+
Groningen University
Department of Finance and Investment
20 February 2007
“Prospect, Private discount and Stock split
theory applied to the bidder process”
“A bridge between Behavioural and Corporate Finance”
Abstract:
This paper empirically tests whether three particular concepts derived from existing theory also have their influence on the accomplishment of the bid premium. First of all Prospect theory states that investors with a negative return should be compensated with a higher premium since these investors are risk seeking. Results based on the Compounded Total Return (CTR) 1-month, 3-months and 1-year prior to the rumour date are used to test this hypothesis. The null hypothesis is rejected for the 1-month CTR, a breakpoint using dummies is found for the 1-month CTR model. Second strong evidence is found for a relation between the liquidity of a stock and the height of the bid premium in model 2 and 3 on a 5% level. One can state that a “private” discount is also applied to listed targets in order to compensate for illiquidity. Finally the share price itself has no influence on the determination of the bid premium according to stock split theory.
Thesis Supervisor: Prof.Dr. B.W. Lensink
Student: M.H. Lamers
Table of content:
Page
1. Introduction 2
2. Models to explain corporate takeovers and the existence of a premium 5
2.1 Ownership structure and the free rider problem 5
2.2 Dilution and replacing current management 6
2.3 The role of a large shareholder 7
2.4 Toehold as a strategic variable 9
2.5 A two tier offer 12
2.6 The Winners curse 12
2.7 Poison Pill 13
3. Overview of earlier research 14
3.1 The toehold as explanatory variable 14
3.2 Market to book value 15
3.3 Acquisition accounting and the bid premium 15
3.4 Method of payment 16
3.5 Managerial Wealth 16
4. Position of the paper 18
5. Different methodologies to determine the height of the premium 20
6. Variables to include in the model 23
6.1 The concept of loss aversion applied on the take over process 23
6.2 Liquidity of the stock 24
6.2.1 Liquidity derived from private discount theory 24
6.2.2 Liquidity derived from stock split theory 27
7. Theory behind the control variables 29
8. Problem, sub questions and hypotheses 31
8.1 Hypothesis for the liquidity and prospect theory 31
8.1.1 Share price 31
8.1.2 Liquidity 31
8.1.3 Compounded Total Return 32
8.2. Control variables 32
8.3. Testing the coefficients 33
8.4. Testing the model 34
9. The dataset 35
10. Methodology 37
11. Presentation of the results 39
11.1 Premium explained 39
11.2 Prospect theory applied to the bidder process 42
12. Conclusion 45
13. Recommendation for further research 47
1. Introduction
The global economy is booming, market indices are on their way back to the 2001 levels. Multinationals are reporting record profits and this creates opportunities for M&A activities. The headlines of the newspapers are currently filled with take over rumours and transactions. Major consolidations, like in the steel-, aerospace-, energy-, transport- and insurance sectors are about to take place. World wide the indices are boosted by these activities since large premiums are paid over the market value of the acquired shares. Especially in the United States of America and the Far East, transactions and rumour in the market boosted the indices in 2006.
Since enormous amounts of money are involved much research has been done on mergers and acquisitions. Most of these studies focus on the share price performance of the acquiring company or target company, and the distribution of the synergies between the target and acquiring shareholders. (Bradley, Desai and Kim (1988), Jovanivic and Braquinsky (2002), Dodd and Ruback (1977))
Studies on the accomplishments of the bid premium are rare. Earlier research has been focusing on several variables, for example the influence of a toehold (initial stake held by the acquirer prior to the bid), method of payment, pre-commitments and bidder competition were tested by Betton and Eckbo (2000).
Other studies like Elisabeth Callison and Linneman (1987) focussed on the explanation of bid premiums based on target liquidity, dividend payout, size, market-to-book value and profitability. This thesis tests whether there are more variables that influence the accomplishment of the bid premium and tests for three specific behavioural finance related theories that might influence the bid premium.
stock prior to the rumour date will demand a lower premium since he is risk averse. Inspired by this theory this paper would like to test whether the performance of the share, measured by the compounded annual return, prior to the rumour date influences the decision whether or not to tender their shares in case of an acquisition. More specifically whether the premiums paid are significantly higher for shares that had a negative return prior to the rumour date compared to shares that had a positive return. The returns are measured on a 1-month, 3-months and 1-year interval prior to the rumour date and tested in three separate models.
Another point of interest lies in the liquidity of a stock in relation to the bid premium. This research will try to find a relation between the premium paid and two different liquidity related theories. At first the concept of a private discount studied by Officer (2005) showed that an acquiring firm receives a discount on the stock to compensate for the illiquidity of a stock that is privately held (Officer, 2005). Different listed entities have different liquidities as well, this thesis examines whether this difference in liquidity also causes differences in the premium paid for different listed entities. Koeplin et al. (2000) also state that non-listed companies are traded at a discount, so it seems reasonable to assume a positive relation between the liquidity of different listed companies and the premium paid. Liquidity is measured as the average percentage of shares outstanding per trading day over the last year before the rumour date.
The other liquidity theory is based on stock split theory. Lakonishok and Lev (1987) and Copeland (1988) give a lot of reasons why stock splits might occur. One of those is that it increases liquidity. When we apply this behavioural theory to the field of M&A, we could assume that the price of a share has potential influence on the bid premium. In general a low share price implies a more liquid stock. For instance there are more potential buyers for a house of €100.000 compared to a house of €1.000.000. Holders of a relative illiquid stock are willing to accept a lower premium, since there are less potential buyers. The acquirer should buy a liquid stock against a high premium because there are other potential buyers which make the selling party indifferent against tendering their shares to a specific party. The average share price prior to the rumour date is used as a proxy of liquidity.
This paper aims to confirm earlier findings represented by the control variables chosen, amongst finding evidence for the pioneering variables; share price, liquidity and Compounded Total Return.
Section 2 will give a comprehensive overview of theories and models that are developed to explain the existence of a bid premium. Empirical results found in earlier studies are presented in section 3. Section 4 elucidates on the scope and positioning of this research, and describes the specific focus for the rest of the paper. Different methodologies to determine the height of the bid premium, the dependent variable, are described in section 5. The explanatory variables will be dealt with in sections 6 and 7. Section 6 presents the variables related to the three theories that will be tested, followed by the theory behind the control variables in section 7. The concrete problems and sub questions are given in section 8. An overview and explanation of the variables to include in the model is given thereafter.
2. Models to explain corporate takeovers and the existence of a premium
The pure theory about takeovers focuses on the price that is paid in an acquisition in relation to the success probability of a take over. It does not elicit on the factors that influence the price that is paid for the target. In the following paragraph an overview of the most relevant theories will be given. The first theories will very basically explain why takeovers exist and focus on the success probabilities of a takeover. Later on models to explain the height of the premium will be discussed. These models are all theoretical; the next section will summarize the most relevant empirical papers in this field of corporate finance.
2.1. Ownership structure and the free rider problem
First of all the free rider problem will be described and the patterns of stock ownership, which influences the outcomes of the takeover. A basic assumption underlying this theory is that every share has equal voting power and that the acquiring party needs a simple majority, or more generally, a fraction қ Є (0,1).
Grossman and Hart (1980) introduced a basic theory in which the existence of a premium is explained based on control rights. The model assumes that shares are diffusely held. “A” represents the range of possible actions that the current management can undertake to increase the firm value. Function f(a) describes the market value of the firm as a function of management’s action “A”. The maximum value that the current management can create is given by:
max f (a) = f (a*) (Equation 1)
aЄA
The abilities of the acquiring party to generate value are supposed to be different from the current management. The acquiring party is able to maximize value with Z, whereby Z is assumed to be positive. This could be the case due to foreseen synergies. The post acquisition firm value in this case can be given by:
v = max f (a) + Z (Equation 2)
aЄA
by the shareholder it will not tender its share until p ≥ v. At the tender offer stage, atomistic shareholders (largely diffused ownership) do not tender unless the offer price matches the post takeover share value. But this condition is clearly uninteresting to the bidding party.
This theory makes takeovers for acquiring parties unattractive. But in real life take overs do take place. One of the most likely explanations is through different valuations due to information asymmetries. In such a case shareholders value, say vs≠ v instead of v. v is not longer known to
the shareholders.
2.2. Dilution and replacing current management
Da Matos (2001) shows another model to explain the existence of a take over in which current shareholders allow dilution in order to stimulate the acquirer and penalize the current management. This idea is comparable with the idea that a not well managed company is vulnerable for a take over and immediately generates vs≠ v. The degree of dilution is measured
by Φ and defined as Φ = vs–v. The bid of the acquiring party should not be lower than vs. If the
cost to the acquiring party is c, the profit to this party can be expressed as:
Π = v – vs– c (Equation 3)
In this model the current management is able to generate value q, shareholders will therefore not be interested to tender for less than q. Therefore, to maximize the profits to the acquiring party, p = max(vs,q) = max (v - Φ,q) (Equation 4)
Or rewriting with equation 3:
Πa = v – p – c = min(Φ,v-q)-c (Equation 5)
The choice of the current management’s action a, gives rise to the market value q= f(a) is therefore dependent on the level of dilution Φ. Suppose now that the manager has a utility U(q) from the market value q, in case of no take over and zero in case of a takeover. The realizations of v and c are random to the management, given as v and c, and let the probability of a takeover be given by:
To put it in other words, the probability of the takeover is given by the chance that the costs of the takeover are either compensated due to dilution or by the difference between the post takeover value and the value that the current management can realize. It now follows that the expected utility for the current management of market value q is:
W(q) = U (q) [1- φ(Φ,q)] (Equation 7) If v and q where not stochastic, takeovers would occur with probabilities of either one or zero, and since the probability of a takeover is a decreasing function of q, the management would chose
q large enough to prevent takeovers. Randomness of the variables v and c is thus essential to
conclude that takeovers occur in real life.
2.3. The role of a large shareholder
Let’s now consider the role of large shareholders. In the two earlier models it is assumed that shareholdings are widely spread and shareholders are atomistic. The next model has been developed by Schleiffer and Vishny (1986). The model assumes there is one large shareholder L which holds a given minority stake α. Z is denoted as the increase in market value by changing the inefficient management. The model assumes that shareholder L has privileged access to new technology. I is given the probability of getting a positive improvement Z from a distribution F(Z) with [0,Zmax]. I can be interpreted as the research intensity. The cost related to the research are
given as c(I) so that c’>0 and c’’>0. Under these conditions the bidder will make a bid when he can buy fraction 0,5 – α of the shares. q is also here modelled as the market value that the current management can deliver. The bid should be, as already earlier mentioned, higher than this value q. p = q + π (Equation 8)
Where π satisfies the equation:
0.5Z – (0.5 – α)π – c ≥ 0 (Equation 9)
The expected value to the small shareholder is now given as E[Z|0,5Z – (0,5 – α) π – c ≥ 0]. Since this is the case shareholders will tender their shares in case π ≥ E[Z|0,5Z – (0,5 – α) π – c ≥ 0]. The bid premium is higher than the expected value. The role of L in this model is to determine the bid premium in such a way that difference between the premium and the expected value is minimized. It can be resumed that the optimal value depends on fraction α held by L. Let π*(α) be the optimal value for which the difference between the premium and the expected value is minimal. In other words in this case the tendering shareholders will get their maximum payoff. Schleiffer and Vishny (1986) continue to show that π*(α) is a decreasing function of α. Notice that if α=0, the same situation as in previous models is examined, namely largely diffused ownership. In this situation Z > π should hold in order to make a bid successful. When this constraint does not hold an acquirer will not make a bid since it is not value enhancing, and no one will tender their share.
The cut-off value Zc(α) is the value for which the shareholders are indifferent between tendering or not tendering. Zc(α) is a strictly decreasing function. Also the willingness to pay for research
I(α) is increasing with the fraction of shares held by L, in order to increase the probability of
getting a positive value Z, and the lower the premium will be that L is willing to pay the other shareholders. L is in this model the bidding party, it is quite straight forward to understand that it is easier for party L to take control over the firm as fraction α increase. It has a better bargaining position if α increases, and therefore L is willing to pay a lower premium. This is why π decreases when α increases, and the take over becomes more attractive to the bidding party.
Equation 10 shows the impact of fraction α on the market value of the company:
V(α,q) = q + I*(α) {1-F[Zc(α)]}E[Z|Z≥Zc(α)] (Equation 10)
The value of the company increases from the original value q, which can be generated by the current management, by an amount that is dependent on three factors: The optimal probability of getting a positive value Z (I*(α)). The second one is the expected increase in the discounted market value by replacing the inefficient management, E[Z|Z≥Zc(α)]. The third variable is the probability that this threshold holds given by 1-F[Zc(α)]. All three factors are dependent on fraction α. The product of the three variables named is equal to the difference between q and the expected value of the firm after changing its management.
2.4. Toehold as a strategic variable
Fraction α can be considered as the toehold of the bidding party. Schleiffer and Vishny (1986) considered this stake as given. Jegadeesh and Chowdry (1994) go one step further and show that the toehold held by L can be seen as a strategic instrument in the bidding process. They consider α as an endogenous variable that can be used to maximize the firm value. Jegadeesh and Chowdry (1994) assume that the bidding party buys a stake α in the open market at a current market price of zero. The bidder can build up an initial stake in the open market up to αmax, for which count
that αmax< 0,5. The value of α signals the motivation of the bidding party to take control over the
firm.
αmax is in most counties regulated by law. In the Netherlands for example bidders have to report to
a special agency called ‘Autoriteit Financiele Markten’ in case their stake exceeds a certain percentage. These threshold values are 5%, 10%, 15%, 20%, 25%, 30%, 40%, 50%, 60%, 75% and 95%1.When their total stake exceeds 30% of the voting shares, they are obliged to make a bid on the remaining shares according to the European law. It countries that introduced the obligatory bid earlier United Kingdom, Germany, France, Belgium, Spain and Italy, practice showed that obligatory bids are very rare. In most cases the bidder builds up a stake just below the 30% and is therefore not forced to make a bid on the remaining shares. (Tweede Kamer, vergaderjaar 2005– 2006)
Continuing with the model, shareholders are assumed to be risk neutral, which means that they are indifferent between receiving a fixed amount against an uncertain amount with the same expected value, and they will not tender their shares unless the bid equals at least the expected improvements of the profit. Both the toehold and the bid are considered as signals in this model. The free rider condition can be expressed as follows:
π ≥ E(Z|α,π) (Equation 11)
The value of the bid to the bidding party is given by:
This value can be interpreted as the money left to the bidder, half of Z since ex post the bidder has a 50% stake, after paying a premium (0,5- α)π in order to build up a majority stake and take control over the firm.
The maximization problem is given by:
max G(π,α,Z) taking into account π ≥ E(Z| α, π) (Equation 13)
The following assumption is then made to obtain insightful results: The success probability of a bid is increasing and concave with π. This means that the higher the premium the more likely it is that the bid will be accepted. But the success probability will increase with a decreasing pace. The assumption made has also the implication that G is concave in π. This can be proved by differentiating equation 12 twice:
δG/δπ = P’(π) [0,5Z – (0,5 – α)π] – (0,5 – α) P(π) (Equation 14)
and
δ2G/δ2π = P’’(π) [0,5Z – (0,5 – α)π] – 2(0,5 – α) P’(π) (Equation 15)
Since 0,5Z – (0,5 – α)π is positive and α < 0,5, the assumption implies that the outcome of the second derivative is negative, and therefore the function is concave. In other words, the value of the bid to the bidding party increases with the height of the initial holding α.
The next step in the model is to determine the height of the premium such that the profitability G to the bidding party is optimal. This optimal value of π is referred to as π(α,Z) and can be found
by setting equation 14 equal to zero. Replacing π with π(α,Z) and differentiating equation 12 with
respect to α gives:
δπ/δα = - P(π) + P’(π) π / P’’(π)[0,5Z – (0,5 – α) π] – 2(0,5 – α) P’(π) (Equation 16)
In contrast to other models Jagadeesh and Chowdry (1994) continue with a signalling model. Therefore a second assumption is made namely; the constraint caused by the free-rider condition is binding for all bidder types. Formally written as:
Z > π(αmax,Z) > π(α,Z) (Equation 17)
Without the last assumption potential bidders would prefer to bid les than their value of Z. and would as low as π(α,Z). The first result can be derived taking into account that in the absence of the free rider constraint the value of bidding high is increasing in bidder’s type Z. Bidding low reduces the probability of success, which will hurt the high bidder more than the low bidders, since the high bidder has more to gain from a take over.
The second outcome relates to the costs involved to build up the initial stake α. The lower the initial stake the higher the cost to acquire the remaining stake ((0,5- α) π ). However these cost are not influenced by the value of Z, unless this fact there will not exist a pooling equilibrium. Assume that the true Zm lies somewhere between Zl and Zh. In this situation the minimal bid should be at least equal to Zm in order to overcome the free rider problem and the initial stake α plays no signalling role. Now the bidders with Zl will bid above their budget and will compensate this by reducing its initial stake α’. The value of α’ is chosen in such a way that another bidder has no incentive to mimic this behaviour. Therefore α’ must satisfy:
G(Zl, α’;Zh) = G(Zm, αm;Zh) (Equation 18)
When this is the case the low bidder might be better of, given by:
G(Zl, α’;Zl) > G(Zm, αm;Zl) (Equation 19)
Other models state that the higher fraction α the higher the price can be in case of competition, because of the fact that the bidding party holding a toehold will gain on its toehold in case of an unsuccessful bid. This advantage provides the opportunity to the bidder owning a toehold to drive up the bid, and therefore increase the success probability.
Betton and Eckbo (2000) researched the impact of a toehold on the accomplishment of the bid premium; in the next chapter their findings will be discussed.
2.5. A two tier offer
A third way to overcome the free rider problem in case of highly diluted shareholdings is to make a two-tier offer. This concept is introduced by Dunn and Spatt (1984). A two-tier offer consist of two parts, in the first part a bid including a premium is made to a selected set of shareholders, just enough to get control over the firm. Then in the second stage a merger plan will be posted. The three crucial variables in this model are the prices paid in either the first and second stage and the third variable is the fraction of shareholder that receive the first bid.
2.6. The winners curse
The last theory that will be discussed is based on a multiple bidder situation. In 1971 Capen, Clapp and Campbell introduced the Winner’s curse. Simply speaking, the winners curse exists when multiple bidders have to estimate the true value of a good they are bidding for in case of an auction. They first discovered this phenomenon by studying the revenue of oil firms, especially those that had invested in the Gulf of Mexico area. They found that in the era after 1950 those companies failed to outperform the local credit union in terms of return. This phenomenon could not easily be explained by bad luck. Capen, Clapp and Campbell (1971) concluded that these companies invested too much in these activities. Each potential bidding party estimated a private valuation for the target/project based on any believes about synergy effects or potential value creation. In this part of the process the winners curse enters. The estimated value of the target is subject to a lower bound given by the market value, since a lower valuation will not result in a bid. Consider the case that there are no takeover gains, then the valuation of the target firm for the bidding party will be a random variable with the mean being the current market value of the target. The take over premium is a random error in the same direction (Jackwerth, 2002), leading to overpayment.
overpayment of the final buyers. The effect of the winner’s curse increases in case of more bidding parties, because the chance that one party overestimates the true value increases.
Empirical evidence even tends to shows that the combined return of both the acquiring firm and the target is zero or even negative. When this is the case, why are managers than still looking for takeovers? Roll (1993) developed the “Hubris Hypothesis” to give an explanation; his main argument was that managers are overconfident with their abilities.
2.7. Poison pill
A Poison pill is a collective term for different sorts of actions a target company can undertake in order to defend itself against a hostile take over. A company can for example increase its leverage to such a high level that the target becomes unattractive. Another action that can be undertaken is to sell a key business unit, which makes the target less attractive.
The action that is focussed on here is called the shareholder rights plan. In the model described in paragraph 2.2 shareholders are wiling to dilute their right in order to motivate a potential bidder to replace the current management.
The poison pill, in the form of a shareholder rights plan, has an opposite objective and is used to protect the current management. The target company is allowed to issue a large number of usually common stock or preferred stock in order to dilute the holdings of the potential bidder in case of a take over threat. This will immediately lower the value of the bidder’s initial holding and, as has been proved before, will increase the costs to take control over the company. The part over which the acquirer has to pay a premium, (0,5 – α), increased and therefore the cost, (0,5 – α)π.
3. Overview of earlier empirical research
Elizabeth Callison and Linneman (1988) came up with a dissertation quite similar to the problem discussed in this paper. They empirically examined the determinants of the price offered to target shareholders in corporate acquisitions of Forbes 500 firms between 1979 and 1983.
They made a distinction between merger and tender offers. To explain the premium paid in a tender offer they found that target liquidity, dividend payout, size and success probability were significant. For mergers profitability and Market-book value were significant in explaining the height of the premium paid.
The following section provides a comprehensive overview of earlier empirical studies specified after each variable that has been tested. Table 1 provides an overview with the papers that will be discussed.
Authors Year Variables Datasource Sample N
Gort, Hogarty 1970 Relative size Multiple 1948-1968 1280
Bradley 1980 Returns distribution Austin Finch and others 1962-1977 258
Walking, Edmister 1985 Leverage SEC 1972-1978 108
Knoeber 1986 Golden Parachutes Compustat 1980-1982 665
Travlos 1987 Method of payment Multiple 1972-1981 167
Nathan 1988 Market to Book value Moody's, WSJ, Compustat 1963-1978 365
Callison, Elizabeth 1988 Liquidity, size, dividend Forbes 500 firms 1979-1983 93
Robinson, Shane 1990 Accounting method DJ News Retrieval Service 1972-1982 95
Jennings, Mazzeo 1993 Bid structure, competition DJ News Retrieval Service 1979-1988 647
Cotter, Zenner 1994 Managerial Wealth DJ News Retrieval Service 1989-1990 141
Betton, Eckbo 2000 Toehold Mergerstat and others 1971-1990 1353
Table 1: Overview of earlier research
3.1. The toehold as explanatory variable
Walking and Edmister (1985) state in their study that the height of a bid premium is a positive function of the foreseen synergies and a negative function of the bargaining power of the bidder. They ran a model with eight different variables. The target debt to assets ratio trend was used in order to determine the leverage of the firm amongst the company’s net working capital and market to book ratio. Further more they distinguished whether the combination was a conglomerate or a non-conglomerate.
models. Model 1 uses as its prediction for each offer the average bid premium for successful takeovers during the previous 12 months. Naïve model 2 made use of the average premium paid, which was equal to 47,6%. The outcomes of both models were both inferior to that of the bid premium model.
Betton and Eckbo (2000) came up with four major findings; 1) Of the 1,353 initial bidders only 53% had a toehold, and initial bidders’ toeholds are the largest in single-bid contests (around 20%) and lowest in multi-bidder contests (around 5%). 2) In multi-bidder contests the initial bidders and rivals on average have toeholds of similar size. 3) Toeholds and premiums are negatively correlated and toeholds are smaller the greater the pre bid increase in the target share price. I.e. in this case it is much more expensive to build up a toehold. 4) The greater the toehold the greater the success probability and the less likely it is that rivals will enter. Finally the greater the toehold the less likely it is that management resistance will occur.
3.2. Market to book value:
Nathan (1988) points out that the height of the bid premium is significantly dependent on the relative pre-announcement market to book value. Walking and Edmister also used the market to book value as a control variable and found the same result for their sample. The market to book value is defined as the market cap of the target company divided by the book value of the shares outstanding. It is assumed that the lower the M/B ratio, i.e. the closer the market value of the stock lies to the book value of the stock, the more upside potential there is to increase a company’s value. This results in a higher premium.
3.3. Acquisition accounting and the bid premium:
3.4. Method of payment
Niden (1988), Frank et al. (1988) and Asquith et al. (1987) provide strong empirical evidence that when an acquisition is financed with cash the bid premium is higher. Travlos (1987) found the same result in his paper independent of the type of take over bid i.e. merger or tender offer and of bid outcomes.
Da Matos (2001) points out three important considerations that support the payment in cash. First of all taxes. Transactions in cash are taxed. Therefore the tendering party might demand a higher price in order to be compensated for that apparent disadvantage. Among others Hansen (1987) and Hayn (1989) provided evidence of this fact. On the other hand Franks, Harris and Mayer (1988) showed the same results for a UK sample, but before the introduction of law taxing capital gains. The role of taxes might be obvious and helps in understanding the choice of means of payment, but it does not provide a complete explanation.
The second consideration concerning cash payments is provided by both Hansen (1987) and Fishman (1989) and is based on signalling. A bidder that chose to pay in cash may simply mean either that he knows the high potential synergetic value of the transaction or that he believes its own share is undervalued. In both cases, paying with cash sends a positive signal to the market, and signals value potential.
The third reason to pay with cash can be derived from Jensen’s (1986) agency theory. Paying with cash creates value since it limits the agency costs related with free cash flows that otherwise could be spent on unprofitable projects. Since all transactions considered in this paper are cash, above considerations should not influence the results of this paper.
3.5. Managerial wealth
Jensen and Meckling (1976) developed a model in which they describe the potential differences of interest between owners and managers of a firm. Very briefly they create a utility based model that shows the cost of outside equity. The trade off between pecuniary and non-pecuniary returns to the manager deteriorates for the outside owner of the company. The costs of investing in non-pecuniary goods are also carried by the outside shareholder and therefore the manager has an incentive to shift it utilities, compared to the situation in which the manager is also owner of al the shares.
large target shareholder return at the time of an announcement. But managers are not always best served in accepting a bid. They gain financially from the rise in the share price or from golden parachutes (a clause in the manager’s contract specifying that he/she will receive large benefits in the event that the company is acquired and the executive's employment is terminated) but managers suffer other gains like compensation, perquisites and intangible benefits of control in case they are replaced after a successful bid.
4. Position of this paper:
The previous two sections provided an overview of theories and empirical results found to explain the height of the bid premium. A broad spectrum of variables has been researched. But not all explanatory variables mentioned can be included in the model of this paper. As a rule of thumb, a multiple regression model should not include more than 25% of the number of observations and preferably not more than six explanatory variables (Kallenberg, 2004). By picking four important control variables, besides the three pioneering variables, and applying certain criteria for the deals included in the dataset this paper tries to come up with a comprehensive model. The scope of this paper is pioneering since three new variables are tested simultaneously, earlier findings and theoretical models are represented by the control variables. The four control variables are chosen in a way that these should cover as much earlier research as possible. Table 2 provides an overview of the variables described in sections two an three and their role in this paper.
Included in the model* Indirectly Omitted
Toehold Ownership structure Bidder competition (two tier offer, winners curse) Eurostocks-1800 Acquired stake Take over defences
Leverage Accounting method Relative size
M/B ratio Method of payment Type of transaction
Managerial Wealth * Elucidated in section 6.
Table2: Model in relation to theory and empirical findings
The explanation of the control variables included, toehold, Eurostocks-1800, leverage and market to book ratio, in relation to earlier models and empirical finding can be found in section six. By taking the toehold into consideration the role of a large shareholder/ownership structure according to Schleiffer and Vishny (1986) is partly taken into account, in a way that the toehold prior to the rumour date provides some insight in the shareholdings. Additionally the toehold is related to the acquired stake, since the acquired stake can be derived from the final stake minus the toehold. Evidently this variable is indirectly taken into account.
5. Different methodologies to determine the bid premium
The specification of the dependent variable, the bid premium, is very crucial for the outcome of the research. Walking (1985) found that earlier studies came up with insignificant results for bid premiums in the explanation for the success of a bid. The latter result contradicts standard economic theory. If this would be true it would be very hard to explain the wide distributions of bid premiums, or more general, why they exist at all. Walking (1985) invented and tested three ways to measure the bid premiums, which will be discussed later on.
Robinson and Shane (1990) describe three methods to determine the height of the bid premium. The first method presented in the paper derives the premium directly from the bidder’s offer price with the share price 40 days prior to the announcement date.
The second method is somewhat more comprehensive, it determines the premium indirectly from returns on target’s common stock for a pre announcement date to the merger communication date. The last method that is used in the paper makes use of estimates about the market’s expectation of the premium amount from an abnormal returns model that evaluates the market reaction to the news of the bid for the target stock. The market reaction, Cumulative Abnormal Return (CAR), is estimated during a test period beginning 40 trading days before the announcement date and ending on the earlier of the fortieth trading day after the announcement date or the last day of trading of the target’s stock before the merger consummation (Robinson, Shane, 1990)
Callison, Joan Elizabeth (1988) found an average premium of 32.75% over their sample of comparing the offer price with the share price 20 days prior to the first public announcement date. Betton and Eckbo (2000) calculate the premium as (p0 – p-60)/p0, where p-60 is the target share price corrected for dividends and stock splits 60 days prior to the offer date. By calculating the CARs over a period prior to the announcement date Betton and Eckbo(2000) determined the interval on 60 days as a good estimation for a non informative base price. Bradley (1980) came up with 40 days based on the same theory and Swert in 1996 found 42 as the best predictor. OCE2 (1985) measured the premium by comparing the price paid to the bidder with the trade price one month before the offer, not adjusting for changes in the market index (Jarrell, Brickley and Netter, 1988)
Jovanovic and Braguinsky (2000) determined the premium as the price paid by the acquirer compared to the share price 60 days prior to the press day or publication date. They even showed that most of the premium is already included in the prices at the press day.
Cotter en Zenner (1994) used two initial dates in order to come up with the bid premium; first they collect the stock data of the target 30-days before the rumour date, if a rumour was present.
Second, they collect stock price information 30 days before the first tender offer announcement. The choice of a 30 day interval prior to respectively the announcement date or the rumour date is somewhat arbitrary. But they state that earlier studies of takeovers report a run up in target stock prices before the first public announcement date due to leakage of information. Cotter and Zenner (1994) further corrected the premiums for the fact that the bidder might not want to buy all outstanding shares in order to gain control using the same method that Walking (1985) used. Walking and Edmister (1985) calculated the bid premium as the percentage difference between the bid price specified in the offer and the target’s market price 14 days prior to the earlier of (1) the offer announcement date or (2) the official SEC filing date.
As already mentioned, Walking (1985) invented three ways to determine the bid premium. The first one is similar to earlier research, namely the percentage premium based on the market price 14 days before the SEC filing date. The second definition is given as the percentage premium based on the market price 14 days prior to the earliest of SEC3 dates or offer announcement. The last one is corrected for the fact that the acquirer might not want to buy all shares in order to gain control. This correction is based on the idea that a bidder assumes an upward sloping supply curve, which requires the payment of a premium in order to get a sufficient amount of shares to gain control over the target. Several conditions could explain the existence of an upward-sloping supply curve. Heterogeneous expectations among the target firm shareholders would result in different estimates of value and different tendering prices. Walking corrects the premiums using the following equation:
BP*S = ((PS– MS1)/MS1) (SS/ST) + ((MS2– MS1)/MS1) (1-SS/ST) (Equation 20)
Taking al these considerations into account, in this paper is chosen to determine the bid premium as the offer price divided by the share price one day before the rumour date. The rationale behind this choice is that Zephyr determined the rumour date based on the very first date at which the information becomes public. Thereby the average number of days between the announcement day and the rumour day is 60 days according to Jovanovic and Braguinsky (2000) and Betton and Eckbo (2000). This secures a good estimation for a non informative base price, giving the following equation:
Bid premium = (Bid price – Sharepricet-1) / Sharepricet-1 (Equation 21)
t= rumour date
Table 3 provides an overview of the bid premium distributions. Six targets were bought below their market value at the rumour date representing 8% of the sample. Three targets were bought at a premium above 60% compared to the market capitalisation at the rumour date. Furthermore the majority of the transactions took place in the years 2002 and 2005.
Calender Year of Offer Frequency distribution of BP
2000 5% BP ≤ 0.0 8% 2001 14% 0.0 < BP ≤ 0.1 22% 2002 25% 0.1 < BP ≤ 0.2 30% 2003 16% 0.2 < BP ≤ 0.3 12% 2004 10% 0.3 < BP ≤ 0.4 16% 2005 25% 0.4 < BP ≤ 0.5 4% 2006 5% 0.5 < BP ≤ 0.6 3% BP > 0.6 4%
6. Variables to include in the model
In this section the theories and empirical findings about the research specific variables will be elucidated. Section 7 will briefly describe the theories underlying the control variables.
6.1. The concept of loss aversion applied to the take over process:
As already mentioned in the introduction this study tries to provide a link between concepts that found their origin within behavioural finance and corporate finance. In classical economic theory it is assumed that individuals act perfectly rational, the so called homo economicus. With the rise of behavioural finance in the seventies of the past century this fundamental assumption was challenged. Researchers came up with irrationalities and anomalies in economic behaviour. An interesting study was done by Tversky and Kahneman (1979). They came up with a utility based model that described investor behaviour. They found out that investors are risk averse in case of a winning situation and are risk seeking in case of running a loss. Investors consider a non-realised loss as a bookkeeping loss, and are risk seeking in order to compensate this loss.
The concept of loss aversion was introduced by Tversky and Kahneman in 1979. Loss aversion refers to the tendency of investors to be more sensitive to reductions in their wealth levels than to increases. Tversky and Kahneman (1979) came up with a utility function that is based on three essential characteristics. The first one is reference dependence. This means that gains and losses are derived from a certain reference point (in figure 1 given as intersect between the x-axis and the y-axis). The second characteristic is diminishing sensitivity, this means that the marginal value of losses and gains decreases with their size (the slope of the utility function decreases when further from the reference point). The third concept is implicit, loss aversion, the function is steeper in the negative domain than in the positive domain. These three factors together are the basis for the s-shaped utility function as given in figure 1.
U(p)
The prospect theory describes a different risk attitude for investors in both intervals. It can be expected that these behaviours play a role in the process of an acquisition. This paper assumes that it influences the willingness to sell a share against a specific price (i.e. the bid) in case of an acquisition, and therefore should have influence on the height of the bid premium. Investors running a loss will require a higher premium since they are risk seeking. On the other hand, investors that have a gain on their investment are risk averse and are expected to tender their shares and prefer a certain outcome, i.e. the bid. Form the graph this behaviour can be easily related to the height of the premium. Consider the case that an investor has a position in the negative interval. The stock that this investor holds has a probability of 0.5 to go up and a probability of 0.5 to go down. If you draw a line between the share prices on the utility curve and connect this point with the utility equivalent, this new point on the utility curve lies on the right of the original point. i.e. a premium is required. In the positive interval the opposite is true, the utility equivalent now lies left from the original point on the curve. i.e. the risk averse shareholder prefers the certain outcome and is even willing to sell its share against a discount. This case is highly theoretical since, in practice the investor always has the possibility to sell against the market price, though a premium is not required in the positive interval following this theory. The Compounded Total Returns (CTR) over one month, three months and one year before the rumour date are used. This means that the returns are based on share price movements as well as possible dividends. This is a good estimator for the position that the current shareholders have in terms of loss or profit i.e. their position on the utility curve.
6.2. Liquidity of the stock:
Robinson and Shane (1990) state that the bid premium may be influenced by liquidity. They state that Jensen’s (1986) agency problem involving free cash flows is applicable in case of an acquisition. They value liquidity as a positive asset for a company, although there are also negative influences on liquidity (Jensen, Meckling, (1976)). Robinson and Shane (1990) though used Jensen’s (1986) free cash flow model, and determined liquidity using a free cash flow measure.
6.2.1. Liquidity derived from private discount theory:
A lot of research is done to explain the liquidity discount that is often used to acquire a non listed company. Empirical evidence from for example Officer (2005) showed that non listed companies are traded at a discount of on average between 15 to 25 percent.
Officer (2005) states that companies have several opportunities to increase their liquidity. A privately held company can make initial public share offer or borrow money. Officer (2005) focuses on companies that increase their liquidity through selling a privately held subsidiary. The cost of obtaining liquidity by selling a subsidiary is researched and compared with for example costs of an Initial Public Offering (IPO).
Officer found that these cost are on average 15 to 25 percent. The reason of this liquidity discount is that the negotiation power of the selling party is less due to the illiquidity.
Koeplin et al. (2000) state that the correct approach to value a company is the Discounted Cash Flow (DCF) method. However, the application of the DCF methodology itself raises certain unsettled questions, including whether the same discount rate should be applied to private companies as when valuing comparable public companies. The value drivers for a DCF model include risk, growth rate, capital structure, the size and timing of cash flows and liquidity. It is quite hard to find a sufficient methodology in order to quantify the valuation differences between private and public companies. Comparable company transactions are another way calculate the valuation discounts for private companies.
Koeplin et al. (2000) continue with an explanation of the valuation differences, of which the most obvious reason is due to the lack of liquidity that private companies face. Stockholders in public companies can use the open market to convert their investment into cash, where private stockholders lack this outlet. For this non-marketability the share is traded at a discount.
Koeplin et al. (2000) distinguish two approaches to calculate the liquidity discount. The first method used is comparable transactions. Recent comparable market transactions are selected and their valuation multiples are applied to the financials of the comparable private company. Using this method Koeplin came up with an average undervaluation between 20-30%. For the non-U.S. firms in his sample he even found an average undervaluation of 40 to 50%. Koeplin et al. (2000) show that the valuation discount is caused by several factors; in order to quantify the discount specifically caused by the lack of marketability two methodologies can be distinguished.
field is done by Silber (1991), he reports an average liquidity discount of 35%. One major flaw of this kind of studies is that the fact that private discount arises from varieties of factors, of which liquidity is only one. The private discount placement might represent compensation for due diligence by an informed investor, leading to valuable equity financing for a firm with few other alternatives. Also, private placement investors usually commit to be active monitors and provide valuable advisory services. Thus the private placement discount might be an unreliable measure of the discount for illiquidity. (Koeplin et al., 2000)
The second method that can be applied to quantify the liquidity discount is a so called pre-IPO (Initial Public Offering) study. By comparing the price at which the stock was initially offered to the public with the transactions occurred when the company was private one can derive the illiquidity discount. A representative study in this field is done by Emory (1994), who found an average discount of 47%. Critics on this type of study are that most of the transactions showed in a private situation are for restricted options offered to the management, implicit, management compensation is often built into the low transaction prices, and since most transaction are restricted option offerings instead of cash for stock transactions. Finally the most important point of critic is that this type of study suffers a serious selection bias, only companies successful enough to go public are included in the sample, and are therefore in the upper side of the value range.
6.2.2. Liquidity derived from stock split theory:
Another way to look at liquidity can be derived from stock split theory. From an economic point of view a stock split is no more that a cosmetic change, so then why should a company split its stock? The first reason is psychological. When the price of a stock gets higher and higher, some investors may feel the price is getting too high for them to buy, for small investors it may feel that the share is unaffordable to them. Splitting the stock brings the share price down to a more attractive level. The effect here is purely psychological. The actual value of the stock doesn't change since no fundamentals have changed, but the lower stock price may affect the way the stock is perceived and therefore might attracts new investors. Splitting the stock also gives existing shareholders the feeling that they suddenly have more shares than they did before.
Another reason for splitting a stock and thereby increasing the stock's liquidity is that when stocks get into the hundreds of dollars per share, very large bid/ask spreads can be asked. The counterparty runs a potential risk of not being able to sell the stock, and converting its investment back into money. By splitting shares and thereby reducing the risk a lower bid/ask spread is often achieved, thereby increasing liquidity
Lakonishok and Lev (1987) state that stock splits are executed by firms that have enjoyed an unusual growth in earnings and stock prices. But an empirical reaction to a stock split in favour of its marketability has never been given. Both Lakonishok et al. (1987) and Copeland (1988) found that a stock split did not increase the liquidity. Lakonishok et al. (1987) state that the hypotheses around this phenomenon can be classified into two groups, signalling and optimal price. Leland and Pyle (1977) introduced a model based on asymmetric information between managers and investors. The manager might use financial decisions, such as stock splits, to convey favourable inferable information to the investor. However, for a signalling device to be valid there should be costs involved in sending false signals. This makes it unattractive for firms with bad performance to mimic this behaviour. The case for costly signalling by stock splits is unclear; therefore this theory is not very strong in explaining the existence of stock splits.
The last reason that is argued in order to explain the existence of stock splits is that certain industry norms for financial ratios create an incentive for companies to split their shares. Managers adjust the price of their stock to that of comparables in order to align financial ratio’s, like for example the dividend per share (DPS) ratio and the earnings per share (EPS) ratio.
None of these reasons aligns with financial theory. Although empirical research has never clearly proved that a stock split increases the liquidity of a stock, it is commonly believed that the lower the share price the higher the liquidity if the stock should be.
7. Theory behind the control variables
The model used includes four control variables. First of all a business cycle indicator is included in order to determine whether there is a relation between the height of the premium and the economic situation. In this model the Eurostocks-1800 is used as an indicator. The nominal value of the index is regressed on the height of the bid premium. The index consists of 1800 different listed entities throughout the Euro zone. This is the most representative index since the dataset includes 73 transactions within the Euro zone from 2000 to November 2006. The correlation between the height of the Eurostocks-1800 and the height of the bid premium is only -.03. The coefficient in the regression can be interpreted as the beta of a stock since the formula of the coefficient is the covariance between the premium paid and the Eurostocks-1800 divided by the variance of the Eurostocks-1800, which is the same formula as to determine the beta of a stock. But in this particular case the share price is influenced by the bid. In other words it is not relevant anymore to compare return in these time periods with the return on the market since the share price will rise up until the height of the bid and won’t rise further, since it is not attractive for a shareholder to buy at a higher price than the bid. Arbitrage opportunities will clear the market and set the share price equal to the bid price. Therefore the return expressed as the premium over this period is not comparable with the market return. The Eurostocks-1800 index is included in the model as a business cycle indicator. In case of high return on this index the economy is expected to be in an upturn, resulting in large free cash flows to fund potential acquisitions. When this is the case more bidders are expected to drive up the price and therefore the premium paid.
Walking and Edmister (1985) argue that the toehold, the preannouncement holding in the target company, plays an important role in the bidding process and in the determination of the premium. This initial holding is inversely related to the bid premium because it improves the bidder’s bargaining power. In section 2 several theories relating to toeholds are discussed. The implications were ambiguous. In section 2.4 Jegadeesh’ and Chowdry’s (1994) signalling theory showed that the lower the toehold, the lower the premium should be, all other theories state the opposite.
Leverage should be taken in to account as well. According to Jensen (1986) a company with a high liquidity has a lower value since there will occur more agency problems. Managers have to reallocate large free cash flows, accompanied by large monitoring and bonding costs. Walking and Edmister (1985) included leverage in their model taking the debt to assets ratio, and found an inverse relation that was significant on a 1-percent level.
According to Gort and Hogarty (1970) the relative size between the target and the acquirer should be taken into account as well. They found that relative larger target received lower premia. Frank and Harris (1988) came up with the same result using a large sample of UK companies between 1955 and 1985. Due to a lack of data availability and the bounded number of explanatory variables that can be included in the model, this variable is omitted from the regression.
The last control variable in the model is the M/B ratio. A relatively low pre-acquisition market-to-book ratio of target equity may indicate management inefficiency. As Jensen (1986) described replacing inefficient management might increase firm value. Nathan (1988) empirically tested this relation and found in his paper that there exists a positive relation between the M/B ratio and the premium paid.
8.
Problem, sub questions and hypotheses
This thesis tests the assumed relations using Ordinary Least Square Regression. This means that for each variable it will test whether there exists a significant relation between the premium paid and the explanatory variables. Further more the expected sign in the regression should
accommodate with the hypotheses. In section 6 and 7 the theories underlying the explanatory variables were elucidated, table 4 provides an overview of the expected signs for every separate variable.
Variable Expected sign
Toehold -Eurostocks-1800 + M/B ratio -Leverage -Aquired stake + Shareprice -Liquidity + CTR 1 Month -CTR 3 Month -CTR 1 Year
-Table 4: Overview of expected signs in the regressions.
8.1 Hypothesis for the liquidity and prospect theory: 8.1.1 Share price
Based on the stock split theory described in section 6.2.2, a negative relation between the average share prices one year prior tot the rumour date and the bid premium is expected. The lower the share price the higher the liquidity, and the higher the liquidity the higher the premium. The correlation between the share price and the bid premium is -0.26, this indicates that the hypothesis might hold.
8.1.2. Liquidity
8.1.3. Compounded total return
To test for the applicability of the prospect theory on the bidder process, the compounded total return for the target shares, 1-month 3-months and 1-year prior to the rumour date are tested. According to the theory there should be an inverse relation between the compounded total return and the height of the premium. The lower the compounded total return on the stock prior to the acquisition the higher the premium should be. Since there is an overlap in the determination of the compounded return over the different intervals, three independent regressions are used.
Furthermore a distinction between negative and positive returns is made as the prospect theory prescribes. To test whether the premiums are significantly lower for the negative return numbers, Dummies are included to the model in order to test for a breakpoint in the dataset around zero (CTR). In case both the dummy and the CTR variable are significant, there is a breakpoint in the dataset around zero. According to Tversky and Kahneman’s (1979) prospect theory the slope in the negative interval should be twice as steep as for the positive interval.
8.2 Control variables:
As we saw in section 2.2 the models supporting the influence of a toehold are ambiguous. Many theories state that there should be an inverse relation between the toehold and the premium paid. Jegadeesh and Chowdry (1994) developed a signalling model that comes to the opposite conclusion. Betton and Eckbo (2000) empirically found results supporting the inverse relation; therefore it can be expected that the sign in the regression will be negative A first indication given by the correlation of -0,17 shows that there is indeed a inverse relation.
The Eurostocks-1800 index is expected to have a positive relation with the bid premium. The influence of the Eurostocks-1800 is tested by benchmarking the premium paid with the return on the index at the particular date. The return is benchmarked to the Eurostock-1800 index at the first transaction date. When the return is high, the economy is in an upward business cycle and premiums are expected to be higher. The correlation of -0.03 does not support the initial hypothesis.
Jensen’s (1986) theory about the M/B value as a measure of inefficient management is supported by the empirical findings of Nathan (1988). It is supposed that the M/B ratio has an inverse relation to the bid premium, and should therefore have a negative sign in the regression. The correlation of -0.26 confirms this hypothesis.
Following Jensen’s (1986) free cash flow theory there should be an inverse relation between the leverage of a target company and the premium paid. Walking and Edmister (1985) found an inverse relation that was significant on a 1-percent level. The correlation between both variables of -0.09 provides the same insight.
8.3 Testing the coefficients:
The t-distribution provides the foundation for testing hypotheses about individual coefficients in the regression. By setting up a multiple regression model it is believed that all the explanatory variables influence the dependent variable, the bid premium. To confirm this belief, it should be supported by the data. If a certain explanatory variable has no bearing on the bid premium, then βk = 0, k Є (1,8). Testing this hypothesis is called a test of significance for explanatory variable xk.
For each of the above variables specified we formally test the hypotheses:
H0: βk = 0 (Hypothesis 1)
H1: βk≠ 0
8.4 Testing the model:
The above mentioned hypotheses test whether the dependent variable, the bid premium, is related to a particular explanatory variable xk using a t-test. Additional an F-test is used to test the overall
significance of the model. This is a joint test of the relevance of all the included explanatory variables. To examine whether the regression provides a viable explanatory model the following hypothesis are set up:
H0: β1=0, β2=0, β3=0, β4=0, β5=0, β6=0, β7=0, (β8=0)4 (Hypothesis 2)
H1: at least one of the βk is nonzero
The joint hypothesis states as a conjecture that each and every one of the parameters βk is zero. If this null hypothesis is true none of the explanatory variables influence the bid premium, and thus the model is of little or no value. If the alternative hypothesis is true, then at least one of the parameters is not zero and thus one or more of the explanatory variables should be included in the model. The F-test implies that the coefficients are jointly significant, and therefore the model is viable. In table 6, the p-values belonging to the F-statistics are given between brackets; this represents the probability that the null hypothesis is incorrectly rejected. When this probability is below 5% we accept the alternative hypothesis.
4β
9. The Dataset:
The selected deals were found on Zephyr by searching at diverse criteria. The final sample contains 73 transactions throughout the Euro zone from 2000 until November 2006. The dataset corrected for outliers resulted in exclusion of five observations. Due to this correction the residuals of the models used are normally distributed, which is one of the key assumptions underlying OLS-regression. Countries included are Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, The Netherlands, Luxembourg, Portugal and Spain. These counties and start date are chosen in order to avoid currency adjustments. In each transaction the acquiring company gets control over the firm by raising its final stake above the minimum of 50,1%. All transactions included in the dataset are completed and paid in cash, and all target companies were listed. The acquired stake is easily derived from the final stake minus the toehold held. An overview of the deals included can be found in table I of the appendix.
Data about share prices, volumes traded, Eurostocks-1800 and market capitalizations are provided by Factset5.
In order to derive the liquidity and share return, daily data on the share price and volume one year before the rumour date are gathered. Only trading days are included resulting in a sample of 258 observations, approximately one year. The average daily volume one year prior to the rumour date is divided by the number of shares outstanding. The number of shares outstanding is simply determined as the market capitalization divided by the share price.
The CTR’s are corrected for dividend payments and are measured on a 1-month, 3-month and 1 year interval. Data about the toehold, market-to-book ratio and financial leverage are found in Zephyr. The M/B ratio and leverage are derived from the latest available financial statement before the rumour date of each transaction. Total assets where given as well as shareholder equity. The book value of debt could easily be derived from these two variables. Leverage is determined as total debt divided by total assets, according to Walking and Edmister (1985). The equity value of the transaction corrected for the premium over the share price at the rumour date gives the market capitalisation at the rumour date for the firms that were not included in the Factset database. This amount is divided by the book value of equity as given in the latest available financial statement before the rumour date to come up with the M/B ratio. One could comment that this estimate does not perfectly equal the actual market to book value, since
shareholder capital also includes reserves, while the market to book value should be determined over the book value of the shares outstanding.
Variable n Mean Std. Dev Minimum Median Maximum P-value
Premium* 73 20,26 18,83 -20,85 16,79 82,17 0,000 Toehold* 73 13,57 18,19 0,00 0,00 48,3 0,003 EuroStocks-1800 73 103,28 1,68 100,00 103,26 105,95 0,129 Acquired stake* 73 82,08 22,15 21,30 94,81 100,00 0,004 M/B ratio 73 2,45 2,65 -5,60 1,91 14,57 0,000 Leverage* 73 59,33 23,18 6,48 58,63 103,46 0,507 Average Shareprice 73 20,40 21,60 0,18 12,02 111,14 0,000 Liquidity* 73 1,75 1,63 0,02 1,20 7,34 0,000 CTR- 1 month* 73 15,52 19,90 -20,46 11,63 72,73 0,009 CTR- 3 months* 73 22,22 29,82 -45,00 18,19 97,58 0,144 CTR- 1 year* 73 18,84 44,15 -91,85 15,66 114,13 0,986 * In percentage
Table 5: Descriptive statistics for the variables used.
10. Methodology
In this thesis the bid premium is defined as the price that the acquirer is willing to pay for the target divided by the value of the outstanding shares at the rumour date. The model used to explain this premium is an Ordinary Least Square (OLS) regression model. The general form will be:
Bid premiumi= α + Σ β factorx + εi (Equation 22)
In order to overcome possible multicollinearity table II of the appendix provides an overview of all cross-correlations between the variables used. None of the explanatory variables has a correlation exceeding the threshold of 80%, except toehold and acquired stake. With a correlation of -0.90 problems concerning multicollinearity can be expected. This is a logical result since toehold and acquired stake depend on each other. The acquired stake is calculated by reducing the final stake with the toehold. To overcome this problem only toehold is included in the regression. The complete models will look like:
BP = α - β1X1+ β2 X2 - β3 X3– β4 X4– β5 X5+ β6 X6 + β7X7 + εi (Equation 23)
For Which:
X1 = Toehold X5= Average Share price X2= Eurostocks-1800 X6= Liquidity
X3= M/B ratio X7= Compounded Total Return (1-month, 3-months or 1-year)
X4= Leverage α = Constant
Also other functional forms are tested. Table III in the appendix provides an overview of the results of different model specifications.
When the coefficients for both the dummy variable and the CTR are significant one can conclude that there exist a significantly different relation for both the positive and the negative interval. The model is now given by:
BP = α - β1X1+ β2 X2 - β3X3– β4 X4– β5 X5+ β6 X6 + β7X7 + β8X8 + εi (Equation 24)
For Which:
X1 = Toehold X5= Average Share price X2= Eurostocks-1800 X6= Liquidity
X3= M/B ratio X7= Dummy
X4 = Leverage X8 = Compounded Total Return (1-month, 3-months or 1-year) α = constant
11. Presentation of the results:
In this section the results are presented. The first section provides the results and interpretation for the first three models. Due to the overlap in the 1-month, 3-months and 1-year Compounded Total Return it was necessary to run three separate regressions. In section 11.2 al regressions are specifically tested for breakpoints around the CTR level zero in order to test whether the prospect theory holds.
11.1. Premium explained:
