Master Thesis Finance
The Influence of Overconfidence and Self-Attribution on
Acquisition Performance:
Evidence from Dutch Publicly Listed Firms
Lennart Klomp – S2453029 Supervisor: Dr. R.O.S. (Raymond) Zaal
2
The Influence of Overconfidence and Self-Attribution on
Acquisition Performance:
Evidence from Dutch Publicly Listed Firms
Lennart Klomp*
Abstract
This paper examines the influence of overconfidence gained through positive self-attribution on the performance in acquisition by Dutch listed firms for a sample of 368 acquisitions. By using the activeness in acquisitions by management as a proxy for overconfidence we find significant negative difference in announcement returns for acquisitions done by frequent and infrequent buyers but are unable to explain this difference with self-attribution theory by looking at the difference between the first and later acquisitions done by frequent buyers. Also when we control for different motives for doing many acquisitions in a short time span by using the management own firm stock trading we find insignificant results, suggesting overconfidence gained through positive self-attribution cannot explain for bad performance in acquisitions by Dutch listed firms. The explanation we put forward for the found difference compared to existing literature is that Dutch listed firms have a dual board structure and thereby an independent supervisory board that does not commit to the acquisitions. With this paper we give empirical evidence on the claim that more independent board members leads to better investment and acquisition performance and contribute to the research on corporate governance and behavioral biases in acquisitions.
Keywords: Acquisitions; Overconfidence; Self-Attribution; Board Structure; Dutch Firms
JEL Classification: G20, G34
3
1. Introduction
4 For years, mergers and acquisitions have shown to not always create value for their shareholders, seen by a decrease in shareholder value upon announcement coming out of too optimistic synergy gains or too extreme valuations (Moeller et al. 2005). The conventional agency theory, in which rational behavior and the overall tendency toward optimization and maximization from individuals towards efficient markets is assumed (Jensen & Meckling, 1976), fails to explain this acquisition behavior by management. Realizing this might also come out of social psychological biases, Billet and Qian (2008) and Petzemas and Doukas (2008) define CEO overconfidence through self-attribution as frequently being involved in acquisitions in a short time span, which shows that announcement returns for these CEOs are more negative than for deemed rational CEOs in deals made by US and UK firms.
In early work by Fama and Jensen (1983), agency costs were ascribed to departures from efficiency out of moral hazard problems due to the fact that the one bearing the risk (shareholders) are not the ones taking the risk (management). From this separation of ownership and control, the board of directors, through corporate governance, is assigned to ensure that CEOs act in the best interests of the shareholders and restrain them if they do not. Having recognized this, Kind and Twardawski (2016) researched whether the significant negative relation between the announcements returns and CEO overconfidence can be better explained by overconfidence by the board of directors. For CEOs, to obtain overconfidence through self-attribution in the first place, one successful acquisition with approval by the board of directors is needed, which should restrain CEOs if it’s not in the best interest of their shareholders. This research indeed provides evidence that measuring overconfidence through self-attribution for the board of director’s gives distinct, and adds to, documented overconfidence by CEOs.
5 In their first work on investment distortion and CEO overconfidence, Malmerdier and Tate (2005) already conclude that strong and independent boards are needed to restrain CEOs from overinvesting. As Dutch publicly listed firms were bounded by the law to have independent supervisory boards, or since 1 January 2013 to have an independent chairman in case of a one-tier board, this research sets out to find empirical evidence for this claim by asking; Can management overconfidence through self-attribution explain bad performance in merger and acquisition activities by publicly listed firms in the Netherlands?
Seeing the evidence on management overconfidence (Malmerdier & Tate, 2005; 2008; Petzemas & Doukas, 2008; Billet & Qian, 2008) and the emphasis on independence in constraining it (McDonald et al., 2008; Kolanski & Li, 2013; Banjer et al., 2015; Kind & Twardawski, 2016) we are surprised to see that this, to our best knowing, has not been researched in a two-tier board system and contribute with this research to the existing literature on management overconfidence and the influence of independence on this.
By examining overconfidence through self-attribution especially in a dual board structure setting we can empirically research the influence of independence on this and see if it can explain the extensive acquisition activities and possibly lower performance that come with it. By doing this research we want to give insight into the influence of different group mechanics and the behavioral biases. Kind and Twardawski (2016) have already shown that board overconfidence is different and adds to CEO overconfidence, but we would like to see if having an independent supervisory board that has not been committed to past acquisitions can completely restrain actions out of overconfidence. This is why we are especially interested in overconfidence coming out of self-attribution in acquisitions because, in the setting of a dual board structure, the supervisory board can be assumed to not be committed to acquisitions, which is the cause of overconfidence out of self-attribution, while they do have the knowledge to judge if an acquisition would be value-creating and have the power to intervene.
6 overconfidence and self-attribution. As self-attribution theory would further suggest that higher-order deals are met with lower returns when compared to the first deal we also compared these for the sample of overconfident frequent buyers to find no significant difference in the abnormal returns. We argue that this is in favor of the corporate governance literature showing the positive effect of independent board members on acquisition performance. Also when controlling for deal, firm or other possible variables that could explain the abnormal returns in our multivariate OLS regression we get insignificant results. To control if our activeness measure of overconfidence is really measuring overconfidence, not other agency costs like empire-building, we also used a different measure to test this. We used the own firm stock dealings from management, arguing that overconfident managers would increase their own firm stock ownership before an acquisition more when compared to a non-overconfident manager as the first is more confident in his valuation and fails to recognize the higher risk. Also on this measure we get insignificant results leading us to conclude that overconfidence out of self-attribution does not play a role in acquisitions by Dutch listed firms like their US and UK counterparts.
7
2. Literature
2.1 Overconfidence and Self-Attribution
It is well documented by researchers in social psychology literature that individuals exert or are driven by some form of overconfidence. Defining overconfidence as the overestimation of one’s knowledge, abilities and precision of their information or to be overly optimistic about the future and ability to control it (Ackert & Deaves, 2010), it also has shown to have impact on financial decision-making. In contrast to many other personal traits, overconfidence has shown not to be a stable but develop over-time and experience. As individuals gather more information through performing an action they get more confident in their judgement while the accuracy of this judgement reaches a ceiling (Oskamp 1965). As personal experience with this action continues, individuals get disproportionally confident in their own skills, believing it’s their doing in case of success and rule out chance, creating an illusion of control (Langer 1975). This error in judgement can create overly optimistic views about own skill and future prospects, especially during actions or projects where individuals are highly committed; they tend to underestimate chance; thereby create an illusion of being in control; resulting in overconfidence and overly optimistic views (Weinstein 1980). Also, with bringing difficult tasks to a success, which can be assumed to need a certain amount of commitment, individuals especially overestimate their relative performance (Haunschild et al., 1994; Moore & Healy, 2008).
8 From the above theory on overconfidence we would expect managers showing this behavior to have destructive influence on corporate value and policy. Overconfident managers think they are acting in the best interest of the firm but all the while would constantly under-or overestimate the firms performance, and their risk-taking and get overly optimistic about future prospects while having the illusion of being in control of it all. There is no reason to believe that individual managers would restrain from this behavior and indeed it’s been put forward in economic literature to explain deviations from rational behavior, seeing the growing field of behavioral finance.
2.2 Overconfidence and Acquisitions
In acquisition activities, overconfident managers would be overly optimistic about the future prospects, overestimating cash flows and future gains out of acquisitions, resulting in overpaying for targets and destruction of firm value. Roll (1986) was the first to recognize that, despite the extensive literature on corporate takeovers and the evidence on overpaying for targets, we still don’t fully understand why this keeps happening and theorizes on it being driven by some sort of hubris. With the hubris hypothesis, he argues that rational bidders would only pay the market price for takeovers and only irrational bidders would pay more than this, as they believe their own valuation is better than that of the market. A market price by definition is already an average of individual irrational shareholders. This would mean that any takeover above market price, which most are, flows out of irrationality and every acceptance of a bid at or below market price flows out of irrationality of the target firm. This means that market efficiency would predict there are no gains to be made by corporate takeovers and there shouldn’t be a market for corporate control in the first place. The fact that this market does exist is in support of semi-market efficiency, the hubris hypothesis and the overly optimistic views on valuation that come with it.
9 higher announcement returns, are more probable to make acquisitions (more active) at any point in time and even more so in case of abundant funds. The overconfidence proxy used in this research is based on overconfidence resulting from (too) optimistic views about future prospects. The results are in support of the hubris hypothesis, seeing that managers being labelled overconfident are more active in the acquisition market, especially when they don’t have to raise any extra capital for it. By considering personal option holdings by managers as a proxy for overconfidence, acquisitions out of empire-building motives, in which managers don’t regard the well-being of shareholders, are being ruled out. Although delaying the exercise of in-the-money options while increasing the risk is a sign of optimism about the future firm performance, this does not necessarily signal overconfidence in the specific investment or acquisition, more so in the firm’s future prospects overall.
10 Based on the finding of a downwards sloping negative value creation curb for higher-order acquisitions, Petzemas and Doukas (2008) proxy overconfidence for UK management as high-activeness in acquisitions within a short time span. They argue that doing this many acquisitions in such a short time can’t be an outflow of rational behavior as the time between the acquisitions is too short to gain knowledge and experience based on which a next acquisition would be done. They find significant lower wealth effects for this overconfidence proxy compared to less frequent acquirers showing evidence in support of the self-attribution theory explaining bad performance in acquisitions in the UK. Besides also controlling for management own beliefs, by looking at own firm stock trading around the announcement date, they control for endogeneity by looking at overconfident deals and firm specifics explaining the acquisition activeness and value creation out of this. They find no evidence of this driving the more negative announcement returns for the overconfident acquirers concluding that overconfidence through self-attribution is the best explanation for management getting involved in high-order deals despite the lower value creation that comes with it.
The existing literature on management overconfidence and its predicted effect on acquisitions all show this to be present and to be negative (Malmerdier & Tate 2005, 2008; Billet & Qian 2008; Petzemas & Doukas 2008; Kind & Twardawski 2016). In line with this reasoning we expect to see the same for our Dutch sample of acquisition, hypothesizing that;
Hypothesis 1: Overconfident management abnormal stock return upon acquisition announcement is more negative compared to non-overconfident management abnormal stock returns upon acquisition announcement
2.3 Overconfidence and Corporate Governance
11 Findings in social psychology on group decision-making and leadership suggest that individuals who share the same characteristics easily identify with each other in groups and let themselves get influenced by this group in decision-making (Hogg & Terry, 2000). Realizing this, Malmerdier and Tate (2008) and Petzemas and Doukas (2008) show in their research that CEO-duality, when the CEO is also the chairman of the board of directors, and a lack of governance have the power to explain the observed overconfidence, by including a dummy in their regression to explain the wealth effects of multiple acquisitions. In more recent work, Kind and Twardawski (2016) research whether the found overconfidence attributed to CEOs and executive management isn’t just an outflow of overconfidence through self-attribution by individual members of the board of directors, knowing they have the task to confirm key policies or otherwise intervene and the board of directors (most of the time) also consists of executive management in the US board structure. Taking overconfidence as proven when being involved in multiple acquisitions, their finding on CEO overconfidence is in line with prior literature. Using the same overconfidence proxy for individual board members they show board overconfidence to be distinct, and add to, the documented CEO overconfidence provig that they do not always carry out their tasks of controlling executive management incentives and also get caught up in the optimism. Also they find board overconfidence to be significantly positive correlated with acquisition premiums.
12 and dominated by outsiders help to make better acquisition performance by limiting the number of acquisitions made by CEOs compared to less ‘strong’ board CEOs. Looking at the role of corporate governance in restraining overconfident CEOs, Banjer et al. (2015) study the effect of a governance code change. Arguing that adequate controls and independent views are needed by board members to control overconfidence, they compare performance before and after a policy change in the US (SOX) which implements rules requiring more independence and control by boards. They conclude that after the implementation there is; (1) less investments; (2) less risk taken; and (3) less acquisitions made by CEOs, confirming their own argument.
Dutch listed firms are fundamentally different to the firms in the existing researches, done by Billet & Quan (2008); Kind & Twardawski (2016) in the US and Petzemas & Doukas (2008) in the UK, in having an independent supervisory board that’s not involved with the day-to-day management. As for overconfidence to come out of self-attribution, one would first have to commit to the acquisition to get the positive representation of a successful bid (Miller & Ross, 1975). Also, seeing the above literature on independence by board members we expect that the negative difference between overconfident and non-overconfident acquisitions cannot be explained by self-attribution theory for Dutch listed firms, hypothesizing that;
Hypothesis 2: Overconfident management 5th and higher order deals abnormal stock return upon acquisition announcement is not more negative than overconfident management 1st deal abnormal stock return upon acquisition announcement.
13
3. Data and Methodology
In this section we present the data and methodology used to answer the research question and hypotheses. First the conceptual model is presented and the variables are discussed after which the sample construction and the methodology.
3.1 Model Figure 1 Conceptual model Management Overconfidence: Independent variable: Acquisitions Frequent Buyers: Overconfident Infrequent Buyers: Non-overconfident Dependent Variable: Difference cumulative abnormal return Overconfident: Self-Attribution Control Variables 1) Type of payment 2) Type of deal 3) Firm characteristics 4) Market circumstances 5) Time and Industry effects
Focus Variable: Difference CAR 1st and 5th
14 In figure 1 we present our conceptual model. In setting up this model we followed the existing researches done by Petzemas & Doukas (2008) and Billet & Qian (2008) to keep our research outcomes comparable. In-line with the existing researches we measure the effect of overconfidence on acquisitions performance by using the acquisition announcement return as dependent variable comparing the abnormal returns of overconfident to non-overconfident buyers. We also recognize there are other variables that may affect the value creation from acquisition and control for this in answering the research question. We further discuss the variables below.
3.2 Measuring Overconfidence
Overconfidence is a personal trait that develops over time and experience (Oskamp 1965). This means that it’s unique for each individual and therefore subjective. Not having a direct and clear meaning as to what being overconfident is also means that there is no way of measuring overconfidence directly. This is why different proxies to measure overconfidence, based upon findings in social psychological research, were created in the existing literature so far. We follow the existing literature on overconfidence and self-attribution to measure overconfidence and present a proxy based on activeness in acquisitions.
16 out to be labeled as a normal acquisition. We do this to be able to use the maximum amount of data.
Like Fuller et.al. (2002) and Petzemas & Doukas (2008), we use a window of three years as we deem this to be short enough to limit the learning and feedback effect from acquisition to assume overconfidence while controlling for management turnover. Although it can also be argued that doing two acquisitions in an even shorter window (two years for example) can be seen as overconfident, we use a three year window with a threshold of five acquisitions as we are especially interested in the possibility of overconfidence coming out of self-attribution and to keep this comparable to the existing literature.
3.3 Measuring acquisition performance
For measuring the acquisition performance we use the abnormal returns at the time of announcement of the acquisition as given by the Zephyr database, in-line with Billet & Quan (2008) and Petzemas & Doukas (2008). Measuring acquisition performance comes down to showing whether or not a good valuation with good assessment of future earnings and possible synergies has been made. This is why we use the announcement returns as overconfident managers would believe in their own valuation more than non-overconfident managers disregarding the valuation by the market and resulting in lower abnormal returns upon announcement. To create this measure we used the event study methodology which is discussed in the methodology section below.
3.4 Control variables
For the creation of control variables in our model we followed the existing literature on the subject of acquisition and self-attribution (Billet & Quan, 2008; Petzemas & Doukas, 2008; Kind & Twardawski, 2016). The variables chosen have shown to have explanatory power for the market reaction of an acquisition announcement for a firm and we want to control for this in our model.
Type of payment
17 type of acquisition financing can signal the beliefs of the management as managers who are confident that their own firm is overvalued are more likely to finance with stock, while managers who are confident their stock is undervalued are more likely to use any form of cash or debt to finance acquisitions (Myers & Majluf 1984). This way, we expect to see that overconfident managers do more acquisitions with cash resulting in lower abnormal returns as they don’t agree with the market valuation. To use this in our comparison between overconfident and non-overconfident buyers, we divided our sample of acquisitions, like Fuller et.al. (2002) and Petzemas & Doukas (2008), in the groups ‘Cash’, ‘Shares’ and ‘Mixed’ according to the method of payment mentioned in the Zephyr database. Under ‘Cash’ we include the acquisitions mentioned in the Zephyr database with Cash, Cash Assumed, Debt, Debt Assumed, Loan Notes or any combination of these; under ‘Shares’ we include pure share acquisitions mentioned with ‘Shares’ in the Zephyr database and no other; and under the last group ‘Mixed’ all other payments or combination of the two are captured. For the placement into payment groups we have ignored the payment methods Deffered payment and Earn-Out as these only signal that management has agreed to pay extra in the future at a set date or when a certain threshold is reached. To control if the signal given by the different types of payment drive the abnormal returns, we have also included a dummy variable in our regression analysis which take on the value of one for the payment method of cash or shares and otherwise zero.
Type of deal
We also want to control for the different types of deals as they have shown to have predictive power for the market’s reaction. It has been shown that diversified companies are valued lower than their undiversified equals (Lamont & Polk 2001). This way we expect to see lower announcement returns if such a diversifying acquisition is done. To control for this we have added a dummy variable in our multivariate regression model that takes on the value of one if the acquisition was non-diversifying and otherwise zero. It also has been shown that cross-border acquisitions are met with lower acquisition returns by the market (Moeller & Schlingerman, 2005). To also control for this we included a dummy variable in our
18 Firm characteristics
The existing literature on acquisitions has shown firm specific characteristics to have predictive power over the market’s reaction of an acquisition announcement. We want to control for this in our model and for this we follow Petzemas & Doukas (2008). Above we have already shown that we expect that the payments in cash for an acquisition can signal overconfidence. Malmerdier and Tate (2005) also show this in their research concluding that management are more likely to go into firm value destroying policies at any point in time in the case of abundant internal funds. To control for this we also included the amount of cash flow in our model, besides a dummy for cash payments, as we expect firms with more cash flow to do more acquisitions resulting in lower announcement returns. The size of the firm has also shown to have effect on the reaction of the market. Moeller et.al. (2004) show that smaller-sized firms persistently outperform bigger firms in abnormal returns for acquisitions. To control for this we included multiple measure to control for firm size. We included the firm’s total assets, relative size to the value of the deal, the CAPEX and the Tobin’s Q to control for the effect of firm size on the abnormal returns upon acquisition announcement. The last firm specific control variable is the debt capacity, as Bruner (1988) argues this is interchangeable with the amount of cash. This means we expect to see firms with a higher debt capacity to be more active on the acquisition market.
19 turning this into a dummy which takes on the value of one if the firm’s debt capacity is above the overall sample median to make this a relative measure. The cash flow, CAPEX and Tobin’s Q variables are used as given above.
Market circumstances
Besides acquisition and firm-specific predictors we also want to control for market circumstances driving the abnormal returns. As well-performing firms have shown to be better buyers (Haleblain & Finkelstein, 1999), we included the average buyer stock returns over the 6 months preceding the acquisition announcement in our model. We also want to control for economy-wide influence by including the average of the market return 6 months before the announcement. Both variables are taken from the Thomas Reuter Financial Datastream.
It’s also been shown that mergers come in waves (Harford, 2005), so to account for this we include the merger activity defined by the log of one plus the total number of mergers in our sample during the six months before the announcement.
Time and Industry Effects
As last we want to control for any time or industry fixed effects in the market reactions on acquisitions. In line with Kind & Twardawski (2016), we use industry fixed effects, to account for any time-invariant differences across industries, and year fixed effects for any fixed differences in abnormal returns over time. For this we use the announcement date and the major sector as provided by the Zephyr database.
3.5 Sample Selection
20 To create a sample of acquisition deals made by Dutch firms we used the Zephyr database starting from 1 January 1997 till 1 July 2016. This is the total amount of time covered by the Zephyr database giving us the maximum amount of data. Even though firms have been allowed to have a one-tier board structure in the Netherlands since 1 January 20131, we argue that we can still use this data as even with the one-tier board structure the chairman must be a non-executive2 and only a few firms thus far have implemented this system. For an acquisition to be taken in the sample it must comply with the following criteria, in line with Billet & Qain (2008) and Petzemas & Doukas (2008):
1. The acquirer is a Dutch publicly listed firm listed at the Amsterdam Euronext Exchange.
2. The acquirer is covered by the Thomson Financial Datastream Database with 5 days return data around the announcement and a minimum of 50 days return data before the announcement on the Thomson Financial Datastream database.
3. The acquirer purchases at least 50% of the targets shares as result of the takeover. 4. The deal value is at least one million dollars, to avoid results being generated by small
deals.
5. The deal has a status of complete or complete assumed in Zephyr.
These requirements result in a list of 448 deals made by 93 firms from the Zephyr database. In addition to these requirements we also exclude financial industry acquirers, in line with Petzemas & Doukas (2008), as this industry is strongly regulated and has fundamentally different asset structure which might cause biases in our used measures and control variables. Also, many financial firms play a fundamentally different role in the market for corporate control, by being involved as a financer in corporate takeovers but not truly being the one that buys, which may cause even more outside biases. To isolate the effect of an announced acquisition we also remove acquisitions done by the same firm announced within the five-day window around the announcement. An exception to this is when the announcement is made on exactly the same day; in this case the two or more deals are perceived as one deal and the deal values are added for that day. This results in an overall sample of 368 deals made by 72 Dutch publicly listed firms of which summary statistics of the acquisitions and firm-specific data are presented in table 1.
21 Table 1
Summary statistics acquisitions and firms Panel A: Acquisition Data
Type of Acquisitions Number of Deals % Total Deal Value (€ mln) % Average Deal value (€ mln) Median Deal Value (€ mln) All 368 100% 215,235.21 100% 584.88 43,24 Frequent buyers 198 54% 47,528.06 22% 240.04 48,50 Infrequent buyers 170 46% 167,707.14 78% 986.51 39,25 Cash 169 46% 76,026.07 35% 449.86 66,64 Shares 26 7% 94,854.60 44% 3648.25 40,89 Mixed 173 47% 44,354.54 21% 256.38 32,00 Frequent Cash 104 53% 28,652.86 60% 275.51 51,44 Frequent Shares 7 4% 6,795.00 14% 970.71 222,00 Frequent Mixed 87 44% 12,080.21 25% 138.85 33,05 Infrequent Cash 65 38% 32,274.33 19% 496.53 29,87 Infrequent Shares 19 11% 88,059.60 53% 4634.72 17,59 Infrequent Mixed 86 51% 47,373.21 28% 550.85 72,00
Panel B: Firm Data
Variables (€ mln) All Frequent Infrequent
Nr Average Median Nr Average Median Nr Average Median Market Value 367 7,074.66 2,064.02 198 9,771.20 3,434.37 169 3,915.41 725.33 Total Assets 367 6,870.77 1,682.09 198 8,028.29 2,315.50 169 5,514.62 606.49 Fixed Capital 365 1,765.82 248.67 198 2,214.06 599.30 167 1,234.37 93.36 CAPEX 365 378.27 70.50 198 457.90 224.86 167 283.86 25.01 Debt capacity 366 703.95 90.64 198 904.89 287.62 168 467.12 10.72 Cash Flow 359 849.40 223.83 198 1,067.79 417.00 161 580.81 77.39 Tobin's Q 367 5.73 1.21 198 3.2 1.25 169 8.7 1.03
Notes: This table presents the summary statistics for the deal and firm specific for the sample of acquisitions done by Dutch listed firms. Panel A presents the deal specific summary statistics for the total sample, divided by the overconfidence proxy, type of payment and combination of the last 2. Panel B gives the firm specific summary statistics for the complete sample and split by our proxy for overconfidence
22 data of the buying firms also divided by our proxy of overconfidence in frequent and infrequent buyers. Comparing the firm-specific data of frequent and infrequent buyers we notice that frequent buyers can be characterized by large firms with high market value and cash flow.
In appendix B we provided the sample of acquisitions set out in time, acquirer industry and target industry as in major sector as given by the Zephyr database. Panel A of this table, where the acquisitions are set out in time, shows two big clusters of acquisition activities in the years 2000 and 2006 till 2008 and Panel B shows acquisitions were mostly done by firms in the ‘Other services’ industry.
3.6 Methodology
To start the analysis we want to isolate the effect of acquisitions on firm value. To do so we use the event study methodology to calculate the cumulative abnormal return (CAR) around the announcement date as given by the Zephyr database. These are calculated from the abnormal returns for a five-day period (-2, +2) for each of the acquisition announcements. A five-day event window is deemed wide enough as Fuller et.al. (2002) shows it captures the first mention of the merger for a sample of 500 announcements and is later on also used in similar researches (Petzemas & Doukas 2008; Malmerdier & Tate 2008; Kind & Twadawski 2016). We estimate abnormal returns using the market adjusted model following Brown & Warner (1980, 1985), given by:
(1.) 𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡− 𝑅𝑚𝑡
23 thereby influencing the β estimation. The assumption of β=1 has shown not to cause any significant estimation errors compared to the market model with the use of a short event window by Brown & Warner (1980). We also have cross-sectional dependence across our sample caused by overlapping observation for different firms at the same time. This is accounted for by using Brown & Warner (1980, 1985) ‘Crude Dependence Adjustment’, using mean average abnormal return in the sample across time t given by:
(2.) 𝐴𝑅̅̅̅̅𝑡= 1
𝑁𝑖 ∑ 𝐴𝑅𝑖𝑡
𝑁𝑖
𝑖=1
Where N is the number of securities for which return information is available at any time t. In line with Brown & Warner (1980, 1985), we use a total of 250 days of return information of which 245 are used as estimation window to determine whether the mean average abnormal return (-2, +2) is statistically significance different from zero. Using this method in combination with more than 200 days Brown & Warner (1985) show we can assume normality already with the use of 50 different observations for N as this method uses the average across time and observation to determine the statistical significance. In their research Brown & Warner (1985) show that with this method a low probability of making the wrong assumption of statistical significance due to non-normality is already reached with the combination of 5 securities and 200 days of return info using a short event window. The limitation of this method compared to the normal market model is the dependence of the results on finding a representative market index. We believe the benefits of the market adjusted model outweigh this limitation in light of this research. The main statistic of interest is the mean average abnormal return for our 5 day window given by:
(3.) 𝐶𝐴𝐴𝑅̅̅̅̅̅̅̅̅ = 1
𝑡 ∑ 𝐴𝑅̅̅̅̅𝑡 𝑡=+2 𝑡=−2
And whether this is statistically significant different from zero, for which the test statistic is given by:
(4.) 𝐶𝐴𝐴𝑅̅̅̅̅̅̅̅̅/ 𝑆̂(𝐴𝑅̅̅̅̅𝑡)
24 another by taking the difference and measure whether this difference is statistically significant by using equation 4, in which the 𝑆̂ is given by:
(5.) 𝑆̂(𝐶𝐴𝐴𝑅̅̅̅̅̅̅̅̅̅̅̅̅) = 𝑖−𝑗)
𝑆̂(𝐶𝐴𝐴𝑅̅̅̅̅̅̅̅̅̅̅)+𝑆̂(𝐶𝐴𝐴𝑅𝑖) ̅̅̅̅̅̅̅̅̅̅)𝑗)
2
Which is Student-t distributed with 2(t-n) degrees of freedom. This method can be used as, even though we may not have the same amount of acquisitions in the different groups, we have the same number of observations over time t for every group from which the standard deviation and the t-value is calculated. This is used to analyze the difference between frequent and infrequent buyers and to see if there is a significant more negative reaction for higher order versus low-order deals made by frequent buyers, as self-attribution theory would suggest.
After this we also want to control for the possibility of the cumulative abnormal returns being driven by any other effects besides overconfidence, as shown by our model. For this we do a multivariate OLS regression using different estimations on the following model:
(6.) 𝐶𝐴𝑅𝑖𝑡 = 𝛽0+ 𝛽1𝑋𝑖𝑡+ 𝛽2𝑍𝑖𝑡 + 𝜀𝑖𝑡
25
4. Results
4.1 Announcement Returns and Acquisition Activeness
In table 2 the main findings of the event study are presented. The five day mean average abnormal return (-2, +2) are given for all, infrequent and frequent buyers and for the different payment methods. Overall we see our findings are consistent with prior research. The overall reaction to acquisition announcements is positive, in line with Petzemas & Doukas (2008) and Kind & Twardawski (2016), and 0.34% at 5% significance level. Overall we record positive average abnormal returns for all methods of payment although only ‘Cash’ is significant. This shows that overall acquisitions are seen as value creating for Dutch listed firms.
Table 2
Cumulative Average Abnormal Returns and Mean Difference
All Cash Shares Mixed
All Buyers 0.34%** 0.30%* 0.89% 0.29% 368 169 26 173 Infrequent Buyers 0.57%** 0.38% 1.09% 0.61% 170 65 19 86 Frequent Buyers 0.14% 0.26% 0.35% -0.02% 198 104 7 87
Mean difference average 0.43%** 0.12% 0.73% 0.63%**
excess return (0.031) (0.633) (0.561) (0.035)
Notes: This table presents the cumulative average abnormal return (-2, +2) for the complete sample of
acquisitions, divided by the proxy for overconfidence and the mean difference between the two. We also present this for the different payment methods where ‘Cash’ are all payments in cash, debt or loan notes; ‘Shares’ are pure share payments; and ‘Mixed’ are all the other payment methods as given by the Zephyr database and mixtures of the before mentioned P-values provided in parentheses; significance level denoted as *10%, **5% and ***1%
26 though we don’t record a significant abnormal return for frequent buyers, the significant mean difference in market reaction provides evidence for hypothesis 1 of overconfident managers having lower announcement returns than non-overconfident managers. Like Petzemas and Doukas (2008), we still record a positive market reaction, also for frequent acquirers of 0.14% even though insignificant.
Several studies show that the method of payment discloses information to the market about own firm beliefs from the management. Through this, behavioral biases, like overconfidence, are likely to influence the choice of payment. Buying a firm with shares signals to the market that management thinks their firm is overvalued (Myers & Majluf 1984). If this overvaluation would drive our results we would expect to see relatively more acquisitions done with shares by frequent buyers when compared to infrequent buyers. Table 2 shows this is not the case; instead, frequent buyers prefer cash payments. This is in support of our proxy measuring overconfidence, as it signals management believes their firm is undervalued but still get involved in multiple acquisitions within a short time span. Furthermore we can see that the returns to frequent buyers are systematically lower than infrequent buyers for all the payment methods, even though insignificant. These findings show that even though management gets involved in multiple deals in a short time span they fail to create extra value for the firm compared to deemed rational acquirers, showing signs of too much optimism and confidence in their own abilities, knowledge and valuation.
4.2 Announcement Returns and Self-Attribution
27 development as we would expect the independent supervisory board not to let this happen and stop management from doing too optimistic acquisitions.
Table 3 presents the findings on this comparing the first deals with the deals made after this by frequent buyers. We can see that the first deal done by frequent buyers is met with higher announcement returns when compared to any of the group of subsequent deals, in line with the finding of Petzemas & Doukas (2008). Even though these findings are in line with the self-attribution prediction none of the mean differences are significant, proving hypothesis 2 and making it impossible to make any inferences. The most noticeable finding is the reaction of the market staying almost level after the first deal, having a maximum difference of 0.04% between the third and the fifth deal group. This could signal the market overreacts on the first deal leading to insignificant results in mean difference.
Table 3
Frequent buyers 1st deal CAR compared to later deals
Deals CAR
Frequent buyer 1st deal 10 0.43%
Frequent buyer 2nd or more deal CAR 85 0.28%
Mean Difference 0.16%
1st minus 2nd deal (0.693)
Frequent buyer 3rd or more deal CAR 75 0.26%
Mean Difference 0.18%
1st minus 3rd deal CAR (0.602)
Frequent buyer 4th or more deal CAR 65 0.27%
Mean Difference 0.16%
1st minus 4th deal CAR (0.641)
Frequent buyer 5th or more deal CAR 55 0.30%
Mean Difference 0.13%
1st minus 5th or more deal CAR (0.704)
Notes: This table presents the cumulative average abnormal return (-2, +2) for the 1st deal made by frequent
buyers and the total average of deals made after this for the 2nd,3rd, 4th and 5th and more deals plus the mean
difference of these groups compared to the 1st deals made. The P-values are provided in parentheses; significance
28 4.3 Regression Analysis of Acquirer’s Announcement Returns
Up till now we have analyzed the market reaction from acquisition announcements and shown a statistical significant difference between non-overconfident buyers and overconfident buyers using our activeness proxy for overconfidence. We argue that the reason for this is that managers who are overconfident keep on buying firms as they see this as a success in a bidding contest, even when it results in less value creation. We looked to find evidence for this argument by looking at the value creation of the first acquisition compared to the following acquisitions made by frequent buyers, claiming that if these were due to overconfidence stemming from self-attribution we would see a significant negative difference between the first and fifth acquisition. Since we could not find evidence for this, we want to test which other factors possibly drive our results and control for endogeneity by doing multivariate OLS regressions on the market reactions (CAR -2, +2) and a set of focus and control variables, discussed in section 3.4. In appendix B we present an overview of the focus and control variables with the summary statistics and their correlation.
29 what theory would predict, we find a negative influence on the announcement returns for these deals.
Table 4
OLS regressions CAR acquisitions
Infrequent Frequent All
(1.) (2.) (3.) (4.) (5.) (6.) (7.) (8.) 1st Deal frequent 0.004 0.003 0.002 0.002 buyer (0.401) (0.496) (0.667) (0.710) 5th Deal frequent 0.002 0.002 0.000 -0.001 buyer (0.190) (0.511) (0.973) (0.655) Frequent buyer -0.002 (0.215) Cash deals 0.000 0.003* 0.003* 0.001 0.001 0.002 (0.990) (0.099) (0.094) (0.292) (0.339) (0.240) Share deals -0.003 0.013** 0.014** 0.003 0.003 0.003 (0.440) (0.036) (0.026) (0.359) (0.369) (0.395) Non-Diversifying -0.004 -0.002 -0.001 -0.002* -0.002* -0.003* deals (0.154) (0.323) (0.386) (0.086) (0.088) (0.076) Domestic deals -0.009** 0.002 0.001 -0.004* -0.004* -0.004* (0.020) (0.576) (0.662) (0.088) (0.081) (0.070) Cash flow 0.000 -0.000 -0.000 0.000 0.000 0.000 (0.251) (0.221) (0.222) (0.762) (0.784) (0.853) Ln size acquirer 0.000 -0.001 -0.001 -0.001 -0.001 -0.001 (0.966) (0.363) (0.487) (0.196) (0.208) (0.345) Size buyer to 0.009*** 0.002 0.000 0.007*** 0.007*** 0.007*** deal value (0.000) (0.897) (0.983) (0.000) (0.000) (0.002) High debt -0.004 -0.001 -0.001 -0.002 -0.002 -0.002 capacity (0.309) (0.694) (0.801) (0.251) (0.254) (0.212) CAPEX -0.000 0.000 0.000 -0.000 -0.000 -0.000 (0.222) (0.431) (0.474) (0.753) (0.730) (0.747) Tobins Q 0.001 -0.000 -0.000 -0.000 -0.000 -0.000 (0.365) (0.354) (0.406) (0.368) (0.330) (0.461) Acquirer Return -2.082 0.726 0.764 -0.125 -0.108 -0.098 (0.131) (0.212) (0.198) (0.866) (0.885) (0.893) Market Return 0.651 -1.090 -1.125 -0.766 -0.783 -0.715 (0.770) (0.477) (0.468) (0.551) (0.556) (0.580) Merger Activity -0.017 -0.001 -0.000 -0.007 -0.006 -0.007 (0.179) (0.946) (0.999) (0.568) (0.575) (0.562) Intercept 0.026 0.027 0.001 0.022 0.029** 0.003*** 0.029** 0.028* (0.244) (0.266) (0.381) (0.342) (0.047) (0.000) (0.046) (0.066)
Industry effects Yes Yes No Yes Yes No Yes Yes
Year effects Yes Yes No Yes Yes No Yes Yes
Observations 146 208 208 208 354 357 354 354
30 Notes: This table presents the regressions estimates with as dependent variable the cumulative abnormal return (-2, +2) for the acquisitions done by Dutch listed firms for the total sample and divided by our overconfidence proxy while controlling for deal and firm characteristics. Columns (5) till (8) present the regression coefficients for the complete sample, whereas column (1) presents the outcomes for non-frequent buyers and columns (2) till (4) for the frequent buyers as divided by the proxy for overconfidence. Columns (3), (4), (6) and (7) present the regression for frequent buyers and all buyers where the ‘1st deal frequent buyer’ is a dummy variable that takes on the value of one if this is the case, and otherwise zero, and ‘5th deal frequent buyer’ is a dummy that takes on the value of one if the deal was the 5th deal or more by a frequent buyer and otherwise zero where in column (4)
and (7) we also control for deal and firm characteristics. In column (8) ‘Frequent buyers’ takes on the value of one if the acquirer is labelled a frequent buyer according to our overconfidence proxy and otherwise zero. In column (1), (2), (4), (5), (7) and (8) we control for deal and firm specific characteristics where cash deals takes on the value of 1 if the deal is made in cash or debt, and otherwise zero. Share deals takes on the value of 1 only if the deal is done with only shares, otherwise zero. Non-diversifying deals takes on the value of one if the acquired target is in the same industry, otherwise zero. Domestic deals takes on the value of one if the acquired target was in the same country, otherwise zero. High debt capacity takes on the value of one if the acquirer debt capacity exceeds the sample median debt capacity. Furthermore are the cash flow presented by the EBITDA, the log of the total assets, the size measured as the market value relative to the deal, the CAPEX presented by investments on PP&E, Tobins Q measured as market value divided by total assets, acquirer stock return and market return in the six months before the acquisition and the merger activity as 1 plus the log of the number of acquisitions done in the sample in the 6 months before the acquisitions used as control variables. As last we control for any industry and time invariant effects by adding year and industry fixed effects. The P-values are provided in parentheses; significance level denoted as *10%, **5% and ***1%
In regression (3), (4), (6), (7) and (8) we introduce the focus variables in the light of this research, in line with the variables used by Petzemas and Doukas (2008). To see if being a frequent buyer has a significant negative influence on announcement returns, as self-attribution theory would suggest, we introduce the frequent buyer dummy in regression (8). Although it has a negative sign as theory predicts, it is insignificant in explaining the announcement returns while controlling for all the other factors. We see no different outcomes in the coefficients for the control variables compared to the regression without the focus variable.
Like Petzemas & Doukas (2008), we also introduce two focus variables to show possible self-attribution. Self-attribution theory would suggest the fifth or higher deal to have a significant more negative coefficient when compared to the coefficient of the first deal. In regression (3) we have done this only for the frequent buyers without controlling for anything whereas we did the same in regression (6) for the whole sample. In regression (4) and (7) we also included the control variables. For all regression we find that the prediction of self-attribution theory is true, but show to have insignificant coefficients.
31 the deal compared to the size of the buyer has a higher predictive power on whether the acquisition is seen as value creating or not with a highly significant positive coefficient, meaning the deal is seen as more value creating the higher the deal value is relative to the size of the buyer for the whole sample. For the frequent buyers the type of payment, and the signal that comes out of this, shows to have the best explanatory power as to if the deal is perceived seen as value creating.
4.4 Robustness Test
The results all point in the direction of overconfidence out of self-attribution not playing a significant role in the acquisition quality of Dutch firms. We do record lower returns to more active buyers but cannot detect a significant negative trend, which suggests overconfidence out of positive self-attribution is fueling this high activeness. Like Billet & Qian (2008) and Petzemas & Doukas (2008), we also acknowledge that this higher activeness can also come out of different motives besides overconfidence and want to test for this. One motive that can explain higher activeness, even though it comes with lower value creation, is empire-building (agency costs). To control for this, and possibly other motives, we use an alternative measure to show overconfidence based on insider dealings. We use the management own firm stock trading before an acquisition as overconfident management believes it can create value with acquisitions and we therefore would expect to see them increase their ownership more than non-overconfident management. Also, if the higher-order acquisition was out of empire building, or any other agency costs driven motive, we would expect to see management not increasing its own stake in the firm knowing it’s taking more risk.
In creating this measure we follow the existing literature and look at the own firm stock trading for the management in the 6 months before the acquisition. The management dealings were manually taken for each individual member of the management from the register on the Dutch financial authority website of the AFM3. The management, meaning supervisory board as well as executive management, from Dutch listed firms is by law obligated to disclose their own firm stockholding and dealing in this to the Dutch financial authority to be made public. The management dealings are taken over the full length of the register, starting from November 2006 and leading us to delete all observations till 2007 for this measure, ending up
3
32 with a total of 179 observations of which 75 are frequent buyers and 104 are infrequent buyers.
To make a good comparison of the observations possible we used a normalized measure of stock trading called the net purchase ratio. To create the measure, the total number of stock dealings by the management, buys and sells including those from option exercise, were taken from the register. From this, the total number of sells was deducted from the total number of buys to create the net purchase of the management in the 6 months before any acquisition. This total net purchase is than divided by the total number of stock dealings, buys plus sells, to create the net purchase ratio for each acquisition. If our measure of overconfidence out of self-attribution is reliable, and not measure something else, we would expect to see the net purchase ratio of the frequent buyers to be higher when compared to the net purchase ratio of the infrequent buyers.
Table 5
Net Purchase Ratio and Mean Difference
All Buyers 25% 179 Infrequent Buyers 23.36% 75 Frequent Buyers 27.46% 104
Mean difference net purchase 4,11%
Ratio (0.5842)
Notes: This table presents the average net purchase ratio for the complete sample of acquisitions, divided by the proxy for overconfidence and the mean difference between the two. The net purchase ratio is calculated by deducting the total own firms stock sold from the total own firm stock bought and dividing this by the total amount of own firm stock trading by management in the 6 month before the acquisition. The P-values are provided in parentheses; significance level denoted as *10%, **5% and ***1%
33 Table 6
OLS regressions net purchase ratio acquisitions Multivariate OLS regression NPR
Frequent Buyers All Buyers
(1) (2) (3) (4) NPR -0.002 -0.004 -0.004** -0.004* (0.590) (0.424) (0.040) (0.062) Cash deals 0.002 0.002 (0.550) (0.295) Share deals 0.017 0.001 (0.099) (0.794) Non-Diversifying -0.002 -0.003 Deals (0.461) (0.119) Domestic deals 0.017 -0.003 (0.002) (0.500) Cash flow -0.000 0.000 (0.057) (0.146) Ln size acquirer 0.013 0.000 (0.013) (0.741) Size buyer to 0.007 0.009 deal value (0.716) (0.000) High debt 0.001 -0.001 Capacity (0.859) (0.702) CAPEX -0.000 -0.000 (0.066) (0.051) Tobins Q 0.000 0.000 (0.870) (0.765) Acquirer Return -0.993 -0.563 (0.541) (0.555) Market Return 0.369 -0.263 (0.872) (0.875) Merger Activity 0.034 -0.010 (0.190) (0.455) Intercept 0.001 -0.232** 0.004*** 0.026 (0.275) (0.017) (0.000) (0.775)
Industry effects No Yes No Yes
Year effects No Yes No Yes
Observations 75 75 179 168
R-squared 0.004 0.386 0.020 0.255
34 In column (2) and (4) we also control for deal and firm specific characteristics where cash deals takes on the value of 1 if the deal is made in cash or debt, and otherwise zero. Share deals takes on the value of 1 only if the deal is done with only shares, otherwise zero. Non-diversifying deals takes on the value of one if the acquired target is in the same industry, otherwise zero. Domestic deals takes on the value of one if the acquired target was in the same country, otherwise zero. High debt capacity takes on the value of one if the acquirer debt capacity exceeds the sample median debt capacity. Furthermore are the cash flow presented by the EBITDA, the log of the total assets, the size measured as the market value relative to the deal, the CAPEX presented by investments on PP&E, Tobins Q measured as market value divided by total assets, acquirer stock return and market return in the six months before the acquisition and the merger activity as 1 plus the log of the number of acquisitions done in the sample in the 6 months before the acquisitions used as control variables. As last we control for any industry and time invariant effects by adding year and industry fixed effects. The P-values are provided in parentheses; significance level denoted as *10%, **5% and ***1%
This insignificance could mean that management from Dutch listed firms is not overconfident in their acquisitions or our results are driven by outliers resulting in insignificant averages. Table 6 above further presents the multivariate OLS regression similar to the one in table 4 but with as focus variable the net purchase ratio (NPR) for every acquisition to measure the effect of this on the abnormal returns. Regression (1) and (3) show the effect of the NPR for the frequent buyers and all buyers. In-line with the findings of Petzemas & Doukas (2008) and what the theory on overconfidence would suggest, it show negative coefficients for the NPR but only significant for the sample with all buyers. Also, when we include the control variables for both in regression (2) and (4) we get the same results with only the NPR coefficient for all buyers having a negative and significant effect on the abnormal returns for the acquisition. All in all we get inconsistent and insignificant results on this alternative measure of overconfidence compared to the existing literature (Billet & Quan, 2008; Petzemas & Doukas, 2008; Kind & Twardawski, 2016) and what theory would predict. These findings are in line with the results on our activeness measure to show overconfidence out of self-attribution for the frequent buyers.
35 effect. Doing this also has the effect of including a bigger variety of firms in the group of frequent buyers as more firms do a second acquisition in one year than the five acquisitions in three years used before. By doing this we take away the possible bias of our results being driven by a small group of firms. Table 7 shows the results of these tests. Due to a rolling window of one year there are less observations removed to ensure that the first acquisition is really the first acquisition for a frequent buyer. This way we have a sample of 383 deals instead of 368.
Table 7
Overview event study: 1 year rolling window Panel A: Cumulative average abnormal returns and mean difference
All Cash Shares Mixed
All Buyers 0.30%** 0.29% 0.72% 0.26% 383 178 26 179 Infrequent Buyers 0.41%** 0.51%* 0.57% 0.30% 209 88 18 103 Frequent Buyers 0.18% 0.07% 1.06% 0.20% 174 90 8 87
Mean difference average 0.23% 0.43% -0.49% 0.09%
excess return (0.220) (0.076)* (0.723) (0.765)
Panel B: Frequent acquirer 1st deal CAR compared to later deals
Deals CAAR
Frequent buyer 1st deal 60 0.05%
Frequent buyer 2nd or more deal 171 0.18%
Mean Difference -0.12%
1st minus 2nd deal (0.511)
Frequent buyer 3rd or more deal CAR 111 -0.01%
Mean Difference 0.06%
1st minus 3rd deal CAR (0.768)
Frequent buyer 4th or more deal 82 0.10%
Mean Difference -0.05%
1st minus 4th deal CAR (0.819)
Frequent buyer 5th or more deal 61 0.16%
Mean Difference -0.11%
1st minus 5th or more deal CAR (0.621)
36 Using this proxy for overconfidence we don’t get different results in significance. We still record a positive significant average abnormal return for all deals and for the acquisitions done by infrequent buyers and we record a more negative but insignificant return for the frequent buyers. Most noticeable is that now the mean difference is also lower and became insignificant. The biggest difference compared to our main findings is that we do get a significant mean difference between frequent and infrequent buyers for the payments made in cash. Also if we look at first acquisitions compared to later we can’t find any significant trend which would signal that the frequent acquisitions are done due to overconfidence out of positive self-attribution.
37
5. Conclusion and Discussion
In this paper we examined if overconfidence through self-attribution could explain bad performance in mergers and acquisitions for Dutch listed firms. We especially looked at overconfidence shown by positive self-attribution for Dutch listed firms as these have a different board structure when compared to the firms in the existing literature so far. By doing this we want to give empirical evidence on the claim in corporate governance literature that more independence by board members results in better acquisition performance and contribute in the literature on overconfidence out of self-attribution.
We used a dataset of 368 deals done by Dutch listed firms from 1997 till 2016. We looked at the announcement returns for the acquisitions combined with the acquisition activeness in a short time span by management to proxy overconfidence. By doing an event-study we retrieved the abnormal announcement returns of the acquisitions for which we see, overall that doing an acquisition creates value for the firms in our sample. After dividing our sample of acquisitions in overconfident and non-overconfident acquisitions we see that, although the first go in to multiple acquisitions within a short time span, they fail to create any extra firm value when compared to the firms that do fewer acquisitions.
38 As last, we tested if our activeness proxy for overconfidence is really measuring overconfidence and not doing acquisitions out of other motives, like empire-building (agency cost). For this measure we used the management own firm stock trading before an acquisition, arguing that overconfident buyers would increase their own firm stockholding more that non-overconfident infrequent buyers. Although we do record a negative difference, the results for this are also insignificant.
39 Doing this research for the first time on a sample of firms with a dual board structure has caused some extra limitations when compared to the existing literature on the same subject. Most of these are data driven as there is simply less data and databases available for Dutch listed firms when compared to US and UK listed firms. Our results may be biased due to the lack of observations compared to comparable research done in the US and UK. Using Zephyr for the maximum amount of time available we had a sample of 368 acquisitions compared to 2,206 US deals for Billet & Qian (2008) and 5334 UK deals for Petzemas & Doukas (2008). To see if the lack of observations has driven our results the same research should be done in the German market, seeing this is a bigger market and is also bounded by law to use a two-tier board.
We measured overconfidence on a management level, even though this is a personal trait. Besides the before mentioned motive for doing this (the management as a whole are responsible for an acquisition and turnover is assumed not to be so high that all experience would be lost within the short time span of our overconfidence measure), there is also no data on the members of the supervisory board and the board of directors going back far enough to create a Dutch sample. Especially as Kind & Twardawski (2016) have shown that board of directors’ overconfidence is distinct from CEO overconfidence, this would be interesting to show if our findings are consistent, and also insignificant, measuring this separately for CEO and supervisory board in a two-tier structure.
Since we do measure a significant negative difference in abnormal returns between frequent and infrequent buyers, further research can be done in trying to explain this difference, as overconfidence out of self-attribution cannot. We already know the type of payment can explain the abnormal returns when management gets involved into multiple acquisitions but, seeing the big difference in abnormal return between the first and later acquisitions done by this group, it could also be an overreaction to the first acquisition. Further research can show if this is the case.
41
References
Ackert, L.F., Deaves, R., 2010. Behavioral finance: psychology, decision-making and markets. Mason: South-Western Cengage Learning.
Banerjee, S., Humphery-Jenner, M., Nanda V., 2015. Restraining Overconfident CEOs through Improved Governance: Evidence from the Sarbanes-Oxley Act. Review of Financial Studies 28, 2812-2858.
Barber, M.B., Odean, T., 2000. Trading is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors. The Journal of Finance 25, 773-806
Billet, M.T., Qian, Y., 2008. Are Overconfident Born or Made? Evidence of Self-Attribution Bias from Frequent Acquirers. Management Sience 54, 1037-1051
Boot, A.W., Macey, J.R., Schmeits, A., 2005. Towards a New Theory of Corporate Governance: Objectivity versus Proximity. Unpublished Working Paper.
Brown, S.J., Warner, J.B., 1980. Measuring Security Price Performance. Journal of Financial Economics 8, 205-258
Brown, S.J., Warner, J.B., 1985. Using Daily Stock Returns: The Case of Event Studies. Journal of Financial Economics 14, 3-31
Bruner, R.F., 1988. The Use of Excess Cash and Debt Capacity as a Motive for Mergers. Journal of Financial and Quantitative Analysis 23, 199-217.
Donaldson, L., Davis, J.H., 1991. Stewardship Theory or Agency Theory: CEO Governance and Shareholder Returns. Australian Journal of Management 16, 49-64.
Doukas, J.A., Petzemas D., 2008. Acquisitions, Overconfident Managers and Self-attribution Bias. European Financial Management 13, 531-577.
Fama, E.F., Jensen, M.C., 1983. Separation of Ownership and Control. Journal of Law and Economics 26, 1-31
42 Gervais, S., Odean, T., 2001. Learning to be Overconfident. The Review of Financial Studies 14, 1-27.
Goel, A.M., Thakor, A.V., 2008. Overconfidence, CEO Selection and Corporate Governance. The Journal of Finance 33, 2737-2784
Haleblian, J., Finkelstein, S., 1999. The Influence of Organizational Acquisition Experience on Acquisition Performance: A Behavioral Learning Perspective. Administrative Science Quaterly 44, 29-56.
Hansen, R.G., 1987. A Theory of Choice of Exchange Medium in Mergers and Acquisitions. The Journal of Business 60, 75-95.
Harford, J., 2005. What Drives Merger Waves? Journal of Financial Economics 77, 529-560 Haunschild, P.R., Davis-Blake, A., Fichman, M., Managerial Overcommitment in Corporate Acquisition Processes. Organization Sience 5, 528-540
Hayward, M.L., Hambrick, D.C., 1997. Explaning the Premiums Paid for Large Acquisitions: Evidence of CEO Hubris. Administrative Sience Quarterly, 103-127.
Healy, P.J., Moore D.A., 2008. The Trouble With Overconfidence. Psychological Review 115, 502-517.
Hogg, M.A., Terry, D.I., 2000. Social Identity and Self-Categorization Processes in Organizational Contexts. Academy of Management Review 25, 121-140
Jensen, M.C., Meckling, W., 1976. Theory of the firm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3, 305- 360.
Kahneman, D., 2011, October 19. Don’t Blink: The Hazards of Confidence. New York Times Magazine. Available at:
http://www.nytimes.com/2011/10/23/magazine/dont-blink-the-hazards-ofconfidence.html?pagewanted=1&_r=1&emc=eta1.
43 Kolasinski, A.C., Li, X., 2013. Can Strong Boards and Trading Their Own Firm’s Stock Help CEOs Make Better Decicions? Evidence from Acquisitions by Overconfident CEOs. Journal of Financial and Quantitative Analysis 48, 1173-1206.
Lamont, O.A., Polk, C., 2001. The Diversification Discount: Cash Flows versus Returns. The Journal of Finance 56, 1693-1721
Langer, E.J., 1975. The Illusion of Control. The Journal of Personality and Social Psychology 32, 311-328
Malmendier, U., Tate, G., 2005. CEO Overconfidence and Corporate Investment. The Journal of Finance 60, 2661-2700
Malmendier, U., Tate, G., 2008. Who Makes Acquisitions? CEO Overconfidence and the Market’s Reaction. Journal of Financial Economics 89, 20-43
McDonald, M.L., Westphal, J.D., Greabner, M.E., 2008. What Do They Know? The Effects of outside Director Acquisition Performance. Strategic Management Journal 29, 1155-1177. Miller, D.T., Ross, M., 1975. Self-Serving Biases in the Attribution of Causality: Fact or Fiction? Psychological Bulletin 82, 213-225.
Moeller, S.B., Schlingemann, F.P., Stulz, R.M., 2004. Firm Size and the Gains From Acquisition. Journal of Financial Economics 73, 201-228
Moeller, S.B., Schlingemann, F.P., Stulz, R.M.., 2005. Wealth Destruction on a Massive Scale? A Study of Acquiring-Firm Returns in the Recent Merger Wave. The Journal of Finance 60, 757-782.
Moeller, S.B., Schlingemann, F.P., Stulz, R.M., 2005. Global Diversification and Bidder Gains: A Comparinson Between Cross-Border and Domestic Acquisitions. Journal of Banking and Finance 29, 533-564
44 Oskamp, S., 1965. Overconfidence in Case-Study Judgements. Journal of Consulting
Psychology 29, 261-265.
Roll, R., 1986. The Hubris Hypothesis of Corporate Takeovers. Journal of Business 59, 197-216.
Sheifler, A., Vishny, R.W., 1997. A Survey of Corporate Governance. The Journal of Finance 52, 737-783
45
Appendices
Appendix A: Development overconfidence proxy over time
Acquirer Announcement date Deal order Is firm Overconfident?
A 1-1-2000 1 No A 1-1-2001 2 Yes A 1-1-2002 3 Yes A 1-1-2003 4 Yes A 1-1-2006 5 Yes A 2-1-2009 1 No
