Anchoring in Mergers & Acquisitions

Table of Contents

Abbreviations

Figures

Tables

1 Introduction
1.1 Objective and Contribution
1.2 Research Methodology
1.3 Main Findings
1.4 Limitations
1.5 Structure of the Study

2 Literature Review
2.1 Traditional MergerTheory
2.2 BehavioralMerger Theory
2.3 Reference Point Thinking
2.3.1 Theory
2.3.2 Anchoring in Economics and Finance
2.3.3 Anchoring in Mergers and Acquisitions

3 Hypotheses

4 Sample Composition and Methods
4.1 Dataset
4.2 Definition of Variables
4.3 Methods
4.3.1 Winsorization
4.3.2 Linear Regression
4.3.3 Gaussian Kernel Regression
4.3.4 Piecewise Linear Regression
4.3.5 Probit Regression
4.4 Descriptive Statistics

5 Results and Findings
5.1 Reference Pricing in Takeover Bids
5.1.1 Linear Analysis
5.1.2 Non-linear Effects
5.1.3 Robustness Checks
5.2 Takeover Success and the Adjustment of Takeover Pricing
5.2.1 Reference Prices and Deal Success
5.2.2 Reference Prices and Offer Adjustment
5.2.3 Robustness Checks
5.3 Discussion of Findings
5.4 Limitation of Results

6 Summary and Perspectives

Bibliography

Appendix

Abstract

Several studies show that people are often influenced by reference points which are derived from the context at hand in estimations or decisions under uncertainty. This analysis deals with the importance of reference points in the UK and to a lesser extent the European takeover market. Applying "prospect theory" to equity investment we would predict that shareholders value gains and losses relative to a reference point in a very human fashion. In a psychological bias called the "anchoring effect", investors adopt irrelevant salient anchor values as reference points and are biased towards these. We follow these predictions and, in this thesis, test historic peak prices as anchor values in corporate takeovers for bidder and target management as well as shareholders. Firstly, we analyze whether bid prices are affected by the 13-, 26-, 39-, 52- and 65-week high prices. Secondly, we are interested in whether for companies whose valuation has fallen far from the historic peaks, these anchor values are of lesser relevance, consistent with the S-shaped form of the prospect theory value function. Finally, bidding above the historic peak prices is tested and analyzed, in whether this entails a higher probability of bidding success.

For the purpose of the thesis, a dataset of in total 1602 takeover bids for listed companies in the United Kingdom (1559), Germany (36) and Poland (7) from Thomson ONE Banker was constructed. The sample includes bids announced and completed in the time period of 1985 - 2011. Data on stock price history for the target companies was extracted from Thomson Reuters Datastream. OLS regression, Gaussian Kernel Regression, Piecewise Linear Regression and Probit regressions are the tools employed to thoroughly analyze the data.

We find that all historic reference values examined exhibit statistically significant impact on the bidder offer prices with diminishing impact of the reference prices observed for extreme values. Importantly, no statistically significant evidence that bidding above historic peak prices secures a higher acceptance rate from target shareholders was discovered. This finding does not support the reference point thinking for target shareholders. The same is true for offer price increases. Offer prices below peak prices do not carry a higher probability of being amended than offer prices ranging above reference prices.

Abbreviations

illustration not visible in this excerpt

Figures

Figure 1: The S-shaped value function according to prospect theory

Figure 2: Historic distribution of bids with average offer- and 52-week high prices

Figure 3: Offer Price Density Distribution

Figure 4: Gaussian Kernel Regression Output

Figure 5: VBA Code for Winsorization of the Dataset

Figure 6: VBA Code to calculate XWKHs

Tables

Table 1: Historic Sample Distribution

Table 2: Descriptive Statistics

Table 3: Linear Regression Results for Takeover Pricing

Table 4: Piecewise Linear Regression Results of Takeover Pricing

Table 5: Takeover Pricing Robustness Checks 1

Table 6: Takeover Pricing Robustness Checks 2

Table 7: Piecewise Linear Regressions for Subsamples

Table 8: Probit Regressions for Deal Success in Takeovers

Table 9: Probit Regressions for the Probability of Sweetened Deals

Table 10: Robustness Checks for Deal Success and Offer Adjustment

Table 11: Historic Sample Distribution of Dataset before Winsorization

Table 12: Descriptive Statistics for Dataset before Winsorization

Introduction

Mergers & acquisitions (M&A) attract the attention of a wider public audience in a way that is rarely found in other fields of economics and finance. The very fact that extremely large sums of money are moved in the course of a merger^[1] is the main reason why M&A and all the issues around them seem to be omnipresent in the media. In the year 2011 alone, global merger activity totaled USD 2203billion with European deals accounting for USD 705billion (mergermarket, 2012). This immense volume in combination with other facts is also a major reason why corporate takeovers are of great interest to the academia. The "market for corporate control" is also probably one of the best researched fields in modern finance because of its importance to the corporate world. In no other way can a company experience such a dramatic change in its strategic focus as by the completion of an acquisition.

The long-term success of acquisitions is questioned in many cases. The key factor for success in acquisitions seems most obviously not to overpay for any future cash flows gained. While according to standard theory, fair takeover prices should be derived from the future free cash flows of the target on a standalone basis discounted at a risk adjusted rate and synergies gained from the combination of the target and the bidder company, this does not seem to reflect real world acquisition pricing appropriately. One explanation for the disappointing performance of acquisitions can be found in psychological biases of those making acquisition decisions. It is the aim of this thesis to examine these biases and their impact on the takeover activity.

1.1 Objective and Contribution

The main objective of this thesis is to test for reference price thinking in the context of the market for corporate control. This approach is based on the so-called prospect theory, developed by Kahneman & Tverksy (1979). In contrast to standard economic theory, "prospect theory" predicts that individuals, in the case of corporate takeovers that means managers and shareholders, value gains and losses relative to a certain reference point. In a psychological bias called "anchoring effect", managers and investors adopt irrelevant salient anchor values as reference points and are biased towards these. If anchoring is present in takeovers, acquisitions pricing and success should be influenced by reference prices.

For this purpose, a dataset of 1602 takeover bids for publicly listed companies in the United Kingdom, Germany and Poland in the time period from 1985 to 2011 is compiled. The main focus of the work performed and is laid on the UK (1559 bids) market as German (37bids) and Polish (7 bids) takeover offers in the sample period are too few to draw proper conclusions.

Different reference prices are thought to influence management and shareholder behavior on both bidder and target side. While the price at which an investor purchased stock represents a reference point that is individual for every investor, historic peak prices are thought to be reference prices that are common for all shareholders and thus influence aggregate market behavior.

This thesis mainly follows the approach taken by Baker, et al. (forthcoming) where the 13-, 26-, 39-, 52- and 65-week high prices are tested as reference prices in mergers for the US American market for corporate control. The analysis is twofold: Firstly, bid prices are tested for a bias towards the peak prices of the different periods. Secondly, target shareholders are tested for the anchoring bias by analyzing the effect of historic reference prices on bid success and offer price adjustments.

The contribution of the study performed is testing for the anchoring bias in Europe, more specifically in the UK. With the UK taking an important position in the market for corporate takeovers, this study can be regarded as a robustness check of the results obtained for the US market. Additionally, while Baker, et al. (forthcoming) focus their analysis on the 52-week high, this study broadens the scope and presents various analyses for historic peak prices from 13 weeks up to 65 weeks.

1.2 Research Methodology

The approach to empirical research can be divided into an inductive and deductive approach. The inductive approach can be characterized as purely data-driven analysis without a pre- defined theory leading to a very open but mostly unspecific research outcome. This methodology is often applied in fields where theory is not established and thus knowledge in the field is too small to form specific hypotheses. The deductive approach on the contrary defines specific hypotheses derived from theory and then tests these on a dataset. This methodology is often employed in fields with extensive knowledge from theory and empirical work and serves to precisely answer hypotheses. For the purpose of this study, a deductive research approach was chosen for mainly two reasons. First of all, the existing theory on reference points and anchoring is very detailed and an advanced stage which gives the opportunity to test various specific sub-hypotheses. Secondly, as this study aims to deliver further evidence on the robustness for the European market of the study on reference points performed by Baker, et al. (forthcoming), following a deductive approach allows for comparable results either supporting or weakening the evidence found for the US market.

1.3 Main Findings

The results obtained give evidence in support of the anchoring bias in bid prices. Firstly, it can be clearly shown that all historic peak prices from the 13-week high to the 65-week high exhibit a statistically and economically significant impact on offer prices. This can be interpreted as evidence for the anchoring bias amongst bidder and target management.

In addition, convincing evidence is obtained on the fact that, in cases where the current share price dropped drastically from peak prices, the anchoring bias was less strongly pronounced. This holds for the 13- to 65-week high.

Thirdly, peak prices with increasing time frame have individual incremental effects on offer prices. This observation strongly indicates that several reference values influence bid pricing with declining economic significance starting from the 13-week high down to the 65-week high. At the same time, each high individually influences bid prices as the impact is not due to the incorporation of a more recent peak price also captured by other peak prices. Thus, the 52- week high is not the single most important historic reference price in mergers, but still occupies a significant position.

Lastly, clear evidence for the anchoring effect could not be demonstrated for target shareholders since bid acceptance rates are not discontinuously positively influenced by bids ranging higher than historic reference prices.

The results obtained confirm the evidence found by Baker, et al. (forthcoming) for the impact of historic reference prices on bid prices but fails to affirm the influence of historic highs on shareholder behavior in corporate takeovers.

1.4 Limitations

While it is often the aim of empirical work to draw universally valid conclusions from the data analyzed, there are always limitations to what extent the results can be extrapolated. For this study, there are mainly five limiting factors to the underlying dataset that limit the degree to what extent the results can be generalized.

Firstly, the dataset employed consists of UK, German and Polish bids. While the UK market for corporate takeovers represents the majority of all transactions in Europe, one has to be cautious with drawing conclusions for the European market as a whole. Although German and Polish bids were included in the dataset, the number is too low in order to draw strong conclusions from the analysis of the country subsamples. Secondly, this study analyses bids for publicly listed companies as the design of the study makes this necessary. The question remains whether reference point thinking is also present in non-public companies as this study cannot answer this question. Thirdly, data quality varies with size and geographic location of target companies with bids for smaller targets and targets located on Poland and Germany exhibiting poor data quality. Also, older bids have less information on pricing and characteristics available. Thus, this study might be biased towards bigger companies located in the UK where transactions attempts were more recent. Fourth, bidder characteristics were not available as control variables, so, inferences drawn are all based on the assumption that bidder characteristics do mainly change the economic and statistical impact of the evidence obtained for the anchoring bias.

Lastly, the dataset available consists of final bids rather than each bid being listed separately if bids were rejected and amended bids were submitted by the bidding company afterwards. So, for example, if a bid was made for a company with price x and target shareholders rejected this bid, afterwards, a second bid was submitted with a price x + y, and this was accepted, the bid price displayed in the sample would be x + y and the bid would be classified as accepted. The first bid would not have been recorded.

1.5 Structure of the Study

This thesis is structured in to six main sections. Section 2 provides an overview over the current state of research on mergers and acquisitions by first elaborating on research findings from traditional corporate finance research then describing the key research in the field of behavioral corporate finance. Within this section, the theory and findings of prospect theory, reference prices and the anchoring effect are presented. Section 3 presents the detailed hypotheses tested. Chapter 4 describes the dataset, methods and variables used in the course of this study. Chapter 5 gives a detailed description of the results and findings obtained and chapter 6 finally summarizes the paper and presents the ideas, perspectives and lessons learned

2 Literature Review

Corporate finance literature generally tries to explain the cause and effect of actions taken by the two sets of agents involved - managers and investors - in relation to: I) Investment decisions including topics such as real investments and acquisitions, II) financial decisions including the issuance of equity and debt, or capital structure decisions, and, III) other corporate decisions such as dividend policy or executive compensation. In all three categories, the examples given are far from exhaustive but they do represent key subjects in the existing corporate finance literature.

Two approaches to investor and managerial behavior can be observed in the literature. Traditional literature is based on the assumption of rationality of the involved parties and the assumption of efficient markets. In contrast, the literature referred to as "behavioral corporate finance" is a subcategory of behavioral economics that has gained momentum in the academic corporate finance literature during the last decades. In the literature following this approach the assumption of rationality is relaxed and anomalies with respect to the homo oeconomicus model are explained by theories originally developed in the field of psychology.

The literature survey in this thesis focuses mainly on takeovers and does not go into much detail on other subcategories of corporate finance. Since each one, the traditional and the behavioral approach to takeovers explain different aspects observed in M&A, first a short overview dealing with the traditional literature on M&A is provided, followed by a description of behavioral literature on M&A in the past. A more detailed summary of the theory and application of "prospect theory" and "reference point theory" in the context of acquisitions will be given.

2.1 TraditionalMerger Theory

Traditional research into the underlying reasons and circumstances why mergers occur, which price is paid for acquisitions and investor reactions to acquisitions can be divided into the theory focusing on synergies and on the other hand agency costs.

Synergies can be defined as the value increase of two merged companies over the sum of the two standalone company values if both companies are managed in a shareholder maximizing way. Synergies are the result of revenue increases, cost reductions, risk reduction on a company level or any combination of the three due to mergers. Examples for revenue increases over the added revenues of the two companies might occur in situations where acquisitions lead to a stronger position in a selling market altering the power to increase prices for a product sold. Cost reductions can result from economies of scale that allow for the shutdown of overlapping functions from either company and thereby reduce costs. Also, a merged company might exhibit higher market power on purchasing markets resulting in lower purchasing costs or joint financing might lead to lower financing costs. Risk reduction of a merged company might be due to natural hedges of the two companies that in the combined state might lead to lower revenue volatility.

Agency theory predicts that a company's shareholders and management follow diverging interests since both are mainly interested in maximizing their own wealth. The only situation where owner and management wealth maximization coincide occurs in the case of owner managers where the management owns 100% of the company. Starting already with marginal fractions of the firm being in control of a third party, agency costs exist. Costs resulting from conflicting interests between owners and management can be grouped into monitoring costs, bonding costs for the management and residual losses (Jensen & Meckling, 1976). Monitoring costs are defined as costs that the principal has to carry in order to check on the management of its company, e.g. costs for auditors, annual reporting and company ratings. Costs for incentivizing management, including variable payment components are referred to as bonding costs. The residual loss results from the fact that even in cases where management is incentivized and monitored, diverging behavior from what maximizes shareholder wealth cannot be completely avoided.

The market for corporate control is generally regarded as a strong control mechanism to lower agency costs. Since companies managed in opposition to shareholder interests trade at a considerable discount compared to companies with lower agency costs, companies with higher agency costs are targeted as possible acquisitions more often. Premiums to the market valuation paid in takeovers accordingly are thought to contain value potential from the reduction of agency costs after replacing management and ensuring shareholder value orientation within the firm (Jensen & Meckling, 1976 and Jensen, 1986).

2.2 Behavioral Merger Theory

While the model of the homo oeconomicus is a valid simplification of reality in order to explain causes and implications of actions taken in finance and other domains, one always has to keep in mind the human factor, meaning that, in the end, humans are not fully rational considering all actions taken. A whole variety of psychological biases influence human thinking and action. Behavioral corporate finance addresses this problem in the standard literature and introduces psychological phenomena as explanations for anomalies observed in managerial and investor behavior.

Although rationality is relaxed as an assumption for the models in behavioral corporate finance, researchers do not completely abolish rationality as an assumption. In fact, rationality is only relaxed for one of the two groups of agents at a time, either managers or investors. While the first set of literature follows the extreme approach of assuming fully rational managers but less than fully rational investors, the second set of research takes to the other extreme and assumes less than fully rational managers but efficient markets (Baker & Wurgler, 2012). These two approaches were dubbed "irrational investors approach" and "irrational managers approach" in the first survey of behavioral corporate finance literature published by Baker, et al. (2004). In the updated survey by Baker & Wurgler (2012) the two approaches are referred to as "market timing and catering approach" and "managerial biases". Not a change in substance of what these two approaches refer to has actually occurred, but one is rather exposed to a major change of perception of the behavioral approach within corporate finance. Rather than representing a sum of anomalies that cannot be explained by traditional theory but still critically observed by most scholars, the self-understanding of this "school" has changed and has established itself in the modern corporate finance literature. I refer to the two approaches in the original way because of the illustrative name they carry.

The reader might justly criticize these two approaches as unrealistic, especially as the two approaches explicitly contradict themselves. However, one can consider these two approaches as a ceteris paribus approach of making corporate finance research more realistic. It is very likely that this clear differentiation of the two approaches in behavioral corporate finance will not be possible in the future with publications slowly emerging that do not fit into one of the two bins. One such example is the introduction of reference points in the analysis - reference points are in fact likely to influence investors as well as managers.

Irrational Investors Approach

In their paper that actually initialized the research of the irrational investors approach, Shleifer & Vishny (2003) propose a model for takeovers that is solely driven by stock market mispricing and rationally acting managers taking advantage of this fact. This model is the key framework of the irrational investors approach, the predictions being thoroughly tested in many empirical studies.

This model involves two publicly traded companies 1, the bidder and 0, the target. Both firms have K1 and K units of capital outstanding and the current market valuation of the companies per unit is Q1 and Q with Q1 > Q. The value of the merged companies is V = S(K +K1) with S being the short-run valuation of the merger per unit of combined equity. S is the "perceived synergy" the market sees in the merger or in other words a misevaluation by the market. In the long run, the model assumes that all assets take a value of q per unit of capital. A combined company is then valued at q(K +K1)in the long-run. This means that this specific model makes the extreme assumption that no synergies between the merged companies materialize in the long run. In the acquisition of the target, a price per unit of P is paid by the target. Putting the pieces together, the following three results can be derived from the model:

(1) The short-run effect of a takeover on the combined firm value is S(K +K1) - Q1K1 - QK, the immediate effect on the target company valuation is K(P - Q) and the short-run impact on the bidder's valuation is K(S -P) + K1(S - Q1).
(2) In acquisitions paid for in cash, the long-run valuation change to the target is K(P - q) and K(q - P) for the bidding company.
(3) In acquisitions with stock, the long-run impact on bidder valuation is qK(1 - P/S) while for the target it is qK(P/S -1).

The key implication of this "market-timing model" is that the main motive of an acquirer, who is assumed to be overvalued, to buy the other company is to convert temporary overvaluation into long-term shareholder value for the his firm. There are essentially three ways to achieve this goal: First, in cash acquisitions, the bidder can preserve current overvaluation by acquiring undervalued target stock. Secondly, in stock acquisitions, bidders might use overvalued stock to acquire less overvalued target stock. The third way for the bidders' shareholders to profit from stock acquisitions, is to deliver a value proposition of perceived synergies so that P < S to the market which will result in an overvalued combined entity in the long run. The result is that long-term investors face a lesser downward adjustment of the stock as compared to the situation when no takeover was achieved (Shleifer & Vishny, 2003). Baker & Wurgler (2012) and Dong (2010) survey the empirical studies performed to test this framework and conclude that various patterns observed in empirical work is consistent with the overvaluation-driven merger activity.

Irrational Managers Approach

The very opposite to the irrational investors approach is taken in the irrational mangers approach. While managers are modeled as acting rational in the framework given by Shleifer & Vishny (2003), the irrational managers approach regards managers as biased in many ways. Importantly, on the other hand, efficient markets are assumed. The literature in this area mainly deals with the psychological biases, such as the overconfidence and optimism of the managers involved in takeovers.

With his seminal contribution on the managerial hubris as key driver for acquisitions, Roll (1986) led the way for many empirical studies testing the framework proposed. Roll (1986) assumes that the market value of a firm is the average of the valuation of all market participants. Thus, if minimal or no synergies in mergers are assumed, takeover bids over market value simply represent an individual valuation error which can be classified as hubris. Takeover prices over market value are the result of overconfidence in the capability to extract value from the target that the current management was not able to realize. The model predicts negative reactions to bidder share prices and value destruction for the combined company. The predictions of the model are confirmed by evidence from 40 empirical studies surveyed by Jensen & Ruback (1983) and other studies on value creation of mergers clearly reveal evidence that points in the same direction (Baker & Wurgler, 2012). Summarizing the research published, the hubris hypothesis seems to be robust for different countries and periods of time and for both listed and private targets. Overconfident managers are more likely to complete a transaction than "rational" managers and the quality of transactions made, measured both in short-run announcement effects and long-run performance, is also worse for overconfident managers (Hayward & Hambrick, 1997; Ben-David, et al., 2007; Brown & Sarma, 2007; Doukas & Petmezas, 2007; Malmendier & Tate, 2008).

Besides hubris, reference point thinking is another psychological phenomenon that is thought to influence managerial behavior in mergers (Baker & Wurgler, 2012). However, as this phenomenon is thought to affect both manager and investor behavior, it does not really fit into the irrational managers approach as mentioned before. I will elaborate this psychological phenomenon and the influence on managerial and investor behavior in more detail in the following section.

2.3 Reference Point Thinking

In order to give the reader a broad understanding of the topic, this section will first go into detail on the phenomenon of anchoring (Tversky & Kahneman, 1974) and prospect theory (Kahneman & Tverksy, 1979), then summarize the evidence found in relation to anchoring in general economic literature and, finally, show how anchoring could have an impact in the pricing of mergers & acquisitions and what empirical evidence was already found in this respect.

2.3.1 Theory

The traditional approach to decision making under risk is "Expected Utility Theory" (EUT). Vastly embedded in economic theory, individuals are perceived to act rationally by making decisions according to the following three axioms (Kahneman & Tverksy, 1979):

illustration not visible in this excerpt

With outcome xr, probability pr, where p1 + —l·pn = 1 and individual wealth level before the decision w. Summarized briefly, EUT constitutes that decisions under risk are assessed in the following way: The utility of a decision is the expected utility of its possible outcomes (1). Decisions are made based on the expected utility of the final wealth resulting from the decision (2). And, individuals are risk averse, so any certain outcome x is preferred to a risky outcome with the expected value x (3).

Prospect Theory

In their seminal paper, Kahneman & Tverksy (1979) show that EUT is systematically violated in decision making and introduce "Prospect Theory" as a new decision theory framework. The key characteristic of the presented framework is that utility is measured based on deviations from a neutral reference point rather than final wealth levels. This leads to several effects that are incompatible with standard EUT:

"Certainty effect": Outcomes that are assumed to be certain are overweighted in comparison to probable outcomes even though the utility calculated according to EUT is higher for the probable outcome.

"Reflection effect": When outcomes are gains, people are overly risk averse, while a risk seeking behavior is observed when people are confronted with losses. The certainty effect works in two directions: for gains, the certainty effect leads to the preference of a certain gain over a higher probable gain. For losses, the certainty effect leads to the opposite. Merely probable losses are preferred to lower losses that are certain, risk seeking behavior is the result.

"Loss aversion": In facing decisions, the response to losses is more extreme than the response to gains. So, a loss of 100 leads to higher dissatisfaction than a gain of 100 leads to satisfaction (Tversky & Kahneman, 1986).

The result of the listed effects is a S-shaped value function (Figure 1) that carries the following characteristics (Kahneman & Tverksy, 1979):

(1) The function is defined on deviations from a reference point.
(2) Convexity for losses and concavity for gains account for the certainty and reflection effect.
(3) The steeper slope for losses than for gains accounts for loss aversion with the value function being steepest at the reference point.

Figure 1: The S-shaped value function according to prospect theory

illustration not visible in this excerpt

While value is driven by the two factors, the asset position that serves as a reference point and the size of deviations from the reference point, the representation of value solely as a function of change from the reference point is a good approximation (Kahneman & Tverksy, 1979).

The Anchoring Effect

While it is obvious that the reference point has critical importance to the Prospect Theory, the characteristics of reference points will now be described in more detail.

Kahneman & Tverksy (1979) state that the reference point for a decision is defined by the context in which the decision is about to be made. In many cases this context is the current asset position, the status quo, but also expectations or targeted levels can serve as reference points.

In the earlier work by Tversky & Kahneman (1974), the phenomenon of anchoring is described as an effect where arbitrary reference points^[2] influence decision making. In the experiments conducted people were first asked to estimate whether the percentage of African countries that are members of the United Nations (UN) was larger or smaller than a number between 0 and 100 determined by a spinning wheel. In a second step they were asked to estimate the percentage of countries that are members of the UN. Interestingly, although the number from the spinning wheel was obviously completely arbitrary, the estimate was significantly biased towards the number obtained by the wheel.

The extent to which the anchor value influences the final value depends on the two psychological mechanisms, the insufficient adjustment process and the knowledge accessibility effect.

In their original study Tversky & Kahneman (1974) suggest that the anchor value is the starting point for the estimation in question, triggering an "insufficient adjustment process". While adjustment from the initial reference point does occur, a bias towards the initial value remains and thus influences the final value.

Other scientists claim that the "knowledge accessibility effect" is the key force behind anchoring. If an initial anchor value in an estimation question is given, humans have the tendency to test a hypothesis in trying to find confirmation for it rather than falsification. This results in far more affirmative than refuting evidence for the initial hypothesis whether the anchor value is the true value (Kayman & Ha, 1987). The knowledge accessibility effect suggests that this imbalance in evidence results in a bias towards the initial anchor value (among others see Chapman & Johnson, 1994; Chapman & Johnson, 1999 and Strack & Mussweiler, 1997).

Epley & Gilovich (2001) and Epley & Gilovich (2005) demonstrate that, depending on the anchor type, either insufficient adjustment process or the knowledge accessibility effect takes effect: For self-generated anchors the insufficient adjustment process is the key factor driving anchoring because an anchor that was taken for a starting point but is known from the beginning not to be the true value does not trigger the accessibility effect. Externally provided anchors trigger the knowledge accessibility effect because the provided reference point, if not completely unrealistic, is regarded as a possibly true value. The adjustment from the anchor value is less strong for externally provided anchors than for self-generated anchors. As a result, the bias towards to the initially provided reference point is stronger for externally provided anchors.

A very important characteristic of the anchoring bias is that the effect observed and well documented for various domains diminishes for extreme values. The further the anchor value from a plausible value, the less influence of the anchor value on the final value was observed (Chapman & Johnson, 1994).

2.3.2 Anchoring in Economics and Finance

Research on the anchoring bias in general knowledge and related domains is manifold and numerous (see Furnham & Boo, 2011) for a good summary of the performed empirical studies). Due to the focus of research limited implications for "real-world" behavior can be derived from these papers. For this reason I will focus on publications that have implications for the research question examined in this thesis. Various economic subdomains can be considered relevant for acquisition pricing. However, I will primarily focus on work that examines consumer behavior, negotiation situations and investor behavior. Additionally, empirical evidence that laymen as well as professionals are affected by anchoring will be discussed.

Ariely, et al. (2003) show that arbitrary anchors (in this case the two last digits of the social security number written down by the participants) significantly influence consumers' willingness-to-pay (WTP) for certain products. High anchor values resulted in distinctly higher WTP than low anchor values. Simonson & Drolet (2004) provide evidence that both the WTP for consumers buying and the willingness-to-accept (WTA) a certain price of consumers selling a good is prone to the anchoring effect in situations of limited information. This bias holds both for anchor values for the same or related goods but also for goods unrelated to the product in question (Nunes & Boatwright, 2004; Adaval & Wyer, 2011).

Northcraft & Neale (1987) show that listing prices of real estate objects serve as anchors in price negotiations. Final prices paid were strongly influenced by the manipulated listing prices. The observation that final prices were strongly biased by first offers in negotiations was also made experimentally by Galinsky & Mussweiler (2001). The anchoring effect was mitigated when participants were instructed to focus on the best alternative of their opponents or to focus on their goals in the negotiation. While the first study was conducted in an information- rich real-world situation, the second was conducted under laboratory situations, however, in both cases participants did not pay with their own money. In a paper on the Troubled Asset Relief Program (TARP), Wilson (2012) shows strong empirical evidence for the anchoring effect in negotiations in real-life transactions with huge economic impact. Banks that started with low first offers in the bargaining process for warrants with the U.S. Treasury ended up paying less. So, anchoring seems to be relevant in small and big negotiations.

Evidence that prospect theory and reference point theory hold in a financial markets setting was delivered by various scientists, Shefrin & Statman (1985) being among the first ones. Their study on investment behavior reveals a "disposition effect" for investors in mutual funds. The disposition effect describes the phenomenon that investors tend to hold on to investments with positive performance too short and to hold on to investments with negative performance too long, the reference point being the purchase price. This behavior results from loss aversion, as investors are reluctant to lock in losses and accept the possibility of accruing higher losses. Amongst many others Odean (1998) and Grinblatt & Han (2005) document evidence of the disposition effect on an investor level. The disposition effect was also found among professional traders in the futures market (Heisler, 1994; Locke & Mann, 2000).

It is important to notice that the described disposition effect was documented for individual reference points, i.e. the purchase price that is unique to every investor. However, there are studies testing whether prospect theory can explain aggregate market behavior. One example where reference prices impact aggregate market behavior is the initial public offering (IPO). Loughran & Ritter (2002) develop a framework empirically tested by Ljungqvist & Wilhelm (2005) where the offer price is the common reference point and gains and losses are calculated from this value. Barberis & Huang (2008) suggest prospect theory as an explanation for IPO related pricing anomalies. Kaustia (2004) tests the disposition effect for the aggregate market for investors after IPOs where all investors have the same purchase price. He observes significantly lower trading volumes for stocks falling below the initial price compared to stocks rising above the reference price. This suggests the aversion to realize paper losses but the propensity to realize paper gains among investors.

Historical high prices are another example of very few reference points that are common to all investors in the market in contrast to purchase prices that are unique to every investor. The importance of historical highs as anchor values to investors is elaborated in several papers. Grinblatt & Keloharju (2001) deliver empirical evidence that investors in the Finnish stock market are significantly more likely to sell their holdings once the stock price has exceeded the historical monthly high. Heath, et al. (1999) observe that employees who have been granted stock option as compensation exercise these options far more often when the 52-week high (WKH) was exceeded than in cases where the stock was worth less than the 52-WKH. In a model developed by Barberis & Xiong (2009 and 2012) in order to investigate whether prospect theory can predict the disposition effect for realized gains or losses, the 52-WKH is determined as an important reference point for when investors sell their investments.

As the cited literature shows, prospect theory and anchoring can be observed in a large variety of domains with many different values, relevant or not to the situation possibly acting as reference values. However, many studies have been performed among laymen, often university students and not professionals. But, as Tversky & Kahneman (1974) stress in their early publication, "[t]he reliance on heuristics and the prevalence of biases are not restricted to laymen." In this specific case, people with "extensive training in statistics" exhibited the same behavioral biases as statistical laymen. Evidence that professionals in their respective field are prone to anchoring is delivered for different fields of expertise. Northcraft & Neale (1987) show that anchoring affects professional housing agents in the same way as university students without prior knowledge of the housing market. Englich et al. (2006) demonstrate that anchoring holds for experienced legal professionals when deciding on prison sentences on court. The anchoring effect did not change with the level of experience. Kaustia et al. (2008) provide evidence that professional investors in the Finland were subject to anchoring albeit less than university students exposed to the same information.

2.3.3 Anchoring in Mergers and Acquisitions

As described in the previous section, the research on anchoring is vast and in essence, the anchoring effect is observed in the most different domains with very different values having been tested as reference prices. The goal of this thesis is to determine whether the anchoring effect is relevant in UK, German and Polish M&A.

As discussed already in chapter 2.1, literature on M&A was mainly developed under the assumption of rationally acting market participants and behavioral aspects (2.2) - with some exceptions - have only entered research in recent years. The anchoring effect, more specifically, was this far only addressed and tested by Baker et al. (forthcoming).

Baker et al. (forthcoming) show that historic high prices of different periods with a strong focus on the 52-WKH act as the anchor values for shareholders in the US market for corporate control. Historic high prices carry two traits that make it plausible to serve as reference points: First, as the high prices externally provided by the market, they might be accepted as indication of true value of the target's stock. Second, historic high prices are salient. Not only are the historic high prices accessible for all market participants but also, as can be observed in reports on stocks, historic highs and especially the 52-WKH is regularly used as a reference value. Also, the literature analyzing trading attitude of investors emphasizes historic highs as anchor values.

An acquisition is a process with the involvement of two parties with deep knowledge in their market and of their company. Acquisitions normally are not handled frequently, so one could be inclined not to regard bidder and target management as professionals in valuing companies and acquiring companies. However, as both sides normally hire financial advisors that are specialized in takeovers, professionals in this field are very well involved in the process. While investment banks or other corporate finance advisors are very well trained on how to calculate the present value of the firm from future cash flows taking different scenarios of business development into account or calculating the value based on trading or transaction multiples of other companies and transactions in the market, the result of all these methods is always a valuation range. So, as this is a situation of limited information where a decision is being taken under risk, past high prices are likely to influence valuation on both sides with or without the involvement of professionals. The anchoring effect can be expected to be less dominant with the involvement of professionals but still observable and significant.

Acquisitions in many aspects resemble classic purchasing situations or negotiation situations and irrational investor behavior influences the success of acquisitions. As in all these domains studied by scholars, reference points evidently influence decisions, it makes sense to study whether the anchoring effect can be observed in M&A. For this purpose, I will first elaborate on the way through which both bidder and target companies can be expected to be influenced by historic high prices. Target considerations will be explained first as they can also be expected to have implications on bidder behavior.

The Target

Whenever an individual or a firm wants to take control of a company, the target board and shareholders have to accept a bid submitted by a bidder in order for a transaction to be completed. Having established that target shareholders act according to prospect theory, two prices in this context are probably employed as reference values: The purchase price and historic peak prices. Both are likely to affect the target shareholders' behavior. But, as the purchase price is different for all shareholders, this reference price will not affect aggregate market behavior. In contrast, historic peak prices are common anchor values to all shareholders and thus can be expected to influence the actions of all shareholders Baker et al. (forthcoming). This study will concentrate on historic peak prices as anchor values to shareholders.

If recent peak prices are used as reference prices by the investors, they will be reluctant to sell their stock below the recent peak price because of loss aversion and the disposition effect. As the acceptance of bid prices that are lower than recent high prices will be regarded as generating a sure loss, shareholders will probably not accept bids below the reference point but rather continue holding the share with the prospect of the stock rising over the peak price sometime in the future. On the other hand, if the offer price exceeds the historic peak, shareholders will be inclined to accept the bid locking in a sure gain. As a result, offers that exceed recent peak prices should carry a higher probability of being approved than offers below peak prices (Baker, et al., forthcoming).

However, the farer away the current market valuation of the stock is from recent peak prices, the less relevance this anchor value is expected to have. This is an implication of the S-shape of the value function, that predicts that shareholders care less about the marginal dollar lost the farer it is from the reference point (Baker, et al., forthcoming). And following the observation of Chapman & Johnson (1994), for stock prices that have fallen extremely from the peak price, the peak will not be considered as plausible valuation reference point and thus the anchoring effect will be much less strong.

Baker et al. (forthcoming) argue that anchoring can influence the behavior of the target board and shareholders in 3 different possible ways. First, under time constraint, with limited information and possibly limited ability to calculate the true value of the company by discounting future cash flows, the recent peak prices will be consulted as indication of true value of the firm. Second, it is not even necessary for this to happen unconsciously, as the target board can employ recent peaks as negotiation anchor in the bidding process with the acquiring company. Third, the target's board faces the possibility of a lawsuit for recommending a too low takeover price to its shareholders.

The Bidder

The bidder's management will be influenced by anchoring both directly and indirectly. Both bidder psychology and perception of the target's psychology can lead to anchoring to the historic peak price.

A bidder will be directly influenced by recent peak prices because the acquisition makes a valuation of the target stock necessary. Same as for the target, the bidder management will have to make a valuation under risk with limited information. Although the bidding company has time to come up with a valuation, past peak prices are very likely to be regarded as indication of true value for the bidding management. An observable anchoring effect is the result. The indirect anchoring effect results from the bidder knowing that the target shareholder and board either regard historic a peak price as true value or use it as negotiation anchor to attain the a high price for their shareholders in the negotiation process.

3 Hypotheses

The general hypothesis that reference point thinking influences takeover activity can be divided into six sub-hypotheses that are derived from theory and past findings for takeovers and related topics. As elaborated above, the bidder and target boards and shareholders are expected to be influenced towards recent peak prices in takeovers. This anchoring effect might either occur because of unconscious bias or intended negotiation behavior. The 13-, 26-, 39-, 52-, and 65-week high prices are all expected to affect takeover pricing:

[...]

---

^[1] The term "takeover", "merger" and "acquisition" slightly differ in their meaning in respect to the attitude of the transaction. However, as there is no clear distinction between the terms, they are used as equivalents next to each other in this paper.

^[2] Anchor and reference point are treated as synonyms in this thesis