Are large deals influenced by
overvaluation?
Author:
Joost Tanis
University of Groningen
Faculty of Economics and Business
MSc Business Administration - Finance
Supervisor:
Are large deals influenced by overvaluation?
ABSTRACT
Several recent theories predict a relationship between the level of overvaluation and acquisition size. They claim larger deals should be more motivated by the overvaluation of the acquirer's equity and should have especially poor synergies. This new theories contrast sharply with traditional neoclassical theories which assume mergers are a disciplining mechanism for inefficient firms and create wealth through synergies. This study investigates if and how misvaluation affected large deals during the dotcom merger wave. For this purpose a sample of 259 large US mergers between 1994 and 2004 is examined. The predictions of behavioural overvaluation theories are contrasted to those made by a neoclassical theory of mergers. This study investigates both short and long term effects and presents results from both operating and stock data. It finds strong evidence to support claims made by overvaluation based theories of mergers and find little evidence to support the predictions of a neoclas-sical theory of mergers. (Mis)valuation is found to be a significant variable in predicting post merger changes in valuation and method of payment choice. Low growth firms tend to acquire high growth firms and misvaluation is higher in acquirers than in targets. Furthermore evidence is found that large firms tend to overpay in equity bids.
Name: Joost Tanis
2. LITERATURE
2.4 Theories about merger waves
2.4.1 Overvaluation hypothesis (OVH)
2.4.2 Q-theory (QT)
2.2 Previous findings 2.6 Hypothesis development
3. DATA AND METHODOLOGY
3.1 Data
3.1.1 Sample selection criteria 3.1.2 Sample characteristics
3.1.3 Data sources and data construction 3.1.4 Measures of operating performance 3.1.5 Measures of valuation
3.1.6 The Residual Income Valuation Model
3.2 Methodology
3.2.1 Event study methodology 3.2.4 Regression analyses 4. RESULTS 4.1 Univariate results 5.2 Multivariate results 5.1 Conclusions 5.2 Limitations REFERENCES APPENDIX
1.
INTRODUCTION
“The market can stay irrational longer than you can stay solvent.” John Maynard Keynes (1883-1946)
As the above quote illustrates market rationality has not always been the prevailing view among economists. Yet for the past thirty years almost all of the research in corporate finance has assumed a broad rationality. Capital markets are supposed to be efficient, with prices fully reflecting all available information. Investors are sup-posed to act rational and have unbiased expectations about the future. Managers driven by self interest and res-ponding to carefully thought out compensation schemes and governance mechanisms are expected to enhance shareholder value by making optimal decisions about the firm and its future.
The 1995-2001 dotcom bubble and what many viewed as irrational behavior during that era made some econo-mists doubt these traditional assumptions and spurred a renewed interest in alternative explanations for corpo-rate behavior. As investors were valuing companies at prices that could not be attained if the whole world bought their products and managers were pursuing crazy plans like selling cat food over the internet. At the height of the IT-hype one internet company raised $12 million from its investors and spent $10 mln on a big party celebrating the launch of the company. This and similar irrational behavior casted doubt on the classical assumptions of market efficiency and investor rationality. Although the evidence of market mispricing is mount-ing (see Kothari et al. (2006)), most studies still assume that markets efficiently prices stocks.
This paper examines the area of corporate finance which lends itself well to empirical study and which is home to some of the hottest debate between the new behavioral explanations of corporate actions and the traditional neoclassical theories, the world of mergers and acquisitions (M&A).
The late nineties had one of the greatest M&A waves in known history (see figure 1). The mergers seemed espe-cially concentrated in sectors, like IT, which had experienced a surge in their stock prices. As figure 1 shows, there seems to be a connection between rising stock market valuation and the number of mergers. It should be noted here that the P/E ratio is a not the best proxy for (over)valuation as it is dependent on a company’s method of financing.
Sources: Mergers: 1895-1920 from Nelson (1959); 1921-67 from FTC; 1968-2002 from M&A. P/E ratios: Homepage of Robert Shiller: http://aida.econ.yale.edu/~shiller/data.htm. Population: Statistical Abstract of United States (several years).
!"
The traditional neoclassical motivation for mergers and acquisitions has always been that they increase share-holder wealth as a result of synergies between the two merging companies or increased market power for the acquirer. After all what rational manager would propose a wealth destroying deal if that meant that his share-holders would dump their stock at the earliest opportunity. Thereby decreasing the value of his option based pay scheme and a few months down the line, during the next shareholders meeting he would have to face angry questions from disgruntled investors and possibly job loss. Consequently as managers are rational and capital markets are reasonably efficient, most mergers should be wealth creating. The only thing that could explain a sharp increase in the number of mergers is some outside influence like an economic disturbance (e.g. lower con-sumption or industry wide shock (e.g., technological innovations or deregulation), which makes other business combinations more profitable and therefore causes a round of acquisitions and divestitures to adapt to the new circumstances
Yet many of the mergers during the IT boom didn’t seem to be motivated by any of such rational reasoning. A lot of deals were between companies which lacked solid business plans and did not have observable synergies as they operated in very different markets. The only observable item that these deals seemed to have in common was that one or both of the firms had hyped expectations and an inflated stock market valuation.
The most obvious deal to illustrate this irrationality is the massive deal between AOL (an internet service pro-vider) and Time Warner (a media and entertainment conglomerate) in 2000. In this deal, that was announced as a merger at the time, AOL in fact acquired the much larger Time Warner for $164 bn. The purchase was paid for in AOL stock which price had risen to enormous heights thanks to the rush on IT stocks. When the deal was announced there was off course much talk about synergies between the two companies. Time Warner would supply the content for AOL’s internet distribution channel. Although the deal was favorably received by the market at the time, even then some doubted the wisdom behind the transaction.
After the merger, the profitability of the Internet Service division (America Online) decreased.Meanwhile, the market valuation of similar independent internet companies drastically fell. As a result, the value of the America Online division dropped significantly. This forced a goodwill write-off, causing AOL Time Warner to report a loss of $99 billion in 2002 — at the time, the largest loss ever reported by a company. In 2003, the company dropped "AOL" from its name, and removed Steve Case (former AOL Chairman) as executive chairman in fa-vor of Richard Parsons, with AOL remaining a part of the company. 1
In the Wall Street Journal of June 2, 2006 Time Warner President and Chief Operating Officer Jeffrey Bewkes stated, what the actions of Time Warner after the merger had already revealed; the fact that the merger did not create any synergies: "While Mr. Bewkes thinks cooperation should be encouraged, he's blunt in assessing the
synergy message his predecessors preached to shareholders: "It's bull-."
The AOL – Time Warner deal and others like it gave many economists reason to question the traditional theo-ries that sought to explain the rationale behind mergers by pure economic reasoning. Although the late nineties saw a boom in internet and telecom use, this did not seem enough to justify the extraordinary rise in stock mar-ket valuations or the subsequent M&A wave. This spurred a group of economists such as Shleifer and Vishny (2003) and Rhodes-Kropf, Viswanathan (2004) and Goel and Thakor (2005) to develop new theories which explain merger waves as the result of managers exploiting the overvaluation in their firm's stockprices. These new theories differ from the traditional neoclassical theories such as Jovanovic and Rousseau (2002), which expect mergers to be wealth creating due to synergy gains, such as increases in operating efficiency or market power. Neoclassical theories also assume that capital markets are reasonably efficient and equity values therefore, on average, reflect a fair estimate of the firm's future cash flows. Significant and long term equity misvaluation, as assumed in the behavioral framework, should therefore be nonexistent.
Recent papers2 have attempted to verify one or more of the predictions made by the contending theories.
The results of these emperical investigations differ. Firm valuation does seem to play a role in acquisition deci-sicions but it is unclear if the effect of valuation is attributable to neoclassical or behavioral causes. There is also a possibility that both theories could be true. Rhodes-Kropf and Viswanathan (2004), who have developed a theory of mergers based on market overvaluation, emphasize that overvaluation may not be the only cause of mergers but can cause mergers independent of other causes such as deregulation or innovation. This study tries to examine if overvaluation is an independent cause in large deals. It does so by examining the predictions of both theories concerning the long and short term effects of a deal. This subject is of interest as mega deals such as the AOL-Time Warner deal are often mentioned as anecdotal evidence for the theory that inflated stock mar-ket prices drive mergers. This notion also has some basis in theory, as the behavioral theories of both Sleifer and Vishny (2003) and Goel and Thakor (2005) predict that acquisitions size rises as the overvaluation of the ac-quirers stock rises. Large deals should therefore be especially motivated by overvaluation. A other reason to study relatively large deals is that any long term effects of (over)valuation in small deals will probably be ob-scured by other changes to the firm in the years following the merger. The long term effect of large acquisitions will more likely be significant. As the deals in the late nineties seemed the motivated by overvaluation this is the time period I will take under investigation. The research question that this study will try to answer is the follow-ing:
To what extent were large deals in the dot com era influenced by overvaluation?
For the purpose of examining the effect of overvaluation on large mergers a sample of 259 US mergers between 1994 and 2004 is examined. The time period ensures that the sample includes deals from the high valuation dotcom era (1998-2001) and also deals during periods of more normal market valuations. Large is defined as the target's equity being at least $100 million.
This study follows earlier papers on the causes of mergers like Mitchell and Mullerin (1996), Mulherin and Boone (2000), Andrade et al. (2001), Hartford (2005), Rhodes-Kropf et al. (2005) and Dong et al.(2006). To examine the effects of overvaluation this study relies on a measure of a firms fundamental value (V) as high market valuations do not necessarily have to be the result of overvaluation but can also be a sign of large growth potential. To separate the two, this study relies on a breakdown of a firms market to book valuation ratio (M/B), in a measure of overvaluation, market to value (M/V) and a measure of growth potential, value to book (V/B). If a stock isn't overvalued, its M/V ratio should be close to one, meaning that its market value is close to its fun-damental value. Such a stock could however have large growth opportunities, which would cause its fundamen-tal value to be much larger than its book value, resulting in a high V/B ratio and also a high M/B ratio.
To investigate predictions concerning synergy gain and operating performance three additional performance measures are used. Return on capital (ROC) and operating margin (OM) are used to study the differences in financial efficiency and operating efficiency respectively. Abnormal return is our last performance measure and is used to measure the market's short term assessment of the synergy gain produced by the merger.
The remainder of this paper is organized as follows. Section 2 gives an overview of the main findings of pre-vious studies about mergers and acquisitions. Section 2 also describes the theoretical developments in the field of M&A, from these the main hypotheses are derived. To analyze these hypotheses we arrive at section 3 where the data and methodology used are presented. Section 4 presents and analyses the results. Section 5 contains the conclusions of this study.
2.
LITERATURE
This section gives a summary of the current theories which try to explain the association between stock market valuation levels and mergers. Following this is a brief overview of the empirical research in the field of M&A and its findings. The remainder of this section is dedicated to developing hypotheses that will be tested in this study.
2.4 Theories about merger waves
Despite the consensus in the corporate finance community about the existence of merger waves, the recent evi-dence about the link between stock market valuation and merger activity created a fierce debate as to the ratio-nale for explaining the clustering of mergers. A vast number of potential motives have been proposed. Almost all of the hypotheses that were put forward can be grouped in two camps. They are either neoclassical in nature or behavioral. The two most prominent theories that were developed are the neoclassical Q-theory (QT) and the behavioral overvaluation hypothesis (OVH).
2.4.1 Overvaluation hypothesis (OVH)
Shleifer and Vishny (2003), Rhodes-Kropf, Viswanathan (2004) and Thakor (2005) all have developed theories of mergers based on stock market overvaluation. Although these theories differ in their details they have a few things in common. The first is that they assume that the market does not always efficiently price stocks and con-sequently a level of significant and long term overvaluation can appear in certain (groups) of stocks. The second is that managers capitalize on this overvaluation by making unwise acquisitions and paying for this with this overvalued equity. The size of these deals rises as the level of overvaluation rises. The market does not react correctly to these unwise investments at their announcement as often investors falsely believe that they are wealth creating so the companies and their managers are not punished for this behavior.
Goel and Thakor (2005) developed a theory that explains merger waves in terms of CEO envy. They hypothes-ize that CEO pay is highly depended on firm shypothes-ize. As such any acquisition in an industry that increases firm shypothes-ize and compensation for one CEO, will trigger other CEO's of similar sized companies in that industry to be en-vious of the resulting pay rise. This compels them to attempt value dissipating but size-enhancing acquisitions of their own. To explain merger waves Goel and Thakor (2005) cite evidence that market values, and thus firm sizes diverges in periods of high valuation and converges during market downturns.
Long periods of rising markets thus create differences in firm size and cause a spiral of ever increasing envy, leading to larger and larger deals with lower and lower synergies. This spiral only ends when the bull market ends and firms sizes start to converge again. The biggest deals should therefore occur when market valuation levels peak and should also be the ones with the lowest synergies. The theory also predicts that more envious CEOs are more likely to engage in acquisitions and pay higher premia.
In their theory the perceived synergy of a deal by the market is higher than the actual synergy (if any at all). As a result there is no negative market reaction at the announcement of the deal. Some months or years after the mer-ger investors will perceive that their expectations are not met and the firm’s stock will fall to the lower valuation levels.
This theory also predicts that cash is preferred when a target is undervalued relative to its fundamentals and that equity is preferred when a target is undervalued relative to its acquirer. Acquisitions are simply a way of exploit-ing misvaluation in the market, as the overvaluation hypothesis assumes no synergies resultexploit-ing from the merger. Because of the lack of synergies any long term wealth effect of a deal is simply the sum of the stock market valuation from target and acquirers correcting itself. Consequently the long term wealth effects of stock acquisi-tions should be more negative than those following cash acquisiacquisi-tions, as stock deals involve two overvalued firms instead of one for cash deals. The theory also predicts that overvalued firms will be more likely to make diversifying acquisitions as the overvaluation is usually related to the entire industry. Therefore acquiring firms from the same industry, in the case of industry wide overvaluation, does not solve the problem.
A logical problem with this theory is the choice of asset that managers use to protect their shareholders from eventual wealth loss. During a period of stock market overvaluation nearly every firm will be overvalued to some extent, although maybe not to the extent of the acquirer. Combine this with the unavoidable takeover pre-mium; buying other companies seems a very expensive way to mitigate the effect of overvaluation. A more log-ical choice would be to issue shares to buy back ones debt or to buy some other asset which the market prices correctly, such as government bonds. Shleifer and Vishny (2003) make the case that such actions would have an adverse signaling effect and would alert investors of the lack of growth opportunities thereby decreasing the value of the firm.
2.4.2 Q-theory (QT)
Unlike the behavioral OVH, neoclassical theories expect mergers to be wealth creating due to synergy gains such as increases in operating efficiency or market power. They also assume that capital markets are efficient and equity values therefore, on average, reflect a fair estimate of the firm's future cash flows. Significant and long term equity misvaluation, as assumed in the behavioral framework, should therefore be nonexistent. Neoc-lassical explanations of merger waves have been around for a long time (e.g. Gort, 1969). These theories assume that economic disturbances, lead to a need within industries to reorganize themselves.
Mitchell and Mulherin (1996) had already presented empirical evidence that differences in the rate of takeover and restructuring activity between industries can be explained by changes in the regulatory or economic envi-ronment. Maksimovic and Phillips (2001) use performance improvements at the plant-level to support a neoc-lassical theory of merger waves.
The neoclassical theory that lends itself best to empirical testing and is most widely cited is authored by Jova-novic and Rousseau (2002). They developed a model in which technological change leads to increased disper-sion in q ratios.
A high Q indicates that a firm is well run and or enjoys good growth prospects. In this view takeovers are a dis-ciplining mechanism, which eliminates inefficient target behavior and lets the target profit from the better growth and investment opportunities of the acquirer. Yet the theory does not rule out that takeovers can also be a means of empire building for wasteful managers.
The neoclassical theory assumes capital markets on average are reasonably efficient The announcement day return of the bid should therefore be related to the difference in Tobin's Q between bidder and target. Because a deal between a competent acquirer and a less able target should produce higher synergies and have more scope for wealth creation than a deal where the reverse is the case. Gugler et al. (2006) noticed that there is a logical problem with this theory, as buying firms with a Q higher than 1 should never be very desirable. New capital equipment always has a Q of 1 (market value equals replacement value), so why would anyone want to purchase for example a used factory for a Q of higher than 1 when a similar factory could be purchased more cheaply by investments in new capital equipment which is always priced at a Q of 1. Following this logic in a time of high stock market valuations we would expect to see a smaller number of mergers, because investments in new capi-tal equipment become more attractive. In the defense of the Q-theory, one can argue that some assets such as (patents, knowhow or consumer goodwill) cannot be bought off the shelf and takeovers are the best method of acquiring them. If this is the case than any deal where the target has a lower Q than the bidder would still be wealth creating and desirable.
2.2 Previous findings
One or both of these theories could offer an explanation for the cause of mergers. Most empirical research sup-ports the idea that mergers happen in waves (Mitchell and Mullerin, 1996; Mullerin and Boone, 2000; Jovanovic and Rousseau (2001); Andrade, Mitchell and Stafford, 2001; and Harford, 2005). All conclude that mergers are clustered in time and by industry.
From findings based on the short term event study method3 we also know thattarget shareholders experience big
wealth gains but the shareholders of acquiring firms on average experience a small wealth loss, earning zero or negative abnormal returns.
This raises the question what the net economic effect is of a transaction. As acquirers are usually bigger than targets this requires the calculation of the total market value created or destroyed during a deal. Moeller et al. (2005) did just this. Using standard event methods they examined the dollar returns of 12.063 takeovers between 1980 and 2001. They discovered that the dollar returns for most acquisitions in the 90s were positive, but that there were a small number of large loss deals between 1998 and 2001 whose value destruction was greater than the value creation in all other deals in the previous periods. Although their short window event methodology is incapable of detecting any of the long term predictions made by the OVH these results do indicate that there is something special with large deals in the late nineties.
Method of payment has also been shown to be associated with the wealth effects of takeovers.
A great number of studies (Bruner (2001)) have shown that deals paid in cash have a better performance than those that use stock as the method of payment. Martin (1996) finds that acquirers which have a high market-to-book ratio tend to favor stock and those with low M/B ratio favor cash. From this the author concludes that firms with better investment opportunities (higher Q) favor stock, as it is more flexible. Rau and Vermaelen (1998) found in their long term event study that high market-to book (glamour) firms underperform after the
merger while low market-to book (value) firms outperform similar firms after the acquisition. Although high M/B firms favor stock the association between valuation and long term merger performance was significant independent of the method of payment used.
Recent papers have attempted to verify some of the predictions made by these new behavioral theories. Hartford (2005) tests both the neoclassical and the behavioral theory. He finds that the measures for deregulatory events and economic shocks are highly significant in predicting mergers. From this he concludes that neoclassical mo-tives are the most likely cause of merger waves. Dong et al. (2006) find that bidders have higher valuations than their targets apparently confirming the behavioral theory. Yet this finding can also been seen to confirm the neoclassical side of the story as more successful companies receive higher valuations by the market and display wealth creating behavior by taking over less successful businesses and converting them to their more successful business practices. The authors therefore conclude that based on their results they cannot reject either of the theories.
Rhodes-Kropf et al. (2005) find that a significant fraction of merger activity is explained by misvaluation. Q-theory suggests that successful transactions have large market-to book differences between bidder and target. However, Rhodes-Kropf et al. (2005) find that failed transactions have larger differences than completed trans-actions, while successful deals display higher levels of misvaluation. Even in industries that appear to have ex-perienced an economic shock, most acquirers come from the highest misevaluation quintile. Based on this they support misvaluation theories based either on behavioral explanations or on asymmetric information between otherwise rational managers and markets. The conclusion states that economic shocks could well be the funda-mental drivers of merger activity, but misvaluation affects how these shocks are propagated through the econo-my. Bouwman et al. (2009) explore the short and long term returns of mergers in low and high valuation mar-kets. They conclude that announcement returns are significantly better for acquisitions announced in high-valuation markets relative to those announced in low-high-valuation markets. The long run returns in their sample show a reverse pattern, with acquirers buying in high-valuation markets significantly underperform relative to acquirers buying during low-valuation markets in the two years following the acquisition. Their finding is con-sistent with behavioral predictions that overvaluation leads to wealth destroying investments and that investors fail to appreciate the wealth effect of the merger in the short term.
2.6 Hypothesis development
As both theories predict that high market valuations are associated with increased takeover activity, this associa-tion cannot be seen as evidence for overvaluaassocia-tion theories. For distinguishing between the two contending theo-ries we will require more elaborate tests.
To examine if the large mergers are influenced by overvaluation or neoclassical motives we need testable hypo-theses. The two contending theories both offer a number of testable predictions, which are summarized in Table 1.
As there is neither one behavioral nor one neoclassical theory this paper will use an overvaluation hypothesis based on Sleifer and Vishny (2003) for the behavioral predictions and the Q-theory based on Jovanovic and Rousseau (2002) for the neoclassical predictions. For these theories and their predictions see the previous para-graphs.
Table 1. Predictions of the overvaluation hypotheses and the Q-theory
This table summarizes the predictions made by the overvaluation hypotheses (OVH) and the Q-theory (QT) respectively. The results column lists the table(s) were the prediction is tested.
OVH Q-Theory
Predictions related to acquisition characteristics
Results Results
Pre merger valuations 1. Overvaluation acquirer is larger than overvaluation target
2. Method of payment is related to level of misvaluation acquirer relative to target 3. Probability of diversification related to level of misvaluation acquirer
4. misvaluation acquirer related to amount of assets acquired
T6 T7, T8 T8 T8,T10
5. Valuation acquirer is higher than valuation target
6.Differences in valuation between acquirer and target are related to differences in operating efficiency and growth prospects.
7. Synergy gains are depended on difference between valuation of ac-quirer and target
T6 T6 T8,T10
Pre merger operating
performance 8. No difference in operating performance between acquirers and targets T6 9. Operating performance of acquirer is higher than operating performance target
T6
Predictions related to acquisition outcomes
Results Results
Long term change in
valuation levels 10 . Long term change in valuation levels is dependent on pre merger level of overvalua-tion acquirer and target.
11. Long term change in valuation levels is associated with the method of payment
T8,T9, T10 T9, T10
12. Long term change in valuation levels not dependent on pre merger variables
T8, T9, T10
Post merger operating
performance 13. There is no change in post merger oper-ating performance T8, T9 14. Post merger operating perfor-mance is related to synergy gains (ARC), registered at announcement.
T8
The OVH for instance predicts that managers buy other firms that are less overvalued than their own to lower the overall level of overvaluation of their firm. (1) Target firms should thus have a substantial lower level of overvaluation than their acquirers.
(3) According to the OVH, overvaluation also leads to a higher proportion of diversifying mergers, as overval-uation is likely to be industry dependent. Diversification into other, less overvalued, industries is the most effi-cient way of lowering the average valuation level of the firm.
(4) The size of the acquisition, relative to the size of the acquirer (measured as market value of equity) is also related to the level of misvaluation according to the OVH. Small acquisitions will do little to alleviate high le-vels of overvaluation, hence more overvalued firms should be driven to make larger acquisitions.
(8) As the OVH assumes that mergers are not driven by the need to create economic value, there need not be any difference between operating performance of acquirers and targets.
The Q-theory also predicts differences in valuation between bidders and targets. (5) Acquirers tend to have higher Q's than targets
(6) The differences in Q valuation are caused by the underlying differences in operating efficiency and growth prospects . As mergers in the Q-theory only take place when there is scope for wealth creation for the acquirer through operating synergies, targets should have a lower operating performance than their acquirers, otherwise there wouldn’t be any possibility for improvement.
If, as the Q-theory predicts, firms are efficiently priced, the markets would recognize immediately that a deal between a highly valued bidder and a target with a low valuation create more value than the opposite. As differ-ences in valuation mostly reflect differdiffer-ences in growth opportunities and operating efficiency, the scope for val-ue improvement is larger when a successful acquirer buys an unsuccessful target than in the opposite case. (7)Announcement day returns should therefore reflect the difference in valuation between bidder and target. (9) Acquirers should have higher levels of operating performance than targets as mergers are a way of eliminating inefficient business models.
The OVH assumes that mergers are a way for firms to exploit long term equity mispricing, and that it takes in-vestors some time to discover the mispricing. Misvaluation should eventually correct itself a few years after the merger, but not directly at its announcement. (10)The change in the level of misvaluation is predominantly dri-ven by the pre merger level of overvaluation. (11) The method of payment should also be related to post merger changes in valuation, as the choice between cash or equity bids is driven by the under- or overvaluation of tar-gets. Targets in cash bids should be undervalued relative to fundamentals, while targets in equity bids should be undervalued relative to their acquirers. When the market eventually corrects the mispricing, cash acquirers should see more positive valuation changes relative to equity bidders.
The neoclassical framework, of which the Q-theory is an extension, asserts that markets are efficient at assessing the wealth effects of mergers. Following the merger, valuation levels on average should remain constant relative to the industry as any change in company value due to M&A activity should be registered at its announcement. (12) No relationship should exist between pre merger valuation information and post merger changes in valua-tion as this would be a strong indicavalua-tion that market pricing is not efficient. Longer term post merger share per-formance should also be indistinguishable from non-merging firms.
The neoclassical theory, of which the QT is an extension, assumes that the market is efficient at assessing the wealth effects of a merger.
(14) Abnormal returns registered on announcement should therefore be a good predictor of the improvements in operating performance following the merger.
3.
DATA AND METHODOLOGY
This section presents the data and methodology used to examine the previously formulated hypotheses. The data used is described in the first subsection, starting with the criteria for sample selection, followed by a description of the data construction and finally a summary of the sample characteristics. The second sub section deals with the methods used to analyze the data.
3.1 Data
This sub-section describes the selection of the sample of acquirers, gives an overview of the sample characteris-tics and describes how the dataset is constructed in order to analyze the results.
3.1.1 Sample selection criteria
The sample of takeover bids is obtained from Mergers & Acquisitions Report and the ZEPHYR database. The sample includes mergers and acquisitions made between 1995 and 2004 between U.S. firms. The sample in-cludes only completed deals subject to the following selection criteria:
- Both the acquiring and target firms were traded on the NYSE, AMEX, or the NASDAQ stock exchanges. - Their price and return data for both the acquiring and target firms are available for 205 days before the acquisition or merger announcement and 30 days after the deal announcement.
- The value of the transaction $100 million or more.
- The offer is announced between January 1, 1995 and December 31, 2004. - Deals where there are multiple attempts to acquire the same target are excluded. - The acquiring firms must have filed annual reports for the 3 years after the merger
- The market value of the target firm must be at least 1% of the market value of the acquiring firm 5 days before the announcement date.
- The merger must be completed 1 year after the announcement. - 100% of target stock must be acquired
- Both acquirer and target must not operate in the financial industry
To limit the number of compounding events such as changes in reporting standards all companies had to be listed on US security markets. This ensures that all the firms comply with the same accounting standards and that all the companies’ filings can be found in the SEC4 database. The estimation of the market model
parame-ters require that sample firms have at least 205 daily returns available prior to the first public announcement of a merger or tender offer. These estimates are used to calculate the abnormal return surrounding the announcement period. To ensure that all companies have at least 3 years of accounting data after the merger announcement the cutoff was determined at December 2004. As this study also tries to investigate any synergistic gains in operat-ing performance, I discard partial deals and deals which are not completed 1 year after the merger as the changes in these firms cannot or only partial be attributed to synergy effects. Firms that operate in the financial industry were excluded because they are hard to value using conventional methods.
3.1.2 Sample characteristics
Table 1 presents the distribution of the deal announcements per year. Valuation levels increase until 2000 and decline thereafter. Bids are categorized as cash deals when 50% or more of the deal is paid in cash, or if inves-tors have the option of being paid out in stock or cash, as previous research has shown that invesinves-tors favor this method of payment. All other acquisitions are classified as stock deals.
Table 2. presents the distribution of deals over the various Value Line industries. Industries for which no firm is present in the sample are not presented. Noteworthy are the large number of deals in the computer and software industry, the drug industry, the telecom sector and petroleum industry. The sample contains very little deals in the internet sector although a great deal of merger activity took place in this sector. The sample selection criteria which required firms to survive at least 3 years post merger was a condition that a majority of firms could not meet. Furthermore the minimum size consideration of $100 mln. and the requirement to be publicly listed for at least two years excluded many internet firms from my sample. The underrepresentation of internet companies will probably make it less likely that predictions of the overvaluation hypothesis will be supported. As the IT sector was probably the sector with the highest valuations, largest drop in equity prices and the lowest profits.
Year #mergers Median deal
value ($) Mean deal value ($) Cash Stock Median M/B Median M/V
1995 12 2.442.000 4.856.083 7 5 3.815 1.588 1996 12 3.961.000 8.110.000 4 8 3.126 1.360 1997 12 13.675.500 10.529.383 2 10 4.362 1.909 1998 12 10.542.500 16.921.833 0 12 4.960 2.533 1999 15 11.600.000 19.315.767 3 12 4.632 2.250 2000 62 1.058.250 6.062.128 30 32 4.643 1.537 2001 54 898.500 2.367.294 23 31 3.674 2.430 2002 33 388.920 2.811.824 18 15 3.127 1.571 2003 33 548.240 1.377.292 19 14 2.645 1.667 2004 14 1.106.985 1.445.673 8 6 3.598 1.176 Total 259 114 145
Table 2. Sample characteristics
Table 4 presents the bivariate correlations between the different performance measures. Most noticeable are the high correlations between the M/B and several other performance variables.
Industry Target Acquirer Target Acquirer
1 Advertising 2 2 32 Household Products 2 2
2 Aerospace/Defense 8 9 33 Industrial Services 6 2
3 Air Transport 0 2 34 Information Services 2 1
4 Apparel 2 3 35 Internet 5 7
5 Auto & Truck 0 1 36 Machinery 6 2
6 Auto Parts 1 0 37 Medical Services 7 7
7 Beverage (Alcoholic) 1 1 38 Medical Supplies 5 12
8 Beverage (Soft Drink) 0 1 39 Metals & Mining (Div.) 2 3
9 Biotechnology 1 1 40 Oil/Gas Distribution 1 1
10 Cable TV 1 1 41 Natural Gas (Diversified) 0 1
11 Chemical (Basic) 3 4 42 New spaper 2 2
12 Chemical (Specialty) 5 1 43 Office Equip & Supplies 1 1
13 Computer & Peripherals 9 6 44 Oilfield Services/Equip. 3 5
14 Computer Softw are & Svcs 31 23 45 Packaging & Container 2 0
15 Diversified Co. 1 7 46 Paper & Forest Products 7 9
16 Drug 19 19 47 Petroleum (Integrated) 1 3
17 E-Commerce 1 1 48 Petroleum (Producing) 17 17
18 Educational Services 1 1 49 Pharmacy Services 5 3
19 Electric Util. (Central) 2 2 50 Precision Instrument 6 6
20 Electrical Equipment 1 1 51 Publishing 1 2
21 Electronics 5 5 52 R.E.I.T. 1 0
22 Entertainment 8 10 53 Railroad 0 1
23 Entertainment Tech 2 3 54 Retail (Special Lines) 8 5
24 Environmental 3 3 55 Retail Store 2 3
25 Financial Svcs. (Div.) 3 4 56 Semiconductor 6 9
26 Food Processing 7 6 57 Semiconductor Equip 2 2
27 Food Wholesalers 1 2 58 Telecom. Equipment 10 7
28 Furn./Home Furnishings 2 1 59 Telecom. Services 15 16
29 Grocery 1 1 60 Tobacco 1 1
30 Homebuilding 2 3 61 Trucking/Transp. Leasing 7 2
31 Hotel/Gaming 2 2 62 Wireless Netw orking 1 1
Total 125 126 134 133
Table 3. Industry distribution of sample
3.1.3 Data sources and data construction
The data used in this paper was obtained from several databases. The acquisition information was collected from the ZEPHYR database. As ZEPHYR does not list mergers pre 1999 the rest of the sample was collected from the journal Mergers & Acquisitions Report which list all mergers and acquisitions undertaken in the US. The ZEPHYR database did have some flaws. In the database the day and month were reversed due to English style dates that were recorded as European dates. For instance a merger announcement that was made on the 4th
of March would actually have been on the 3rd of April. Because of this all merger were manually looked up in
EDGAR, the online SEC database, for press releases or other statements which contained the announcement date of the intention to merge.
Accounting data and daily stock returns (Total return index includes dividends and stock splits) were obtained from the Thompson Datastream database. It lists the balance sheet, income statement and cash flow statement for most listed companies in the world. The Thompson database does contain rather a large number of irregulari-ties. For instance in some years certain amounts are listed in thousands and other years in millions. Some years are completely missing from the database and had to be reconstructed using 10-K filings in EDGAR, the SEC database. The accounting data obtained from the Datastream database consist of the following variables: EBIT, bookvalue of debt, bookvalue of equity, cash, net profit, sales. Companies that did not comply to the sample selection criteria or could not be found in the Thompson Datastream database were discarded. The eventual sample consists of 259 mergers and acquisitions announced between 1994 and 2005 in the U.S. As a measure of a firm's growth potential the industry growth rate of sales is used. Koller (2005) notes that the growth rates of individual firms have the tendency to move toward the industry average over time. The industry average there-fore forms a better approximation of a firms growth potential than the individual growth rate, which can be
dis-M/BT M/BA ROCT ROCA OMT OMA ARA ART ARC M/VT M/VA gT gA V/BT V/BA M/BT 0,35 0,06 0,02 -0,08 0,02 -0,06 -0,02 -0,04 0,82 0,05 0,07 0,05 -0,06 0,01 *** *** M/BA 0,17 0,24 -0,02 0,10 -0,19 -0,02 -0,17 0,25 0,52 0,15 0,21 0,04 0,13 *** *** *** *** *** *** ** *** ** ROCT 0,13 0,25 0,07 -0,10 -0,04 -0,12 0,00 -0,06 0,15 0,19 0,03 -0,02 ** *** ** *** ROCA 0,16 0,62 -0,13 -0,04 -0,13 0,04 0,21 -0,08 -0,13 0,04 0,19 ** *** ** ** *** ** *** OMT 0,25 -0,07 -0,15 -0,06 -0,13 0,12 0,00 0,01 0,09 0,02 *** ** ** OMA -0,09 -0,08 -0,11 0,02 0,24 -0,09 -0,11 0,01 0,14 *** ** ARA 0,09 0,80 -0,02 -0,09 -0,13 -0,13 -0,03 0,06 *** ** ** ART 0,30 0,01 0,00 -0,04 -0,01 -0,14 0,09 *** ** ARC -0,03 -0,05 -0,15 -0,11 -0,01 0,02 ** M/VT 0,03 0,05 0,04 0,03 0,03 M/VA -0,01 0,04 0,01 0,07 gT 0,73 0,01 0,00 *** gA 0,01 0,04 V/BT 0,00 V/BA
This table provides the matrix of bivariate correlations for the variables used to measure performance from both acquirers and targets in the entire sample (N=259). M/B is the market to book ratio. M/V is the market to fundamental value ratio. V/B is the value to book ratio. OM stands for operating margin. ROC stands for return on capital. g is the average grow th rate of revenue for the industry to w hich the firm belongs. All variables and their defintions can be found in paragraph 3. ***, **, * denote that the correlation is significant at the 1%, 5%, and 10% level, respectively.
torted by non recurring events such as divestitures or purchases of business units which may significantly impact sales growth in a specific year but have no influence on the long term growth potential. A overview of all va-riables used and their definitions can be found in the appendix.
The market values of the acquiring and target firms were obtained from the Thompson Datastream database. The Thompson database lists daily market values by multiplying the share price with the number of shares out-standing. Industry information about operating performance was obtained from a dataset from the New York University5 coming from the Value Line Database. The Value Line Database: tracks about 7000+ US-firms and
which are divided in 100 industries and provides accounting data on them. The variables obtained from the da-tabase are the industry variables for growth rate of revenue, operating margin, market to book ratio, return on capital and the cost of equity capital (re).
3.1.4 Measures of operating performance
For testing the predictions regarding operating performance this study uses two measures of operating perfor-mance. As neither theory is very specific in what constitutes operating perforperfor-mance. To measure operating effi-ciency, operating margin (OM) is used and to measure financial effieffi-ciency, return on capital (ROC) is used. The measures of operating performance that are used are similar to those used by Agrawal and Jaffe (2003). They defined operating performance both in terms of operating return on assets (OPA) and operating return on sales (OPS). I use OM and ROC and not OPA and OPS because I want to compare the firm values to the industry variables from Value Line and Value Line doesn't use OPA and OPS. Although the variables used are different I think my results can be compared to those of Agrawal and Jaffe (2003) because the variable definitions only differ slightly. The Value Line variables are defined as follows:
OMit = EBITit(1-Tind) / Salesit
ROCit = EBITit (1-Tind) / (Dit + Bit - Cit)
OMit is the operating margin for firm I at date t.
EBITit is income before interest and taxes for firm I at date t.
(1-Tind) is 1 minus the effective tax rate, averaged across the industry.
Sales are total revenues for firm I at date t. ROCit is the ROC for firm I at date t.
Dit is bookvalue of all interest bearing liabilities for firm I at date t.
Bit is the bookvalue of the firm’s capital and other shareholders’ funds for firm I at date t.
Cit is the bookvalue of cash and cash equivalents for firm I at date t.
To compare these variables before and after the merger they are adjusted to account for changes in industry op-erating performance. The industry adjusted variables are calculated as follows:
OMit is the operating margin for firm i at date t.
OMind.t is the operating margin for the industry of firm i at date t.
OMind.adj.t is the industry adjusted operating margin of firm i at date t.
ROCit is the ROC for firm i at date t.
ROCind.t is the ROC for the industry of firm i at date t.
ROCind.adj.t is the industry adjusted ROC of firm i at date t.
3.1.5 Measures of valuation
To test the OVH we require a measure of overvaluation. Measuring overvaluation is inherently difficult. After all if anyone can locate overvalued firms so in theory can the market and once that happens these firms would quickly cease to be overvalued.
Until recently a widely used method to measure overvaluation was to examine long run abnormal stock returns. Because one would expect after a period of overvaluation, share prices would gradually return to a level reflect-ing normal valuations. The results of these long-run abnormal return studies should be interpreted with care because of statistical and conceptual problems with this method. (see, for example, Barber and Lyon, 1997; Fa-ma (1998), Loughran and Ritter, (2000). One important issue is that buy and hold studies assume that observa-tions are independent, while for example Mitchell and Stafford (2001) and Andrade et al. (2001) find that they are positively cross-correlated. Conclusions drawn from studies using long term event-methodology should therefore be approached with caution. For this study I rely on an alternative method which relies on accounting based valuation techniques to calculate a firm's fundamental value which is contrasted to the market value of equity to determine overvaluation.
The predictions of the Q-theory rely on differences in Q between companies as an explanation for merger activi-ty. Lang and Stulz (1989,1994) claim that Q is a good indicator of managerial quality and firm efficiency, which makes it an ideal instrument for comparisons between firms, this in contrast to stock returns and account-ing measures which have to be normalized and risk adjusted. They also note that Q is only a good inidicator if we assume that marketpricing is efficient. Lang and Stulz (1989) also mention that Q might be seen as a meas-ure of market mispricing. As Q is hard to measmeas-ure I use M/B as a proxy for Q as previous literatmeas-ure (Lang et al. (1989) and Servaes (1991)) has found a high correlation between M/B and Tobin’s Q.
Ang and Cheng (2006), Dong et al. (2006) and Rhodes-Kropf et al. (2005) have investigated predictions of the OVH and QT using various measures of overvaluation. Their tests predominantly used a ratio of market to book value of equity or some similar measure. A potential problem with the M/B ratio, is that it dependent on capital structure of the specific firm. The M/B ratio may therefore be an imperfect measure of misvaluation and hence not suited as for testing any predictions of the Overvaluation hypothesis.
To measure overvaluation specifically Rhodes-Kropf et al. (2005) devide the market to book ratio in two com-ponents; market value to fundamental value (M/V) and fundamental value to book value (V/B).
(Mit/Bit) = (Mit/Vit) x (Vit/ Bit)
Mit is the market value of firm I’s equity at date t.
Bit the book value of equity of firm I at date t
Vit is the fundamental value calculated by a Residual Income Valuation Model of firm i at date t
For M/V to be a relevant measure of overvaluation we have to assume that there is an accurate way to measure fundamental value. Both Ang and Cheng (2006) and Dong et al. (2006) use M/V or a similar measure of over-valuation. They calculate a firms fundamental value (V) using a residual income model and compare this to the market value of the firm to detect misvaluation. They claim this market to fundamental value ratio (M/V) filters out the extraneous information about growth and managerial agency problems inherent in M/B and is therefore a better measure of misvaluation.
Analogous to Dong et al. (2006) I classified firms with a negative M/B as having high valuation. These firms were therefore assigned the median M/B value of the quartile of firms that had the highest value of M/B in the takeover sample of that year. This adjustment was made as it is counter intuitive to judge firms with negative M/B as having a low valuation. A firm with a tiny amount of equity would have a very high M/B ratio, yet where this amount of equity to decline only a little and become negative the same firm would suddenly have a very low valuation while little had changed. A negative M/B should therefore be seen as a sign of great relative valuation by the market. Only 5 firms in the sample had a negative M/B ratio.
The calculation for the M/B and M/V ratio at the date of the deal relies on the market value 5 days prior to the deal announcement. The book value of equity and the fundamental value of equity are derived from the last annual report prior to the deal announcement. Industry variables necessary for the calculation of fundamental value are derived from Value Line and come from the same year as the last annual report prior to deal an-nouncement.
3.1.6 The Residual Income Valuation Model
The finance literature suggests several techniques that can be used to estimate a company’s fundamental value. The dividend model, the cash flow model and residual income model (based on earnings) are all valid candi-dates.
Empirical testing by Bernard (1995) and Penman and Sougiannis (1998) showed that earnings techniques have many practical advantages over approaches based on dividends or cash flows. For instance it can produce accu-rate valuations using shorter forecast horizons than the alternatives. Although al methods require the calculation of a terminal value (TV), which is the liquidation cash flow beyond the forecast horizon. D’Mello and Shroff (2000) found that the TV only accounted for 11% of their total equity value using the residual income model (RIM), 33% using the cash flow model and 58% using the dividend discount model. Both authors also report that the RIM model produced lower valuation errors than alternative techniques when compared to actual mar-ket valuations.
There is also strong support that value is an indicator of mispricing. Lee et al. (1999) for instance report that the aggregate price to residual income value ratio predicts the one-month-ahead returns of the 30 stocks that make up the Dow Jones Industrial Average Index better than aggregate price-to-book. Frankel and Lee (1998) find that V is a good predictor of cross-sectional stock returns for 1 year and beyond and that these returns are not attributable to book to price ratio, firm size or beta.
(1) !" # ! (2) $ ! " # !
Vit is the residual income fundamental value of the firm i at date t.
Bit the book value of equity of firm I at date t
Et : expectation based upon information available at date t
Niit : Net income for period t
re: cost of equity capital for the industry of firm I at date t
ROEit : Return on book equity for period t
The cost of equity capital is estimated using the capital asset pricing model: Cost of Equity = Riskfree Rate + Beta (Risk Premium)
The average beta for the industry is used. The long term treasury bond rate is used as the riskfree rate, and the risk premium is assumed to be 5.5%.
This equation states that firm value is the sum of book value plus the sum of all discounted future earnings in excess of a capital charge for the use of equity capital. For practical application the above infinite sum has to be replaced by a sum of finite periods plus a terminal value calculation. Lee et al. (1999) found that expanding the forecast period beyond 3 years added little predictive power to their value estimate. I also use a 3 year forecast period and discount the last period’s income as a perpetuity.
(3)
$% ! "
# !
$%& ! % "
# !& $#%' !&! %&"
In equation (3) the estimation of Vt requires the expectations of future earnings (Et), the cost of capital (re) and
book values (Bt). Expectations are off course unobservable, and research suggests that analyst forecasts are a
poor proxy (See Brown 1993; O'Brien 1988 and Dechow and Sloan 1997). Because of this I use realized post event earnings, bookvalues and industry level estimates of cost of capital, for the post merger calculation of Vt,.
Thereby assuming that investors on average have perfect foresight of future earnings.
For calculating the terminal value (TV) the assumption is made that expected residual earnings remain constant after year 3. (Lee et al. (1999), D’Mello and Shroff (2000) and Dong et al (2006) make the same assumption in their calculations). Just as D’Mello and Shroff (2000) and Dong et al (2006) To protect against the terminal value being negative multiple calculation methods are used. This choice is motivated by the assumption that it is unlikely that long term value of the firms is negative. D’Mello and Shroff (2000) and Dong et al (2006) also use multiple calculation methods to protect against the option of a negative TV . This method also ensures that V has the largest possible value and isn't biased toward detecting overvaluation. For calculating TV I use three methods and use the largest. All methods are reasonably conservative as they assume that return on equity and cost of capital will not deviate much from past values or the industry average. The first method assumes that the growth rate of equity is equal the firm's average ROE in the preceding two periods minus industry re , the second
method assumes that the growth rate of equity (ROEit - re) is 5% which seems common among most Value Line
industries. The third method assumes that the growth rate of equity is equal to that of the industry, re andROEit
3.2 Methodology
In this part the methodologies used to analyze the data are discussed. First, the event study methodology used to measure the abnormal announcement returns is explained. Second a description of the non-parametric tests used to determine significance variations in the data. Third, an explanation of the regression analysis, which tests the impact of deal characteristics on other variables.
3.2.1 Event study methodology
The Q-theory assumes market to be reasonably efficient. As such the short term market reaction to a bid an-nouncement should on average give a reasonably accurate reflection of the synergy gains produced by a deal. To study if shareholder value is lost or created in an acquisition we have to estimate abnormal returns. To do this we use standard event study methods (see Brown and Warner (1985) and MacKinlay (1997)) and compute mar-ket model abnormal returns using the S&P 500 as the marmar-ket index. The data used is the return index (RI) from Thompson Financial Datastream, which is adjusted for stock splits and dividends. and derived Following Moel-ler et al. (2005) I estimate the parameters for the market model over the (-205,-5) day interval.
Because the total wealth effect of the merger is of interest I follow Moeller et al. (2005) in measuring the impact of the acquisition announcement on the combined value of the acquiring firm and target firm in percent returns. If an acquisition involves synergy gains, the loss in value for the acquiring firm is more than offset by the gain for the shareholders of the acquired firm. Bradley, Desai, and Kim (1988) show that such an outcome is typical for their sample of takeovers. The abnormal return for both target and acquirer are reported, as well as the ab-normal return synergy gain (both firms combined), in both dollars and percentages, following the method of Bradley, Desai, and Kim (1988)
For each of the acquiring and target firms the market model regression is calculated using 200 trading days, starting 205 days before the first announcement and ending 5 days before the first announcement of a tender offer or merger. Abnormal returns are defined as the difference between the actual return and the expected re-turn of an individual stock measured of the period 3 days before the announcement and ending 3 days after the announcement. The expected return is estimated by calculating the alpha and the beta of the market model, by using the daily stock- and benchmark return.
(1) ARit = Rit – ( i + i * Rm ) t=(-3, 3)
ARit = abnormal return for stock i at time t
Rit = the total return on stock i at time t; and
Rm = the total return on is the S&P 500 index at time t
i and i = market model parameter estimates estimated in (-205,-5) days before bid announcement
The combined abnormal return (abnormal return synergy gain) is calculated as follows: (2) MA =MA * ARA
(3) MT =MA * ART
(4) ARC = ( MA + MT ) /( MA + MT )
ARA = abnormal return of the acquirer during event window
ART = abnormal return of the target during event window
MA = acquirer's market value of equity before event window (5 days before announcement)
MT = target's value of equity before event window (5 days before announcement)
3.2.2 Pre and post merger comparison
To adequately examine the impact of the merger on firm performance, the performance of the firm 1 year before the merger is compared to that of the firm 3 years after the merger. As we have two firms before the merger and one firm following the merger some arithmetic is required to make this a fair comparison. Data for acquirer and target is added together to create a virtual firm before the merger. For instance to calculate the pre merger M/B the market value and book values of equity for both firms are added together. The same method is used for the calculation of other variables
The calculation for the operating margin of the (virtual) combined firm pre merger looks like this:
()* $ + # +!, $ + # +!-!
./0123 ./0124
Where OMC is the operating margin of the combined firm (acquirer plus target)
EBIT(1-t)A and EBIT(1-t)T are the after tax operating returns for acquirer and target respectively
SalesA and SalesT are the revenues of acquirer and target respectively
For the combined firm pre merger ROC, M/B and M/V are calculated likewise by adding the accounting and or market variables for the acquirer and target firm as if the two firms have already merged. The calculation me-thods for all the pre merger variables for the combined firm are summarized in the appendix.
3.2.3 Non-parametric tests and data presentation
To compare differences in mean characteristics in various samples and subsamples of deals I use two non-parametric test. In the case of two related samples (eg. firm ROC before and after merger) the Wilcoxon signed-rank test is used. I chose the Wilcoxon signed-signed-rank test, which is a non-parametric test that, unlike the paired Student's t-test, as it can be used in the case that the population can't be assumed to be normally distributed. The second method used is the Wilcoxon rank-sum test, which is a non-parametric test for assessing whether two samples of observations come from the same distribution (e.g. cash and stock deals). The Wilcoxon rank-sum test is almost identical to performing an ordinary parametric two-sample t -test on the data after ranking over the combined samples. Its advantage lies in its robustness as it is less likely than the t test to give a false significant result because of one or two outliers.
As a way of presenting my data I use a method, similar to that of Dong et al. (2006), of ranking the data and dividing it in 4 equally sized groups. The ranking variables I use are the valuation variables (M/B and M/V) and synergy gain ARC because these have the most predictive power according to the two contending theories. I then
acquisition size of the highest M/B group to the relative deal size of the lowest M/B group we can determine if there is any association between acquirer valuation and acquisition size.
3.2.4 Regression analyses
To identify the impact of certain deal characteristics a number of least squares regressions are performed. To test the OVH, the impact of valuation levels of acquirer and target is measured on post merger changes in performance and valuation. Valuation levels should determine the potential synergy gains according to the QT. To test this the effect of (M/B) on synergy gain (ARC) is measured. Furthermore the impact of the relative size is
studied as this should impact changes in valuation according to the OVH. Fama and Frech (1993) showed that the M/B ratio of a firm can be seen as a measure of risk. Others claim that M/B also holds information about operating efficiency, size and growth prospects, while the OVH claims that it is mainly driven by industry and market wide overvaluation. A number of control variables are added to the regression to separate the effect of valuation from other factors that might be correlated with the valuation variables.
Table 5. presents the results of the normality test of the dependent variables. All of the variables have a non normal distribution.
ARA ART ARc Relative acquisition size LOG(MT/MA) %change M/B post-merger %change M/V post-merger Mean -0,030 0,202 0,006 -0,875 0,105 0,336 Median -0,021 0,156 -0,001 -0,776 -0,260 -0,172 Minimum -0,320 -0,220 -0,260 -2,741 -0,988 -0,909 Maximum 0,303 1,443 0,401 0,537 10,554 23,785 Variance 0,010 0,055 0,008 0,466 0,934 0,151 Skew ness -0,184 1,815 0,451 -0,670 5,780 1,553 Kurtosis 1,024 5,553 2,291 -0,204 2,764 0,302 N 259 259 259 259 259 259
This table gives the descriptive statistics of the variables used in the OLS regression. To determine the normality of the dependent variables in the estimation w indow the Skew ness and Kurtosis are presented. The Skew ness and Kurtosis determine w hether thenull hypothesis of normally distributed has to be rejected. Subsequently, the mean, median, minimum, maximum and variance of the dependent variables are presented.
4.
RESULTS
This section tests if the relationships between various firm and deal characteristics are in accordance with the predictions made by the Q-theory (QTH) and Overvaluation hypothesis (OVH). Section 4.1 presents the results of the univariate tests and section 4.2 the results of the OLS regressions.
4.1 Univariate results
The differences between acquirer and target characteristics at deal announcement are reported in Table 6. The M/B ratio is used as the rational measure of valuation, while the M/V ratio is used to detect misvaluation. For the entire sample both the median M/B and M/V ratios of acquiring firms are significantly higher than that of target firms. Ang and Cheng (2006) and Dong et al.(2006) also reported that acquirers on average have higher valuations than targets.
Both the Q-theory and the OVH predict that high valuation companies take over lower valued companies. The crucial distinction is that the Q-theory beliefs that the high valuations are rational and the OVH assumes they are not. As both measures are significant neither theory can be ruled out.
N Acquirer Target A-T p-value
M/B 259 3,899 3,261 0,638 0,007 *** M/V 259 3,026 1,994 1,032 0,000 *** ROC 259 13,58% 12,66% 0,009 0,175 OM 259 13,54% 10,39% 0,032 0,000 *** V/B 259 1,308 1,502 -0,193 0,000 *** g 259 9,22% 9,40% -0,002 0,287 M/B 114 2,863 2,476 0,388 0,423 M/V 114 2,102 1,428 0,674 0,000 *** ROC 114 14,63% 13,07% 0,016 0,157 OM 114 13,78% 10,29% 0,035 0,000 *** V/B 114 1,432 1,516 -0,084 0,000 *** g 114 8,03% 8,65% -0,006 0,264 M/B 145 4,808 3,645 1,163 0,004 *** M/V 145 3,755 2,339 1,416 0,000 *** ROC 145 12,16% 11,76% 0,004 0,558 OM 145 13,29% 10,44% 0,029 0,043 ** V/B 145 1,203 1,474 -0,271 0,000 *** g 145 10,02% 10,10% -0,001 0,737
Table 6. Median acquirer and target characteristics
This table presents the median characteristics of acquirers and targets of the w hole sample (N=259) and tw o subsamples of cash(N=114) and stock acquisitions (N=145). To determine if acquirer and target have the same characteristics a Wilcoxon signed rank test is performed. The characteristics compared are the Market to Book ratio of equity (M/B), the Market to RIM value ratio (M/V), Return on Capital (ROC), operating margin (OM), value to book ratio (V/B) and industry grow th rate of revenue (g). M/B and M/V are
calcualted 6 days prior to the bid announcement. ROC, OM and g are based on annual reports and Value Line data from the end of the year prior to deal announcement. ***, **, * denote that the difference in means is significant at the 1%, 5%, and 10% level, respectively.
Full sample
Cash sample
Table 6. also reports the difference between acquirer and target characteristics in cash and stock deals separate-ly. This further analysis shows that the difference in valuation ratios is mainly driven by the subsample of equity financed takeovers. The cash subsample shows no difference in median target and acquirer M/B. The overvalua-tion variable M/V on the other hand, does differ significantly between acquirer and target in the two subsamples. Interestingly V/B, the measure for growth potential is larger for targets than for acquirers in all samples which means that misvaluation and not growth potential is the cause for the higher valuations of acquirers. Rhodes-Kropf et al. (2005) also found that acquirers had significantly higher M/V ratios than their targets yet signifi-cantly lower V/B ratios.
In the end value is determined by business efficiency. The results show acquirers on average have superior oper-ating margins relative to their targets. This seems to confirm the neoclassical prediction that acquirers with supe-rior business models take over less successful firms. It should be noted that the financial performance measure of efficiency, ROC, does not differ between bidders and targets. The ROC represents a more complete measure of operating performance as unlike OM it accounts for the amount of capital that is employed to earn a dollar of profit. There does seem to be a difference between the cash and stock subsample, as the difference in operating margin is much more significant in the cash subsample.
Table 7 tests the differences between firm characteristics in cash and stock deals. Just as Dong et al.(2006), I find that both acquirers and targets in stock deals have significantly higher M/B and M/V valuations than their counterparts in cash deals. This can be interpreted as an indication that stock deals are more motivated by over-valuation, just as the OVH predicts. It could also mean that firms in stock deals are more efficient or have better growth opportunities. If this is the case than their valuations would be justified according to the Q-Theory, so I also test for this possibility.
stock cash stock-cash p-value
M/B Acquirer 4,808 2,863 1,945 0,001 *** Target 3,645 2,476 1,170 0,015 ** M/V Acquirer 3,755 2,102 1,653 0,000 *** Target 2,339 1,428 0,911 0,000 *** ROC Acquirer 12,16% 14,63% -0,025 0,142 Target 11,76% 13,07% -0,013 0,592 OM Acquirer 13,29% 13,78% -0,005 0,196 Target 10,44% 10,29% 0,002 0,536 V/B Acquirer 1,203 1,432 -0,229 0,000 *** Target 1,474 1,516 -0,042 0,075 * g Acquirer 10,02% 8,03% 0,020 0,002 *** Target 10,10% 8,65% 0,014 0,012 ** AR Acquirer -5,65% -0,12% -0,055 0,000 *** Target 13,00% 20,99% -0,080 0,002 *** N 145 114
Table 7. Median cash and stock subsample characteristics
