of a potential acquirer signals to the market that the target company potentially could have a market-to-book ratio similar to the acquirer. Therefore in the targets’ current state it could be undervalued and need a revaluation. Unfortunately the amount of observations becomes very small which again limits the significance and certainty of the results.
In Table VII the results of cash and stock offers combined are shown. The first proxy for information asymmetry is positive and significant. Implying that for opaque targets that receive stock offers the revaluation effect is higher than for transparent targets that receive stock offers. The cash component also has a positive effect on the revaluation. However, the combination of Cashand Forecast error is negative. This implies a lower revaluation for opaque cash targets in regard to transparent cash targets. The same holds for the second proxy, Analyst Dispersion, although, the differences in revaluation are considered to less apparent comparing the magni-tude of the coefficients. The third proxy, Analyst coverage is not significant. The results seems to partly confirm and partly contradict hypothesis 2. The level of information asymmetry does affect the results whitin cash or stock targets differently. Opaque stock targets receive a higher revaluation compared to transparent stock targets and opaque cash targets receive a lower reval-uation compared to transparent cash targets. Overall, the results of Malmendier et al. (2016) hold since cash targets have a significantly higher revaluation than stock targets. Regarding the
control variables, the market-to-book ratio for the target has a negative effect on the revaluation, and the market-to-book ratio of the acquirer a positive effect on the revaluation. This confirms the theories and findings of Malmendier et al. (2016). The causal relationship for this findings can be explained as follows. The theory by Hansen (1987) and Fishman (1989) predicts that stock is used for opaque targets. So whitin the opaque stock targets it is possible that investors see the signal of being target to a takeover as a positive signal of undervaluation of the target. Transparent stock targets, however, are assumed to be fairly valued. Therefore, the signal of being subject to a potential takeover does not affect the price.
For cash targets the transparent targets are perceived to be fairly valued, but because of the bid, investors receive the signal of an undervaluation of the market (Dong et al., 2006). How-ever, when the cash target is opaque the signal of being undervalued is not valued because the firm remains hard to value. Therefore, the best reference point is the price pre-announcement.
By comparing the whole cash and stock sample, I find that information asymmetry con-verges the revaluation effect. A causal explanation for this phenomenon is not yet investigated. Reasoning from Baker et al. (2012) the best reference point after the revaluation is the pre-announcement price. Therefore, for opaque targets the revaluation naturally is attracted to that point. However, for transparent targets this might not hold since there is less information asym-metry and therefore less magnitude of the reference point.
Table VI
Information Asymmetry on Revaluation
Dependent Variable = Revaluation CAR
Cash Offers Stock Offers
(1) (2) (3) (4) (5) (6) Forecast error -0.06 0.18* (0.101) (0.087) Analyst dispersion 0.09 0.28*** (0.076) (0.068) Analyst coverage 0.06 -0.06 (0.080) (0.139) Offer premium 0.10 0.08 0.05 0.22 0.33*** 0.23 (0.133) (0.138) (0.139) (0.126) (0.102) (0.163) q of target -0.04* -0.04* -0.05*** -0.06*** -0.05*** -0.06*** (0.019) (0.019) (0.016) (0.012) (0.009) (0.019) q of acquirer 0.03*** 0.04*** 0.03** 0.06*** 0.06*** 0.06*** (0.010) (0.010) (0.012) (0.010) (0.008) (0.019) Transaction Size 0.02 0.03* 0.04** -0.03 -0.03 -0.05 (0.023) (0.016) (0.016) (0.026) (0.026) (0.041) Intercept 0.21* -0.01 0.09 -0.33 -0.56*** -0.01 (0.125) (0.136) (0.124) (0.200) (0.133) (0.180) Observations 45 45 45 20 20 20 Adjusted R-squared 0.0863 0.1099 0.0854 0.4908 0.6595 0.3721
Note: Table VI shows the result of the proxy’s for information asymmetry on the revaluation of tar-get firms. The dependent variable is a the CAR of the stock price measured from 25 days before the announcement until 25 days after the withdrawal. the sample consist 55 withdrawn deals pure cash and stock deals. The EPS forecast − actual, Analyst dispersion and Analyst coverage are measured in the quarter before the deal announcement date. Standard errors are reported in parentheses. Statistical significance at the 1%, 5% and 10% level is indicated by ***, **, and *
Table VII
Information Asymmetry on Revaluation of the target
(1) (2) (3) Cash*Forecast error -0.29** (0.136) Cash*Analyst dispersion -0.19* (0.114) Cash*Analyst coverage -0.03 (0.132) EPS forecast-actual 0.19* (0.097) Analyst dispersion 0.29*** (0.087) Analyst coverage -0.05 (0.114) Cash 0.52** 0.39** 0.17 (0.199) (0.171) (0.213) Offer premium 0.14 0.18 0.13 (0.103) (0.118) (0.104) q of target -0.03* -0.03* -0.03* (0.016) (0.016) (0.018) q of acquirer 0.04*** 0.04*** 0.04*** (0.006) (0.005) (0.007) Transaction Size 0.01 0.02 0.01 (0.017) (0.014) (0.019) Intercept -0.29* -0.49*** 0.03 (0.169) (0.169) (0.182) Observations 66 66 66 Adjusted R-squared 0.1957 0.2589 0.1414
Note:Table VII shows the result of the proxy’s for information asymmetry on the revaluation of target firms. The dependent variable is a the CAR of the stock price measured from 25 days before the announcement until 25 days after the withdrawal. the sample consist 55 withdrawn deals pure cash and stock deals. The EPS forecast − actual, Analyst dispersion and Analyst coverage are measured in the quarter before the deal announcement date. Standard errors are reported in parentheses. Statistical significance at the 1%, 5% and 10% level is indicated by ***, **, and * respectively.
6
Conclusion
In summary, this thesis elaborates on the revaluation effect that exists following the failure of mergers and acquisitions. It is the first paper that investigates the relationship between information asymmetry and the revaluation effect in a broader mergers and acquisitions context. By using three measures of information asymmetry, I first test whether the choice of the method of payment for acquisitions depends on the level of information asymmetry. Additionally, I create an exogenous withdrawn sample to discover how information asymmetry affects the revaluation effect.
The results show that two of the three measures of information asymmetry confirm the relationship between information asymmetry and method of payment. While Forecast error does not show a significant impact, following Analyst dispersion as a proxy for information asymmetry, the results show that opaque targets are more likely to be acquired with stock as method of payment. The proxy Analyst coverage also indicates that transparent targets are more likely to be taken over with cash. This confirms theories presented by Eckbo et al. (1990); Hansen (1987) and Fishman (1989) which assert that stock offers can be used to reduce the problems arising by information asymmetry.
Furthermore, my paper confirms that the revaluation effect is higher for cash targets than for stock targets, which is in line with findings by Malmendier et al. (2016). By dividing the sample by the measures of information asymmetry into two subsamples, I investigate whether opaque or transparent targets receive a higher revaluation. For the cash bids no statistical differences are found. For the stock bids only the separation of the sample by analyst forecast dispersion shows a positive significant effect for opaque targets. I find this evidence is too weak to support my hypothesis of opaque targets receiving a higher revaluation. However, when controlled for other variables, such as offer premium and the Tobin’s q for the target and the acquirer, there seem to be a significant difference in revaluation. For opaque cash targets compared to transparent cash targets, the results show a higher revaluation for transparent cash targets. Regarding stock acquisitions, the revaluation is higher for opaque targets than for transparent targets. This shows a difference in the effect of information asymmetry on the target’s revaluation. Viewed from a broader perspective, opaqueness converges the difference between cash and stock revaluation.
This paper contributes to the existing literature by using information asymmetry in failed mergers and acquisitions. My findings support the findings of Malmendier et al. (2016); Liu (2017). The thesis also brings attention to the theories of the choice for cash or stock by
