CEO overconfidence, market valuation and corporate acquisitions.
An empirical investigation of the short-term wealth effects for European acquirers.
Thesis
Master of Science in Business Administration Specialization: Finance
University of Groningen Faculty of economics and business
Author: C.T. Hattink
Student number: 1553658 Supervisor: dr. P.P.M. Smid
Abstract
In this paper I empirically test the effect of CEO overconfidence and market-valuation on acquirer returns. The sample consists of 684 acquisitions, made by companies from the Euro-zone in the period 1999-2009. Depending on the proxy used, evidence is found for an CEO overconfidence effect on acquisitions. This finding implies that overconfidence negatively impacts acquirer returns. Acquisitions undertaken during high-valuation markets receive more favorable market reactions than acquisitions undertaken during low-valuation periods. Furthermore evidence is found for a combined effect of CEO overconfidence and market-valuation on acquirer returns. The results show that acquisitions undertaken by not-overconfident CEOs in high-valuation periods lead to the largest positive short-time wealth effect. When controlling for deal and firm characteristics only the acquirer size and the method of payment seem to have a significant effect on the acquirer cumulative abnormal returns.
Key words: Acquisitions, overconfidence, market-valuation, acquirer returns.
Table of Contents
1. Introduction ... 4
2. Literature Review ... 7
2.1 Overconfidence ... 7
2.1.1 The general idea of overconfidence ... 7
2.1.2 Overconfidence in finance ... 7
2.1.3 Overconfident managers ... 8
2.1.4 Measuring managerial overconfidence ... 9
2.2 Acquisitions ... 11
2.3 Market valuation ... 12
2.4 The interaction of market valuation and managerial overconfidence ... 13
2.5 Empirical evidence ... 13
2.5.1 The wealth effects of acquisitions... 13
2.5.2 Managerial overconfidence and acquisitions ... 14
2.5.3 Market valuation and acquisitions ... 16
2.6 Research question and Hypotheses ... 17
3. Methodology ... 18
3.1 Event study ... 18
3.2 Constructing overconfidence proxies ... 19
3.3 Measuring market valuation periods ... 21
3.4 Multivariate analysis ... 21
4. Data ... 23
4.1 Sample selection ... 23
4.2 Managerial overconfidence proxy data ... 24
4.3 Control variable data ... 25
4.4 Deal statistics ... 27
5. Results ... 28
5.1 CEO overconfidence and acquisition announcement returns ... 28
5.2 Market valuation periods and acquisition announcement returns. ... 30
5.3 Combined effect of CEO overconfidence and market valuation periods. ... 31
5.4 Multivariate analyses ... 34
6. Conclusion ... 39
References ... 42
Appendices ... 44
A1 Acquisition announcement returns. ... 44
1. Introduction
Empirical evidence suggests that acquirer returns in case of mergers and acquisitions are typically small and often negative. This relatively poor acquirer performance makes the reasons why acquisitions take place even more interesting (Eckbo, 2009). According to Hayward and Hambrick (1997) the motives to undertake an acquisition can be categorized into two categories: rational decisions to maximize value and irrational actions that derive from managerial biases. The former assumes that synergies are a rational argument to undertake acquisitions. Another motive, that is consistent with rational decision-making, is that the market displaces poor management. The motive that managers may not always make rational decisions when they undertake acquisitions stems from the Hubris theory of Roll (1986). Successful acquirers may be optimistic and overconfident about deal synergies. With this study I focus on the Hubris explanation for acquisitions, and investigate how managerial biases affect acquirer returns. Studying these managerial biases might help to better understand why some acquisitions receive favorable market reactions as compared to others. The idea that people are overconfident and that this possibly affects decision-making among individuals elaborates on findings in Psychology. According to de Bondt and Thaler (1995, pp. 385), “perhaps the most robust finding in the psychology of judgment is that people are overconfident”. One implication of overconfidence is that people overestimate the precision of their knowledge. People also overestimate their own capabilities. Kahneman and Lovallo (1993) studied if these psychological biases affect managerial decision-making: they found that managers underestimate risks due to these biases.
Traditional finance research has extensively focused on managerial incentives and actions, and how those affect corporate policies like acquisitions. Behavioral assumptions have entered the field, but mainly focusing on investor behavior; they have significant less effect on the corporate finance literature. Since Rolls’ Hubris theory, explanations that stem from the behavioral finance paradigm gained support in explaining merger and acquisition activities. The empirical studies that try to investigate the effect of irrational managerial behavior on M&A decisions are limited, and mainly focus on the US (Hayward and Hambrick, 1997; and Malmendier and Tate, 2008) and the UK (Doukas and Petmezas, 2007 and Croci, Petmezas, and Vagenas-Nanos, 2010). This study elaborates on this literature and investigates if optimism and overconfidence are also prevalent among managers of Euro-zone firms, and if overconfident managers have the same, value destroying, effect on acquisition decisions.
acquisitions. One of them is market-valuation. Rhodes-Kropf, Robinson and Viswanathan (2005) show that the number of acquisitions in high-valuation markets exceeds the number of acquisitions in low-valuation markets. Bouwman, Fuller and Nain (2009) find evidence that not only the number of acquisitions, but also the acquirer returns are affected by market-valuation. Croci, Petmezas and Vagenas-Nanos (2010) combine the findings of overconfidence and market-valuation, and test how they commonly affect acquirer returns. For a sample consisting of UK firms, they find that acquirer returns are highest for acquisitions made by not-overconfident CEOs in high-valuation markets. With this study I try to combine the studies on managerial overconfidence and market-valuation, and investigate their effect on acquisitions by Euro-zone firms. I focus on this geographical area to see if the outcomes are in line with studies that focus on the US and UK, or that the effects of
overconfidence and market-valuation differ between geographic areas. The results of this study might explain how managerial behavior influences decision making in firms and help to understand the outcomes of acquisition decisions. It can also provide evidence for the interaction between market valuation and managerial overconfidence.
The research question is as follows: Are acquirer returns affected by managerial overconfidence and market-valuation?
I use two proxies to measure overconfidence, one is based on acquisition frequency, and the other is based on the recent performance of the bidder. To test for the market-valuation effect I use the P/E ratio of the market to classify months as high- and low-valuation markets. For the period 2000-2009 684 acquisitions by companies from the Euro-zone are analyzed. Event study methodology is used to research the short-time wealth effect of acquisitions. This is followed by a regression analysis to further explain the CARs. Variables for overconfidence and market-valuation are included, as well as variables to test for deal- and firm characteristics.
when acquisition frequency is used to measure CEO overconfidence, the conclusion can be drawn that overconfidence has a negative wealth effect on acquisitions being announced. Testing for overconfidence and market valuation together leads to the conclusion that acquirer returns are highest for acquisitions made by not-overconfident CEOs in high-valuation markets. Furthermore, the results of the regression analysis show that deals that are completely paid for with shares receive significant negative market reactions, and that the size-effect leads to significant negative acquirer returns.
2. Literature Review
2.1 Overconfidence
In this section I review the idea of overconfidence as it is persistent in psychology and how this affects decision-making in finance.
2.1.1 The general idea of overconfidence
Overconfidence has been investigated extensively in psychology; Weinstein (1980) found that a vast majority of the people claim they are above average. He also found that people believe good things to happen more often to themselves than to others. Illusion of control, limited feedback, abstract reference points and high commitment increase this belief. People believe they have greater control and ability over events than they actually do have. In the psychological literature overconfidence can explain several phenomena. People are overconfident in the way they calibrate probabilities; they show excessive confidence in the precision of their estimates of probabilities. The calibration theory has demonstrated that people are overconfident in answering moderate to extremely difficult questions, and people are under-confident when they have to answer easy questions (Gervais, Heaton, and Odean, 2003). People also overestimate their own skills and knowledge, and
overestimate the precision of their information. Another implication of overconfidence is known as the self-attribution bias: people take credit for successes but blame failures on bad luck. Most people overestimate the degree to which they are responsible for their own success: this reinforces
overconfidence.
2.1.2 Overconfidence in finance
Since the rise of behavioral economics, the implications of overconfidence have been studied in finance as well. An important finding of the behavioral finance research is that people do not always behave rational, as assumed in the traditional finance theory. Irrational managerial behavior is “behavior that departs from rational expectations and expected utility maximization of managers” (Baker, Ruback and Wurgler, 2005, pp. 35).
risks. This is an implication of narrow framing, managers make decisions one at a time and neglect future decision opportunities. Risk taking is increased by optimism and the illusion of control, because it leads to unrealistic forecasts and unrealizable plans. Managers are willing to bear risks because they fail to recognize that they have to bear them.
Several studies have incorporated overconfidence in theoretical models to explain how
overconfidence could affect financial decisions. In these models the behavioral biases are usually divided into two groups. The first way in which managers are biased stems from optimism; managers overestimate the firms cash flows, the second way stems from overconfidence, managers either underestimate the volatility of the cash flows, or overweight private signals relative to public information (Ben-David, Graham and Harvey, 2007). However, in most studies these two terms are used interchangeable.
Heaton (2002) found managerial optimisms leads to an underinvestment/overinvestment tradeoff for managers. Optimism makes managers believe that capital markets undervalue the stocks of their company, causing managers to decline positive NPV projects because they believe external financing is too costly. On the other hand, optimism inclines managers to upward biased cash flow forecasts, causing managers to overvalue projects and ultimately overinvestment.
2.1.3 Overconfident managers
There are good reasons to believe that managers are vulnerable to the biases of overconfidence and optimism. Gervais, Heaton and Odean (2003) argue that managers are more prone to overconfidence than the normal population. Managers have to make difficult decisions, and research has proven that overconfidence is most prevalent in situations where difficult decisions have to be made. Another reason may be that overconfidence is triggered by the position managers fulfill; the illusion of control makes them become overconfident. A selection bias based explanation argues that people who are overconfident about their prospects as a manager are more likely to apply for a management position. Firms may also select people with these characteristics if they are perceived as a sign of greater ability. The model developed by Goel and Thakor (2008) shows that overconfident managers have a higher likelihood of being promoted to the position of CEO than rational managers, because they perceive less risk and therefore take more chances.
capital restructurings are relatively rare and are difficult to compare with similar situations that happened before: this seems to increase biased decision-making.
Weinstein (1980) found that people are more optimistic about outcomes that they believe they can control and people are more optimistic about decisions to which they are highly committed. Managers have control over the decision-making in the organization and are likely to be highly committed to their firm. Both characteristics increase the tendency for managers to be overconfident and optimistic.
Several empirical findings confirm the ideas that managers are affected by overconfidence. March and Shapira (1987) find that overconfidence is persistent under managers; managers believe outcomes are largely controllable and projects are less risky than they actually are. Ben-David, Graham and Harvey (2007) studied miscalibration among CFOs, and concluded that CFOs highly overestimate the precision of their predictions on the performance of their companies’ stocks. This miscalibration is the effect of personal threats and company characteristics.
2.1.4 Measuring managerial overconfidence
Now that it is clear that managers are influenced by overconfidence, I discuss how this managerial overconfidence can be measured. As Malmendier and Tate (2005b, pp. 652) put it; “The biggest challenge for the analysis of overconfidence is to construct a plausible measure of overconfidence”. It is difficult to measure overconfidence because there is no instrument to directly measure a
personality trait.
Table 1 lists the overconfidence proxies that were used in previous studies. Maybe the most direct way to measure overconfidence is by using questionnaires to test whether a person is overconfident or not. Brettel (2008) uses a survey to ask managers general knowledge questions representing a moderate to high level of difficulty. The underlying idea is to test the subjects’ awareness of the limits of their own knowledge. Evaluating the confidence levels then derived the measure of
overconfidence, finding that almost all surveyed CFOs show overconfidence. Ben-David, Graham and Harvey (2007) operate a different survey. They surveyed US CFOs and asked them to predict stock market performance for the coming ten years. Overconfidence is defined as a general miscalibration of beliefs; overconfident managers have a narrower confidence interval. By analyzing the narrowness of the confidence interval managers were classified as being overconfident or not.
options. These studies measure overconfidence by looking at CEO holding options behind rational thresholds. CEOs are under-diversified: their human capital is invested in their firm and they often receive stock and option based remuneration. Rational managers should exercise in-the-money options or sell company stock on a pre-committed schedule. However, overconfident CEOs maintain their stock options until the expiration date and are exposed to a high level of risk because they believe that under their leadership the company’s stock will perform better.
Another measure is based on outsider perception by looking at press portrays (Malmendier and Tate, 2005b and 2008 and Hayward and Hambrick, 1997). Evaluating the description of CEOs by words related to confident or unconfident are used to classify the CEOs as overconfident or
not-overconfident.
Table 1
Overconfidence proxies
This table displays proxies for managerial overconfidence that were used in previous studies. The table names the study and proxy that was used to measure overconfidence.
Study Overconfidence proxy
Hayward and Hambrick -Recent performance acquiring firm
(1997) -CEO media portrays
-CEO self-importance
Malmendier and Tate -Timing of CEO option exercises
(2005a) -Acquisition of company stock by CEO
Malmendier and Tate (2005b and 2008)
-CEOs’ personal portfolio transaction in their companies’ stock and options
-Press portrays of the CEO
Ben-David et al (2007) -Miscalibration of CFOs forecasts compared to the stock market forecasts
Billet and Qian (2007) -Acquisition frequency Doukas and Petmezas (2007) -Acquisition frequency
-Insider dealings
Brettel (2008) -Level of overconfidence is obtained from answers to a questionnaire
Croci et al (2010) -Executive stock options
appropriate measure of overconfidence. This reasoning is supported by the findings of Malmendier and Tate (2005a) that overconfident managers undertake more mergers and acquisitions. Doukas and Petmezas (2007) qualify a manager as being overconfident if he acquired five or more targets in a three-year period. They find evidence that managers are not able to evaluate acquisitions that occur in quick succession. Managers often experience over-exuberance when they acquire, and this makes them ignore inference from past acquisitions. According to the authors, managers that acquire multiple companies over a short period have been consistently overconfident about the companies’ prospects. Billet and Qian (2007) define acquirers as frequent if they acquire at least two targets within a period of five years. They argue that a five year time-span is sufficient to allow an acquisition history to develop, but short enough so that historic acquisitions are likely to be informative.
Another overconfidence proxy used by Hayward and Hambrick (1997) is recent organizational success. Recent success is measured as the stockholder return in the twelve months before the acquisition was announced. They argue that CEOs whose company recently was successful receive favorable attributions by and the commitment of people from inside the organization. This leads managers to increasingly believe in their own ability. So, the more successful the company has been in the past, the more likely it is the CEO is overconfident.
2.2 Acquisitions
Mergers and acquisitions have had considerable attention from financial research, what resulted in substantial theoretical and empirical evidence. It is not intended to give an exhaustive overview of these findings, therefore I try to discuss the main theories that are relevant for this study.
According to Hayward and Hambrick (1997) there exist three main motives why takeovers take place. The first argument for undertaking mergers and acquisitions is the value that is obtained from synergies. The value of the new combined entity exceeds the value of the sum of the separate parts. The second motive is poor management of the target company. The third explanation stems from the Hubris theory of Roll (1986), successful acquirers may be optimistic and overconfident about deal synergies.
According to Morck, Shleifer and Vishny (1990) some argue that returns for the acquiring companies can be explained by an adverse information problem: companies that finance a takeover with shares release adverse information about the company. This phenomenon is mostly explained by the signaling theory. This theory suggests that how the market reacts on an acquisition announcement is not completely attributable to the value of the acquisition; the market reaction could also be a reaction on other information about the bidding firm that was signaled with the announcement.
2.3 Market valuation
There are two general classes of theory that try to explain market valuation periods and the effect on mergers and acquisitions; a neoclassical and a behavioral theory. Both are discussed briefly, and I explicate on which theory I elaborate with this study.
According to Harford (2005) the neoclassical argument of market valuation and M&A waves relies on an economic disturbance that causes industry reorganization. The collective reaction of companies on a shock leads to a reallocation of assets through mergers and acquisitions. An example of an economic disturbance that can cause industry- or market reorganization is technological shock. Another possible disturbance is a liquidity shock that causes capital reallocation and leads to an increased number of mergers and acquisitions.
2.4 The interaction of market valuation and managerial overconfidence
From the two discussed theories on the impact of market valuation on M&A activities, I elaborate on the behavioral one in the remainder of this study. I focus on the behavioral explanation because it is in line with the theoretical background of managerial overconfidence, which also is rooted in the behavioral paradigm. In the case of overconfidence, managers behave irrational because they are too confident about future events and their own knowledge and skills. This irrational behavior is also present when managers time the market and make acquisitions when their companies’ stock is overvalued. During high valuation periods managers are likely to be optimistic, like investors.
Because overconfident managers are too optimistic about the prospects of the firm and the decisions they make, it is likely that this overconfidence is reinforced when market valuation is high. Referring to the underinvestment/overinvestment trade-off model by Heaton (2002), it is likely that optimism inclines managers to upward biased cash flow forecasts, causing managers to overvalue projects and ultimately overinvestment by making acquisitions. According to Croci, Petmezas and Vagenas-Nanos (2010) managers overestimate the synergies from deals during high market valuation periods and this negatively influences the quality of deals. Managers have an incentive to make acquisitions in high-valuation periods because it is likely that even bad deals receive favorable market reactions, as a consequence of investor optimism. In low-valuation markets investors will recognize bad deals more easily, resulting in negative announcement returns.
2.5 Empirical evidence
After having discussed some of the theory behind overconfidence and market-valuation, and a brief overview of the merger and acquisition literature, now the focus will be on the empirical findings.
2.5.1 The wealth effects of acquisitions
Travlos (1987) demonstrated that the way in which a bidder pays affects the market reaction on the bidding announcement. He finds that acquisitions that are financed with shares have a stronger negative market reaction than acquisitions financed with cash. This finding is often attributed to the signaling theory. When managers think their companies’ shares are overvalued, they prefer to acquire a target with shares signaling to the market the shares are overvalued. On the other hand, cash-financed acquisitions signal to the market the shares are undervalued.
Another empirical finding is the size effect, as showed by Moeller, Schlingemann and Stulz (2004). It turns out that the announcement return is on average two percentage points higher for small acquirers, in comparison to large acquirers. Large firms offer larger acquisition premiums and engage in takeovers with negative dollar synergy gains. This evidence suggests that hubris is more profound in the acquisition decisions of large firms.
Fuller, Netter and Stegemoller (2002) show that the target status is a significant factor in explaining announcement returns. They find significant positive abnormal returns in the period after the
announcement when the target is a private firm, and find negative returns when the target is a public firm. Acquisition returns also seem to be influenced by the fact whether a target is a national or foreign firm. Cross-border acquisitions generate lower returns than national acquisitions. Morck, Shleifer and Vishny (1990) present evidence that some types of bidders systematically overpay, and they relate this to managerial objectives that may impact acquisition decisions. They reason that managers may not only consider the market value of the firm when making an
acquisition, but also consider personal benefits. When an acquisition provides large personal benefits to a manager they may overpay for the target.
2.5.2 Managerial overconfidence and acquisitions
Table 2
Overview of previous studies on the effect of overconfidence on acquisitions
This table gives an overview of the previous studies that were conducted on the effect of managerial overconfidence on acquisitions. The authors, sample-size and period, country and the event window are given. The last three columns expose the cumulative abnormal returns for the bidder when the CEO is (i) not-overconfident, (ii) the CEO is overconfident and (iii) the difference between CARs for overconfident and not-overconfident CEOs. (***, ** and * means significant at the 1, 5 and
10% level).
Study Period sample
size Country Event window CAR Non- overconfident CEO (i) CAR Overconfident CEO (ii) CAR Difference (iii) Billet and Qian
(2007) 1985-2002 3537 US -1,+1 -0.12% -1.51% 1,39%*** Doukas and Petmezas (2007) 1980-2004 3844 UK -2,+2 1.34% 0.79% 0,55% *** Malmendier and Tate (2008) 1980-1994 477 US -1,+1 -0.12% -0.90% 0,78% ** Croci et al (2010) 1990-2005 848 UK -2,+2 1.26% 0.16% 1,10% **
Another study that linked overconfidence with acquisition frequency is Billet and Qian (2007). Acquirers are defined as frequent if they acquire at least two targets within a period of five years. They find insignificant mean abnormal return for first deals while that for high-order deals is – 1.51% and significant, the difference between first and high-order deals is significant, 1.39%. From this they conclude that value destroying high-order acquisitions are motivated by previous acquisition
experience, and overconfidence affects managerial acquisition decisions.
Malmendier and Tate (2008) found that overconfident managers overpay for target companies and undertake value-destroying mergers. The likelihood that an overconfident manager undertakes an acquisition is 65% higher, if compared with a not-overconfident manager. If the merger is
diversifying, and is funded with internal financing, the effect of overconfidence is largest. Another implication of the study is that the market reaction is significantly more negative for a merger undertaken by an overconfident CEO than when undertaken by a not-overconfident CEO, the difference is a significant 0.78% in three days.
Croci, Petmezas and Vagenas-Nanos (2010) find a significant mean difference (1.10%) in abnormal returns for acquisitions undertaken by overconfident and not-overconfident managers. As a proxy for managerial overconfidence they use a measure based on stock options. This difference is largest in periods of low stock market valuation, indicating that the impact of overconfidence is particularly relevant in these periods. The results also show that not-overconfident managers are not sensitive to the stock market conditions, only overconfident managers are.
2.5.3 Market valuation and acquisitions
2.6 Research question and Hypotheses
From the discussion of the theories on overconfidence and market valuation, and in line with the existing empirical evidence the main research question is: Are acquirer returns affected by managerial overconfidence and market-valuation?
From this research question two testable hypotheses are derived.
H1.0: Acquisition announcement returns are not affected by managerial overconfidence.
H1.1: Managerial overconfidence has a negative influence on acquisition announcement returns.
H2.0: Acquisition announcement returns do not differ between high and low market valuation periods.
3. Methodology
This section discusses the methodology that is used to test the hypotheses. The effect that
overconfidence and market-valuation have on short run acquisition returns is tested with an event study methodology, and a multivariate regression model in which I control for other factors that may influence the acquirer returns.
3.1 Event study
For this research, event study methodology is used to test the market reaction of an acquisition announcement. The daily returns around the announcement date will be analyzed.
The abnormal returns are computed by using the market-adjusted return model, this model compares the actual returns with the market index returns. The market adjusted model is used instead of the OLS market model, because it is likely that the estimated beta would be influenced by acquisition announcements in the estimation period. This is the case because many bidders are multiple acquirers. According to Fuller, Netter and Stegemoller (2002) if this is the case, the beta estimation becomes less meaningful and the market-adjusted model is an appropriate way to measure abnormal returns. Furthermore, Brown and Warner (1980) prove that for short-window event studies, the OLS market model does not significantly improve estimation.
The announcement day (t=0) is the date the acquisition is reported in the Zephyr database.
The abnormal return is computed as follows:
ARi,t=Ri,t-Rm,t (1)
Where ARi,t is the abnormal return of company i on day t, Ri,t is the actual return of company i on day t, and Rm,t is the return of the market index on day t. The S&P Europe 350 index is used to calculate the market returns. This is a value-weighted index that covers 17 major European markets. Total returns are used for the index as well as the companies’ stock returns. Calculations are based on trading days and closing prices.
AARt = 1/N Σ ARit (2)
To aggregate abnormal returns through time for an individual company the cumulative abnormal return (CAR) is calculated. The CAR of company i for the period t1 to t2 is.
CARi (t1,t2) = Σ ARit (3)
The average cumulative abnormal return (ACAR) for the period t1 to t2 is.
ACAR(t1,t2) = Σ AARt (4)
The AARs and ACARs are first calculated for a period of 40 days around the acquisition
announcement, as it can be seen from table A1 in the appendix. Based on these findings I choose to test the hypotheses making use of an five-day event window (-2,+2). Because this period captures all days with significant abnormal returns directly around the acquisition announcement, and almost has the same significant t-value as compared to a three-day event window (-1,+1).
To test the hypotheses the five-day ACARs of the different sub-groups (overconfident and not overconfident, high- and low valuation periods) are compared to see if they differ significantly. Because the variance of the means is unequal between the sub-groups, a t-test is used that assumes unequal variances.
3.2 Constructing overconfidence proxies
From the described overconfidence proxies in section 1.1.3 some are more appropriate for this study than others, here I will explain which proxies I use and how they are constructed.
Acquisition frequency
plausible measure of managerial overconfidence. The proxy is consistent with the argument of Heaton (2002) that overconfident managers undertake more projects, and with the finding of Malmendier and Tate (2008) that overconfident CEOs undertake more mergers. According to Billet and Qian (2007) it can be argued that acquisition frequency is driven by firms’ investment
opportunities. However, they show that investment opportunities fail to entirely explain acquisition frequency, and that managerial overconfidence plays an important role in this.
Doukas and Petmezas (2007) qualify a manager as overconfident if he acquires at least five targets in three years. Billet and Qian (2007) qualify a manager as overconfident if he acquires at least two targets in five years. For this study I use a five-year period because Billet and Qian (2007) argue that a five-year period is sufficient to allow for an acquisition history to develop, but short enough so that previous acquisition are informative. However, in my opinion two acquisitions is inadequate to qualify a manager as overconfident, so I use a minimum of three acquisitions. A CEO is qualified as being overconfident if he has undertaken three or more acquisitions in a period of five years, and not-overconfident if the CEO undertakes one or two acquisitions. This definition is also relevant for the sample I study. From table 4 in the data section it follows that there is a large portion of firms that acquire one or two targets, and there are few firms that acquire five or more targets. The definition of Doukas and Petmezas would result in an insufficient number of overconfident managers to test for statistical significance. For all acquisitions announced, it is analyzed who the CEO is that undertakes the acquisition. This to sort out cases in which a company announces multiple
acquisitions, but these acquisitions are undertaken by different CEOs.
Recent bidder performance
Hayward and Hambrick (1997) argue it is likely that the success of a company reinforces the
confidence of its managers. They become self-servant instead of self-critical, and expectations about their own capabilities will further rise. So, for successful companies it is more likely the CEO is overconfident, and as a result, will undertake more acquisitions. Recent bidder performance is measured as the shareholder return for the one-year period before the acquisition is announced. This proxy is based on the performance of the firms stock, relatively to the performance of the other firms in the sample. Because, when a company performs well relatively to the market, it likely reinforces the overconfidence of the CEO .
is below the median.
3.3 Measuring market valuation periods
Market valuation is defined as the valuation of the market as a whole. The market is divided in high, low, and neutral valuation periods, the methodology to do so is based on Bouwman, Fuller and Nain (2009). The classification of the market as high, low, or neutral is based on the price/earnings (P/E) ratio of the European market index for each month1. Like the P/E ratio of a company is used to measure the over- or undervaluation of the company’s stock, here the P/E ratio of the market index is used to measure market over- or undervaluation. The P/E ratio of the market index is the weighted average of the P/E ratio of the individual stocks. In this way the impact individual constituents have on the market index depends on their relative size. The earnings are based on the last 52-week reported earnings, the price is based on the closing price of last day of the month. The P/E ratios for each month are ranked top down, and divided in three groups of equal size. The highest third is referred to as high market valuation period, the lowest third as low market valuation period, and the middle group as neutral market valuation period. The acquisitions that are announced in these months are then named high, low, or neutral market acquisitions, according to the classification of the month.
3.4 Multivariate analysis
A multivariate analysis is conducted to analyze cross-sectional regression of the returns and the variables that influence the CARs. The OLS model is used to see how the abnormal returns are effected by firm- and deal characteristics and the proxies to measure overconfidence and market valuation periods. In section 1.5.1 I discussed some variables that are likely to influence acquirer returns, these are included in the regression. The dependent variable in this OLS model is the five-day CAR. The independent variables are:
• Recent bidder performance. This is a dummy variable for the overconfidence proxy. It takes the value one if the acquisition is undertaken by an overconfident CEO, and zero otherwise. A CEO is qualified as overconfident if the return of the companies’ stock is above the median of the sample returns, and a CEO is qualified not-overconfident if the companies’ return is
1 Bouwman, Fuller and Nain (2009) find that alternative classifications for market valuation periods, i.e. index
below the median of the sample return. From hypothesis 1 it follows this dummy variable is expected to have a negative influence on CAR.
• Acquisition frequency. This is a dummy variable for the overconfidence proxy. It takes the value one for overconfident acquisitions, and zero for not-overconfident acquisitions. Overconfident acquisitions are undertaken by managers that acquire three or more targets within a five-year period, not-overconfident acquisitions are undertaken by managers that acquire less than three targets in a five-year period. From hypothesis 1 it follows this dummy variable is expected to have a negative influence on CAR.
• High market valuation period. Dummy variable is one if the acquisition is announced during a high-valuation period, and zero otherwise. For each month the P/E ratio of the market index is calculated. The months that belong to the highest third are classified as high market valuation periods. From hypothesis 2 it follows that this dummy is expected to have a positive effect on CAR.
• Low market valuation period. Dummy variable is one if the acquisition is announced during a low-valuation period. For each month the P/E ratio of the market index is calculated. The months that belong to the lowest third are classified as low market valuation periods. From hypothesis 2 it follows that this dummy is expected to have a negative effect on CAR. • Private or public target. The dummy variable takes the value one if the target is public, and
zero for a private target. Fuller, Netter and Stegemoller (2002) show that the acquisition of a private firm has a relative positive effect on CAR when compared to the acquisition of public targets.
• Cross-border or national acquisition. If the acquisition is national the dummy variable is zero, if it is border the variable is one. Fuller, Netter and Stegemoller (2002) find that cross-border acquisitions generate lower returns than national acquisitions.
• Cash. The dummy variable is one if the acquisition is entirely financed with cash, and is zero otherwise. Travlos (1987) find a positive influence of cash financed acquisitions on the acquirer returns.
• Shares. The dummy variable is one if the acquisition is entirely financed with shares, and is zero otherwise. Travlos (1987) find a negative influence of stock-financed acquisitions on the acquirer returns.
4. Data
4.1 Sample selection
I collect from the Zephyr database a list of successful acquisitions of Euro zone companies, with initial bids announced between 1/1/1999 and 31/12/2009. To be included in the sample, the following conditions must be satisfied:
1. The focus of this study is on the Euro zone. The euro was introduced to world financial markets on 1 January 1999 as an official accounting currency. For this study I analyze companies from countries that have been participating in the Euro since the launch in 1999. The twelve countries that participate in the Euro since the launch are: Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, The Netherlands, Portugal, and Spain.
2. The deal has to be completed at the moment this research is imposed.
3. Only acquisitions are being analyzed. The final stake after the acquisition has at least to be 50 percent. The initial stake, before the acquisition is announced, has to be not more than 1 percent.
4. The value of the deal has to be at least 10 million Euro. This minimum is set to exclude minor deals that could have a marginal impact on the acquiring company.
5. Financial and insurance, and utility acquirers and targets are excluded from the sample because those companies are regulated, this regulation could impact managerial biases. This selection is based on the NACE rev. 2 code of the European Union, that lists the different industry classifications.
6. The acquiring company has to be quoted; in order to analyze the market returns of the acquirer it must be possible to collect the return data.
4.2 Managerial overconfidence proxy data
Recent bidder performance
Table 3 shows the statistics of the recent bidder performance, which is used to measure overconfidence, and is calculated as the shareholder return for the one-year period before the acquisition has been announced. The median return of all companies in the sample is 7.8%, acquirers with a return above (below) the median are labeled as overconfident (not-overconfident). This overconfidence proxy results in 342 overconfident and 342 not-overconfident acquisitions.
Table 3
Descriptive statistics recent bidder performance proxy
This table contains the descriptive statistics of the bidder recent performance, which is used to measure overconfidence. For each acquirer the return in the one-year period before the acquisition announcement is calculated. From these returns
the median is calculated. A CEO is qualified as overconfident if the return of the companies’ stock in the year before the announcement is above the median of all one-year-before-announcement returns. A CEO is qualified not-overconfident if
the companies’ stock return is below the median.
Mean 14.38 Median 7.80 Maximum 332.91 Minimum -81.72 St. dev. 44.75 Acquisition frequency
Table 4 shows that 374 companies were responsible for 667 acquisitions. For this study, CEOs who acquire one or two targets in a five-year period are classified as not-overconfident, CEOs who acquire at least three targets in a five-year period are classified as overconfident. 311 of the acquirers are CEOs that acquire one or two targets, and are classified as not-overconfident. This means that 394 acquisitions are made by not-overconfident CEOs. 63 CEOs acquired at least three targets in a five-year period, and are classified as overconfident. These 63 overconfident CEO are together
Table 4
Acquisition frequency bidders
This table shows how the acquisitions are divided between the CEOs in the sample. The first column lists the number of acquisitions made by CEOs, and column 2 how often these acquisition frequencies occur. Column three is the product of the
former two columns. CEOs that acquire one or two targets are classified as not-overconfident, CEOs that acquire at least three targets in five year are classified as overconfident. The total of (not) overconfident acquisitions is the sum of all acquisitions made by (not) overconfident CEOs. The total number of 667 acquisitions is different from the sample size of
684 because for the remaining acquisitions it was uncertain who the CEO of the bidding company was at the time the acquisition was announced. The acquisitions for which the CEO is unknown are left outside the analysis that use acquisition
frequency to proxy managerial overconfidence.
Acquisition frequency Number of acquirers Number of acquisitions
1 (Not-overconfident) 228 228 2 (Not-overconfident) 83 166 3 (Overconfident) 29 87 4 (Overconfident) 17 68 5 (Overconfident) 5 25 6 (Overconfident) 3 18 7 (Overconfident) 5 35 8 (Overconfident) 1 8 9 (Overconfident) 0 0 10 (Overconfident) 2 20 11 (Overconfident) 0 0 12 (Overconfident) 1 12 Total Not-overconfident 311 394 Total Overconfident 63 273
Total all acquirers 374 667
4.3 Control variable data
Table 5 lists data from all control variables, these data are used in the regression analysis to control for deal- and company-characteristics. Panel A shows that only 99 of the 684 targets are public companies, 585 are private companies. From panel B it follows that the majority of the acquisitions are cross-border (467), the remaining 217 acquisitions are domestic. For 296 out of the 684
Table 5 Control variable data
This table displays the data for variables that control for deal- and company-characteristics. In panel A the acquisitions in the sample are divided regarding the status of the target as private or public companies. Panel B splits the sample in cross-border and domestic acquisitions. When a company acquirers a target with a different nationality the acquisition is counted
as cross-border, when the target has the same nationality as the acquirer the acquisition is domestic. Panel C shows the payment method of the acquirer. The acquisitions for which the payment method is known, are split between acquisitions that are entirely paid with cash and acquisitions that are entirely paid with shares. In panel D the descriptive statistics of the
acquirers market capitalization can be found. The market capitalization is measured at last available value before the acquisition announcement.
Panel A: Target status Number of acquisitions As percentage of total acquisitions
Private 99 14.47
Public 585 85.53
Total 684 100
Panel B: Domestic/Cross-border Number of acquisitions As percentage of total acquisitions
Domestic 217 31.73
Cross-border 467 68.27
Total 684 100
Panel C: Payment method Number of acquisitions As percentage of total acquisitions
Cash 317 46.35
Shares 71 10.38
Unknown/Other 296 43,27
Total 684 100
Panel D: Market capitalization acquirer Value in thousand Euro
Mean 5,272,798
Median 1,075,160
Maximum 104,460,683
Minimum 7,380
Standard Deviation 12,251,487
over seven million Euro, while the largest acquirer has a market capitalization exceeding 100 billion Euro. The market capitalization is measured as the last available value before the acquisition announcement.
4.4 Deal statistics
Table 6 presents some financial statistics from the deals. The mean of the deal values is about 453 million Euro, in combination with figure 1 it seems that this mean is upwardly affected by a relative small number of 56 deals that have a value of more than a billion Euro. Most acquisitions have a deal value ranging from 10 to 500 million Euro, however the largest acquisition has a value over 41 billion Euro.
Table 6 Figure 1
Descriptive statistics of the deal values Graphical presentation of deal values
Mean 453,264,560
5. Results
5.1 CEO overconfidence and acquisition announcement returns
Table A1 in the appendix lists the average abnormal returns (AAR) and the average cumulative abnormal returns (ACAR) in the 41-day period around the acquisition announcement for the entire sample. On the announcement date an average abnormal returns is found of 0.68%, significant at the 1% level. Also for the days -2, +1 and +2 significant positive returns are found. For the other days (except t=-13) no significant abnormal returns are found. Over the duration of a five-day period around the acquisition announcement an average cumulative abnormal return is measured of 1.8% (significant at the 5% level). This means that on average acquirers’ market value increases with 1.8% in the five days around the announcement. From the empirical evidence it followed acquisition announcement can lead to negative as well as positive abnormal returns. The positive acquirer ACAR is in line with studies that focus on the effect of managerial overconfidence on acquisitions, that also find positive CARs (Doukas and Petmezas, 2007 and, Croci, Petmezas and Vagenas-Nanos, 2010).
Table 7 lists the average cumulative abnormal returns in the event period (-2,+2). Panel A shows the effect of managerial overconfidence, measured by the recent bidder performance proxy. The sample is split in an overconfident group with firms that performed better than the median performance of all companies in the sample, and a not-overconfident group with firms that performed worse as compared to the median performance of all companies in the sample.
Panel A shows the ACAR is moderately higher for the low performance group (2.181%) than for the high performance group (1.418), the difference between the groups is 0.762%, and insignificant. It was argued that CEOs whose company has recently been successful receive favorable attribution by and the commitment of people from inside the organization, this increases the likelihood the CEO will develop a feeling of overconfidence. The results seem to support this idea but no conclusion can be drawn, due to the absence of significant evidence.
Table 7
The effect of managerial overconfidence on average cumulative abnormal returns.
This table shows the average cumulative abnormal returns (ACARs) in the five-day period (-2,+2) surrounding the acquisition announcement by managerial overconfidence. Abnormal returns are calculated using the market-adjusted
model: ARi,t = Ri,t-Rm,t
Where ARi,t is the abnormal return of company i on day t, Ri,t is the actual return of company i on day t, and Rm,t is the
returns of the S&P Europe 350 index on day t. Average cumulative abnormal returns are calculated as: ACAR(-2,+2) = Σ AARt
Where AARt is the average abnormal return on day t.
In panel A managerial overconfidence is measured by using the recent bidder performance proxy. Bidder recent performance is calculated as the one-year buy-and-hold return of the companies’ stock. For each acquirer the return in the
one-year period before the acquisition announcement is calculated. From these returns the median is calculated. A CEO is qualified as overconfident if the return of the companies’ stock in the year before the announcement is above the median of all one-year-before-announcement returns. A CEO is qualified not-overconfident if the companies’ stock return is below
the median. In panel B acquisition frequency is used to measure managerial overconfidence. CEOs are labeled as overconfident if they acquire at least three targets in a five-year period, and not-overconfident if they acquire one or two
targets. The right column contains the difference in ACARs between acquisitions made by overconfident and not-overconfident CEOs. The ACARs are in percentage and the t-values are reported in parentheses. Statistical significance at
the 1%, 5% and 10% level is indicated by ***, **, and *, respectively.
Panel A: recent bidder performance
All acquirers Not-overconfident (low performance) Overconfident (High performance) Difference ACAR (-2,+2) 1.800** (2.423) 2.181** (2.467) 1.418** (1.895) 0.762 (0.886)
Panel B: acquisition frequency
All acquirers Not-overconfident (one or two acquisitions)
Overconfident (more than two acquisitions) Difference ACAR (-2,+2) 1.800** (2.423) 2.749*** (3.182) 0.603 (1.124) 2.146* (1.778)
CEOs. The difference in average cumulative abnormal returns between the subgroups is a marginally significant 2.146%. From this it can be concluded that the ACARs for bidders whose CEO acquirers more than three targets in a five-year period are lower than for bidders with a CEO that acquirers less targets . This supports the idea that overconfident CEOs overestimate their ability to generate superior returns. Managers that undertake multiple acquisitions in a short period might signal to the market that they are overconfident about outcomes of future successive acquisitions and that they pay too much for the target, this leads to negative market reactions. There are other possible explanations why multiple acquisitions in a short period leads to negative market reactions. Markets might expect that multiple acquirers are unable to realize synergy gains because they fail to
efficiently integrate subsequent acquisitions in a short period. Using acquisition frequency to measure CEO overconfidence, the conclusion can be drawn that overconfidence has a negative wealth effect on acquisitions being announced. This means that not-overconfident CEOs outperform overconfident CEOs when announcing an acquisition, the market seems to react more favorably on an acquisition undertaken by an not-overconfident CEO as compared to an acquisition that is made by an overconfident CEO.
5.2 Market valuation periods and acquisition announcement returns.
What effect market valuation periods have on acquisition announcement returns can be seen in table 8. In line with the results of Bouwman, Fuller and Nain (2009) and Croci, Petmezas and
Vagenas-Nanos (2010) acquisition announcement returns in high market valuation periods are higher than in low market valuation periods. Over the five-day event period, the difference between
acquisitions announced in high-valuation (3.638%) and low-valuation periods (0.831%) is 2.535%. This difference is statistically significant, indicating that the hypothesis that acquisition returns are higher in high market valuation periods can be supported. The market seems to react more
Table 8
Announcement returns in high- and low market valuation periods.
This table shows the average cumulative abnormal returns (ACARs) in the five-day period (-2,+2) surrounding the acquisition announcement by market valuation periods. Abnormal returns are calculated using the market-adjusted model:
ARi,t = Ri,t-Rm,t
Where ARi,t is the abnormal return of company i on day t, Ri,t is the actual return of company i on day t, and Rm,t is the
returns of the S&P Europe 350 index on day t. Average cumulative abnormal returns are calculated as: ACAR(-2,+2) = Σ AARt
Where AARt is the average abnormal return on day t.
Acquisitions are divided in high- and low valuation groups, according to the valuation of the market. The market valuation is computed with the P/E ratio of the market index. The P/E ratios for each month are ranked top down, and divided in three
groups of equal size. The highest third is referred to as high market valuation period, the lowest third as low market valuation period. The right column contains the difference in ACARs between acquisitions made in high- and low-valuation
periods. The ACARs are in percentage and the t-values are reported in parentheses. Statistical significance at the 1%, 5% and 10% level is indicated by ***, **, and *, respectively.
All acquirers High market valuation Low market valuation Difference ACAR (-2,+2) 1.800** (2.423) 3.365*** (3.238) 0.831 (0.996) 2.535** (2.281)
5.3 Combined effect of CEO overconfidence and market valuation periods.
Table 9
The effect of market valuation, CEO overconfidence, and interaction on the average cumulative abnormal returns in the event period.
The effect of CEO overconfidence and market valuation on the average cumulative abnormal returns in the event period (-2,+2), and the joined effect are given in this table. Abnormal returns are calculated using the market-adjusted model:
ARi,t=Ri,t-Rm,t
Where ARi,t is the abnormal return of company i on day t, Ri,t is the actual return of company i on day t, and Rm,t is the
returns of the S&P Europe 350 index on day t. Average cumulative abnormal returns are calculated as: ACAR(-2,+2) = Σ AARt
Where AARt is the average abnormal return on day t.
In panel A managerial overconfidence is measured by using the recent bidder performance proxy. Bidders recent performance is calculated as the one-year buy-and-hold returns of the companies’ stock. For each acquirer the return in the
one-year period before the acquisition announcement is calculated. From these returns the median is calculated. A CEO is qualified as overconfident if the return of the companies’ stock in the year before the announcement is above the median of all one-year-before-announcement returns. A CEO is qualified not-overconfident if the companies’ stock return is below
the median. In panel B acquisition frequency is used to measure managerial overconfidence. CEOs are labeled as overconfident if they acquire at least three targets in a five-year period, and not-overconfident if they acquire one or two targets. The market valuation is computed with the P/E ratio of the market index. The P/E ratios for each month are ranked
top down, and divided in three groups of equal size. The highest third is referred to as high market valuation period, the lowest third as low market valuation period. The lowest row represents the difference between high- and low-valuation periods, the last column represents the difference between not-overconfident and overconfident CEO’s. The cell in the right-low corner contains the difference between not-overconfident CEO’s in high-valuation periods and overconfident CEO’s in low-valuation periods. The ACARs are in percentage and the t-values are reported in parentheses. Statistical
significance at the 1%, 5% and 10% level is indicated by ***, **, and *, respectively. Panel A: recent bidder performance
CEO overconfidence
Market valuation Not-overconfident Overconfident Difference
Panel B: acquisition frequency
CEO overconfidence
Market valuation Not-overconfident Overconfident Difference
High 4.733*** 1.480 3.253** (4.593) (1.245) (2.184) Low 1.581** -0.149 1.703 (1.994) (-0.236) (1,584) Difference 3.152** 1.630 4.883*** (2.369) (1.371) (2.678)
returns and overconfident acquisitions in low-valuation markets the smallest returns. In panel A of table 9 recent bidder performance is used to measure CEO overconfidence, in panel B acquisition frequency is used to measure overconfidence.
Panel A shows that average cumulative abnormal returns are higher in high-valuation periods than in low-valuation periods. The difference between high-valuation and low-valuation periods is more pronounced for not-overconfident CEOs, a significant 3.639% difference, than for overconfident CEOs (an insignificant 1.089% difference). There is no significant difference between not-overconfident and overconfident acquisitions, when measured by bidder recent performance. ACARs for
acquisitions in a high-valuation period by not-overconfident CEOs are significant and 2.733% higher than for acquisitions in low-valuation periods undertaken by overconfident CEOs.
When acquisition frequency is used to measure CEO overconfidence (see panel B), high market valuation periods as compared to low-valuation periods, have a significant positive effect on ACARs for acquisitions undertaken by not-overconfident CEOs (3.152% difference). The difference between high- and low valuation markets is less pronounced for overconfident CEOs. This means that not-overconfident CEOs gain the most when they announce an acquisition during high-valuation markets. For overconfident CEOs the decision when to announce the acquisition, seems to have less impact on the returns. This finding is contrary to the finding of Croci, Petmezas and Vagenas-Nanos (2010) who find that market valuation has more impact on overconfident CEOs. ACARs of acquisitions
low market valuation periods there is an insignificant difference of 1.703%. This also indicates that the influence of managerial overconfidence is particularly relevant in high market valuation periods. These findings also imply that infrequent acquirers can benefit from announcing the deal in a high-valuation market, this benefit does not hold for frequent acquirers. There is a significant 4.883% difference between acquisitions in high-valuation periods by not-overconfident CEOs and acquisitions in low-valuation periods by overconfident CEOs. This shows that the interaction of managerial overconfidence and overall stock market valuation is an important factor in explaining acquirer returns.
From these results it becomes clear that the effect valuation periods have on acquisition announcements returns is stronger than the effect that CEO overconfidence has on acquisition announcement returns. This finding becomes also apparent from the separate investigations of overconfidence and market valuation periods. The effect that market valuation periods have on announcement returns is significant. The influence that CEO overconfidence has on announcement returns depends on the overconfidence proxy used. Evidence is found for an interaction effect of overconfidence and market-valuation on acquirer returns.
5.4 Multivariate analyses
0.292 between the acquisition frequency proxy and the acquirer market value. This result seems intuitive since it is plausible to expect that larger firms undertake more acquisitions.
In the regression the dependent variable is the acquirers cumulative abnormal returns (CAR) over the event period (-2,+2). The explaining variables are the proxies for overconfidence and for market valuation. In all regressions I include variables for the deal- and firm characteristics. To measure the overconfidence and market-valuation effects, dummies are included.
Table 10 shows the results of the regressions. From regression (2) it can be seen that the acquisition frequency variable, which is a measure of CEO overconfidence, is significantly negative. The
coefficient of the acquisition frequency proxy is -0.019, meaning that the abnormal return for
frequent acquirers is less than that of not-frequent acquirers. When the acquirer recent performance is used to measure CEO overconfidence (regression 1), this overconfidence effect is less declared.
Table 10 Multivariate analyses
This table presents results from cross-sectional regressions of cumulative abnormal returns on CEO overconfidence and market valuation period, and control variables for deal and organizational characteristics. Abnormal returns are calculated
using the market-adjusted model: ARi,t=Ri,t-Rm,t
Where ARi,t is the abnormal return of company i on day t, Ri,t is the actual return of company i on day t, and Rm,t is the
returns of the S&P Europe 350 index on day t. CARs are calculated for the individual companies as the sum of the abnormal returns in the five day event period (-2,+2), where day 0 is the announcement date. The variables for bidder recent performance and acquisition frequency are included to measure CEO overconfidence. For bidder recent performance, a CEO is qualified as overconfident if the return of the companies’ stock in the year before the announcement is above the median of all one-year-before-announcement returns. A CEO is qualified not-overconfident if the companies’ stock return is
below the median. For the acquisition frequency variable, CEOs are labeled as overconfident if they acquire at least three targets in a five-year period, and not-overconfident if they acquire one or two targets. The high-valuation variable is a dummy and takes the value one if the acquisition is announced during a high valuation period and zero otherwise. The
low-valuation variable is a dummy and takes the value one if the acquisition is announced during a low-low-valuation period and zero otherwise. The market valuation is computed with the P/E ratio of the market index. The P/E ratios for each month are ranked top down, and divided in three groups of equal size. The highest third is referred to as high market valuation period, the lowest third as low market valuation period. The national/cross-border dummy is zero for national acquisitions, and is one for cross-border acquisitions. The Private/public dummy variable takes the value one if the target is public, and zero for
a private target. The cash dummy variable is one if the acquisition is entirely financed with cash, and is zero otherwise. The shares dummy variable is one if the acquisition is entirely financed with shares, and is zero otherwise. The acquirer market value variable takes the value of the log of the bidder’s stock market capitalization. T-values are reported in parentheses,
Variable (1) (2) (3) (4) (5) (6) (7) Intercept 0.062*** (3.057) 0.043** (2.101) 0.047** (2.273) 0.079*** (3.942) 0.084*** (4.233) 0.064*** (3.144) 0.070*** (3.431) Bidder recent performance -0.001 (-0.350) -0.001 (-0.325) -0.001** (-2.254) -0.001** (-2.162) Acquisition frequency -0.019*** (-3.400) -0.019*** (-3.457) -0.017*** (3.092) -0.017*** (-3.114) High-valuation period 0.018** (2.129) 0.023*** (2.694) 0.020* (1.921) 0.021** (2.495) Low-valuation period -0.021*** (-4.003) -0.024*** (-4.385) -0.021*** (-3.907) -0.023*** (-4.269) National/Cross border -0.007 (-1.108) -0.005 (-0.821) -0.005 (-0.872) -0.007 (-1.133) -0.007 (-1.199) -0.005 (-0.895) -0.006 (-0.957) Public/Private 0.001 (0.007) -0.002 (-0.216) -0.002 (-0.214) 0.001 (0.118) -0.001 (-0.001) -0.001 (-0.161) -0.002 (-0.271) Cash 0.010* (1.825) 0.010* (1.825) 0.009* (1.725) 0.009* (1.780) 0.010)* (1.762) 0.009* (1.693) 0.009* (1.654) Shares -0.018* (-1.838) -0.020** (-2.103) -0.021** (-2.195) -0.019** (-2.092) -0.019** (-2.068) -0.022** (-2.382) -0.022** (-2.379) Log (Market Value) -0.003** (-2.103) -0.001 (-0.781) -0.001 (-0.896) -0.003*** (-2.679) -0.004*** (-2.806) -0.002 (-1.489) -0.002 (-1.614) Adjusted R2 0.019 0.047 0.048 0.079 0.092 0.101 0.113
When both measures of CEO overconfidence are regressed together, only the acquisition frequency has a significant effect on acquirer CARs. From these findings, and the event study, it can be
For the influence of market valuation periods on cumulative abnormal returns, stronger evidence is found (regression 4). High-valuation periods have a significant positive effect on bidders CARs, whereas low-valuation periods have a significant negative effect on bidders CARs. This finding is in line with Bouwman, Fuller and Nain (2009) and Croci, Petmezas and Vagenas-Nanos (2010) that find the same market-valuation effect on acquirer CARs.
When the recent bidder performance proxy and the market valuation variables are implemented together (regression 5), the bidder past performance variable becomes statistically significant, although the economic effect it has on CARs is still very small. Running the acquisition frequency proxy and the market valuation period variables together, results in a significant negative influence of acquisition frequency on CARs. These outcomes are the same when both of the overconfidence variables, together with the market valuation variables are regressed (7). Frequent acquirers abnormal returns are lower as compared to not-frequent acquirers.
From the variables that control for deal and firm characteristics the share-dummy is significant and negative (for all regressions), indicating that deals that are paid with only shares receive less favorable market reaction. Deals that are paid exclusively with cash seem to result in positive CARs, although no significant effect was found. This is in line with the finding from Travlos (1987) that the way in which a bidder pays affects the market reaction on the bidding announcement. Acquisitions that are financed with shares have a stronger negative market reaction than acquisitions financed with cash. This can be explained by the idea that managers that finance acquisitions with shares signal to the market that the companies' shares are overvalued. When the market valuation period variables are included (regressions 3-5), the acquirer market value has a significantly negative effect on cumulative abnormal returns. This confirms the conclusion from Moeller, Schlingemann and Stulz (2004) who find that the announcement return is higher for small acquirers than for large acquirers. This means that the short-time wealth effect of acquisitions is positive for small acquirers in
comparison to large acquirers. The effect of the public/private dummy variable on CARs is small and insignificant, the status of the target has no influence on the acquirer returns. The nationality has more influence on the CARs, although the effect is statistically insignificant.
The predictive power of the regression model rises strongly when the variables for the market valuation periods are added to the regression, in comparison to the case where the CEO
overconfidence variables are included. Indicating that market valuation can better explain acquirer CARs than managerial overconfidence.
6. Conclusion
The main goal of this study is to explain acquirer returns by focusing on managerial behavior. I test the effect of managerial overconfidence on acquisition announcement returns, and examine whether this effect is different in high- and low-valuation markets. Literature suggests that managerial
overconfidence has a negative effect on acquirer returns, and acquisitions in high-valuation markets receive higher short term returns than acquisitions during low-valuation periods. Both hypotheses are tested, making use of event study and regression model methodology. This study focuses on the Euro-zone countries and the sample consists of 684 acquisitions in the period 1999-2009.
Two methods to classify CEOs as overconfidence are operated. The first one is based on the recent bidder performance. The rationale behind this proxy is that for successful companies it is more likely the CEO is overconfident, and as a result, will undertake more acquisitions. The other overconfidence proxy is based on the acquisition frequency of managers. The rationale behind this method of
measuring overconfidence is the idea that undertaking multiple acquisitions in a short period is an poor investment strategy and an appropriate measure of overconfidence. This reasoning is
supported by the findings of Malmendier and Tate (2005a) that overconfident managers undertake more mergers and acquisitions. The classification of the stock-market in high- and low-valuation periods is based on the P/E ratio of the market and is from Bouwman, Fuller and Nain (2009).
The outcome of the event study methodology shows a significant positive average cumulative abnormal return (ACAR) of 1.8% over the five-day event period (-2,+2). From this it can be concluded that on average bidders received a positive acquisition announcement return, acquisitions lead to a short-time wealth increase for bidding firms. The first overconfidence proxy, recent bidder
performance, has no significant influence on acquirer returns. Although acquirers in the high-performance group receive favorable returns, the difference with the low-high-performance group is insignificant. Operating the second overconfidence proxy, CEO acquisition frequency, results in a significant 2.146% higher ACAR for firms that acquire one or two targets in comparison to higher-order acquirers (at least three acquisitions in a five-year period). So, when acquisition frequency is used to measure CEO overconfidence, the conclusion can be drawn that overconfidence has a negative wealth effect on acquisitions being announced.
