The emergence of multiple-bidder takeover contests and the shareholder wealth effects

(1)

The emergence of multiple-bidder takeover

contests and the shareholder wealth effects

Author: R.M. Rijk

Student number: 1323318

E-mail address:

r.m.rijk@student.rug.nl

Telephone number: +31 (0)6 50 26 18 88

Supervisor: dr. ing. N. Brunia

(2)

2 Abstract

The purpose of this paper is to analyze the emergence of multiple-bid takeovers and multiple-bidder takeover contests and the effects on shareholder returns. The predictions of the preemptive bidding theory (Fishman, 1988 and Fishman, 1989) are tested. Mergers and acquisition from different countries in the period 1997-2006 are considered. The event study methodology is applied to determine the harmonized cumulative abnormal returns in single-bid takeovers, multiple-bid takeovers, and multiple-bidder takeover contests. The hypothesis of the preemptive bidding theory with respect to the emergence of multiple-bidder takeover contests are all rejected. However, with respect to the

abnormal returns, evidence is found that target firm shareholders earn significantly higher abnormal returns in single-bid takeovers than in multiple-bidder takeover contests.

(3)

Table of contents

1. Introduction 4

2. Literature 7

2.1 Overview of theory 7

2.1.1 The English auction 7

2.1.2 The preemptive bidding theory 7

2.1.3 Some notes in analyzing expected profits in the preemptive bidding theory 9 2.1.4 The expected profits in the preemptive bidding theory 10 2.2 Alternative models of multiple-bidder takeover contests 11

2.3 Empirical evidence 13

2.3.1 Variables explaining the emergence of multiple-bidder takeover contests 13 2.3.2 Abnormal returns in single-bid takeovers versus multiple-bidder takeover

contests 14 3. Data 17 3.1 Sample selection 17 3.2 Collection of data 19 3.3 Descriptive statistics 20 4. Methodology 23

4.1 The emergence of multiple-bidder takeover contests 23

4.1.1 The initial premium 23

4.1.2 The costs of acquiring information 24

4.1.3 The method of payment 25

4.1.4 Cross sectional analysis 25

4.2 The event study methodology 26

4.2.1 The event and the event window 27

4.2.2 The normal returns 28

4.2.3. The standardized cumulative abnormal returns 30

5. Results 33

5.1 The emergence of multiple-bid takeovers and multiple-bidder takeover contests 33

5.1.1 The initial premium 34

5.2 Abnormal performance 38

5.2.1 Abnormal performance of the target firm 39

5.2.2. Abnormal performance of the first bidder 40

(4)

4 1. Introduction

In an attempt to acquire a firm, it is possible that the initial offer is not the one that leads to an acquisition. Target firm management could oppose against the offer, which will often lead to a subsequent bid by the same bidding firm or a new bidder (Jarrell, 1985). Furthermore, the bidding firm can make a new offer if there is a threat of a second bidder to make a bid (De Sankar et al., 1996). Cases in which one bidder makes more bids to acquire the target company are called ‘multiple-bid takeovers’.

However, it is possible that a second bidder gets involved. This means that, after the first bidder, a second bidder will make an official offer to acquire the target company. The second bidder sometimes serves as a ‘white knight’, a friendly alternative against a hostile takeover bid. The different bidders can bid against each other. If the shareholders of the target firm are rational investors (i.e. utility maximizers), the highest bidder will win the takeover contest and takes control over the target firm. Contests in which two bidders are competing (i.e. both firms conduct at least one official offer) to acquire a firm are called ‘multiple-bidder takeover contests’.

Recently, multiple-bidder takeover contests gained much public attention (e.g. ABN Amro, Corus). Off course, it is not always the case that more bids are necessary to acquire a company. To answer the question why and in what circumstances multiple-bidder takeover contests are observed and what the effects are on the shareholder returns, several theories are developed (e.g. Baron, 1983; Giammarino and Heinkel, 1986; Fishman, 1988; Fishman, 1989).

(5)

offered in single-bid takeovers is higher than in multiple-bidder takeover contests. Moreover, in single-bid takeovers, the initial bid is more often in cash than in multiple-bidder takeover contests. Finally, the costs of acquiring information are expected to be higher in single-bid takeovers, compared to multiple-bidder takeover contests.

With respect to the shareholder returns, the preemptive bidding theory analyzes the target company and the first bidder (i.e. the firm conducting the first bid). Only first bidders acquiring the target firm are considered here. The preemptive bidding theory predicts that target firms will earn higher profits in single-bid takeovers than in multiple-bidder takeover contests. Furthermore, expected profits are higher when the initial offer is in cash, compared to offers in shares. Finally, target firms are expected to profit from a decline in the costs of acquiring information. First bidders are expected to have higher profits in single-bid takeovers than in multiple-bidder takeover contests. Profits are expected to be higher when the initial offer is in cash, compared to initial offers in shares. Furthermore, first bidders are expected to profit from an increase in the costs of acquiring information.

To test the predictions by the preemptive bidding theory, a sample of mergers and acquisitions from different countries in the 1997-2006 period is used. Other studies on multiple-bidder takeover contests focus on the United States only. The event study methodology is applied to test the predictions of the preemptive bidding theory. The methodology in this study differentiates from other studies in 3 ways:

1. Normal returns are estimated using an extended market model, including the world market index, the domestic market index, and the exchange rate.

2. Because of differences in the time between the first bid and the final bid, a harmonized cumulative abnormal return (HCAR) is computed.

(6)

6 In this study, the results show that the initial premiums are not relevant in the emergence of multiple-bid takeovers or multiple-bidder takeover contests. Moreover, an initial offer in cash will result more often in multiple-bid takeovers or multiple-bidder takeover contests than an initial offer in shares. Furthermore, low costs of acquiring information will lead to more multiple-bidder takeover contests than high costs of acquiring information. The latter result is also significant in a cross-sectional analysis.

With respect to the abnormal returns, the results for target companies show significantly higher abnormal returns in single-bid takeovers, compared to multiple-bid takeovers and multiple-bidder takeover contests. An initial offer is shares results in significantly lower abnormal returns to the target firm than an initial offer in shares. Finally, low costs of acquiring information will lead to significantly higher abnormal returns for the target company.

The abnormal returns of the first bidder are not influenced by multiple-bid takeovers or multiple-bidder takeover contests. Furthermore, the initial method of payment does not have impact on the abnormal returns. When costs of acquiring information are low, the first bidder will earn significantly lower abnormal returns. This relationship disappears when only successful first bidders are considered. However, in a cross-sectional analysis there is a significant relationship between the abnormal returns of the (successful) first bidder and the costs of acquiring information.

(7)

2. Literature

2.1 Overview of theory 2.1.1 The English auction

A multiple-bidder takeover contest is similar to an English auction. In an English auction, several bidders can make an offer to acquire an item. Bidders submit successively higher bids until no one will increase the bid. The item is sold to the highest bidder. An English auction takes in general several minutes. Bidders can acquire information about the item before the auction starts. During the auction, the only additional information can be learned from the bidding itself (Fishman, 1988). On the other hand, a multiple-bidder takeover contest takes several weeks (Bradley et al., 1988). Observing an offer, bidders can learn about the other firm’s valuation of the target company. At a minimum they learn that the valuation equals or exceeds the bid. In multiple-bidder takeover contests, bidders can acquire information after observing an offer. This possibility has an impact on the way bidders act in a multiple-bidder takeover contest. The preemptive bidding theory (Fishman, 1988 and Fishman, 1989) is an example of a theory explaining the behavior of bidding firms in a multiple-bidder takeover contest.

In the following section the preemptive bidding theory (Fishman, 1988 and Fishman, 1989) is explained in further detail.

2.1.2 The preemptive bidding theory

(8)

8 bidder makes a bid. Fishman (1988) assumes that the management of the target and both bidders act to maximize the expected wealth of their respective shareholders.

Before making a bid, the first bidder has to identify whether the acquisition of the target is potentially profitable. The potential profitability is indicated by the random variable s~ , which is denoted by s+ when the acquisition is potentially profitable and s- when it is not.

If, and only if the acquisition is potentially profitable (s~= s+), the first bidder will incur information costs (c1 ) to observe the private valuation of the target (~v ), as proved by 1 Fishman (1988). Only the first bidder knows this private valuation. The higher the expected synergy gains, the higher v will be. If ~1 ~v is higher than the current valuation of 1 the target company (v ), the first bidder will conduct an offer (p). 0

(9)

The first bidder knows that the second bidder knows from the bid (p) made, that ~v >₁ v₀ and that the second bidder will make an offer if v > p. The first bidder can take into ~₂ account this behavior when the first bid is made. The higher the first bid (p), the higher the signaled ~v and the smaller the probability that 1 ~v > p, and thus the smaller the 2 probability that a multiple-bidder takeover contest emerges. If a preemptive bid is made, the signaled v is such that the second bidder will not make an offer (i.e. the expected ~1 payoff from competing is negative). Fishman (1988) proved that in equilibrium, the first bidder will make a high, preemptive premium bid that deters the second bidder from competing (v is ‘high’ in this case) or a zero or low premium bid resulting in an English ~₁ auction (v is ‘low’). ~₁

The premium offered by the first bidder is not the only variable that signals ~v . The ₁ method of payment of the initial offer (cash or shares) also contains information about ~v ₁ (Fishman, 1989). If the bidder is uncertain about the value of ~v , an offer in shares will 1 cause the shareholders of the target firm to share the risk that the bidder may have overpaid, since the value of the shares will decline in that case (Hansen, 1987). The higher the value of v , the lower the probability that the bidder overpays, given the value ~₁ of p. Therefore, the higher the value of v , the fewer offers will be in shares. This implies ~₁ that an initial offer in shares signals that ~v is lower than a cash offer. Fishman (1989) ₁ shows that in equilibrium, ‘high’ valuing bidders (i.e. bidders making an preemptive bid) will always offer cash and ‘low’ valuing bidders will always offer shares. Therefore, more multiple-bidder takeover contests are expected if the initial offer is in shares.

2.1.3 Some notes in analyzing expected profits in the preemptive bidding theory

(10)

10 making a high, preemptive bid. To analyze whether the strategy of preemptive bidding will result in higher expected profits for the first bidder, both successful and unsuccessful first bidders should be used.

In Fishman (1988), expected profits observing high and low initial premiums are analyzed. It is assumed that the fist bidder acquires the target company, also in multiple-bidder takeover contests. So, in order to test the predictions of the preemptive bidding theory, the first bidding firms that acquire the target should be used. These are the ‘successful first bidders’.

Fishman (1988) identifies problems in distinguishing between single-bidder and multiple-bidder takeover contests (single-multiple-bidder takeovers are takeovers in which one multiple-bidder is observed). It is possible that a second bidder, after observing p, pays c to observe 2 ~v 2 and not conducts an offer, because ~v < p. p is in this case not high enough to deter the 2 second bidder from considering an offer, however, it is high enough to deter the second bidder from actually making an offer. When the second bidder does not make a bid, no English auction will emerge and there is no multiple-bidder takeover contest. If p is not high enough to deter the second bidder, the fist bidder can revisit its offer before the second bidder makes a bid. The threat of the second bidder to make an offer is enough to revisit the initial offer in this case. To test whether the threat of a second bidder has the same effect as a bid by the second bidder, the predictions of the preemptive bidding theory are also tested using multiple-bid takeovers in stead of multiple-bidder takeover contests.

2.1.4 The expected profits in the preemptive bidding theory

(11)

and ‘low’ in multiple-bidder takeover contests. The higher the value of v , the more the ~₁ bidder is prepared to acquire the target company (because of higher expected synergies). Therefore the expected profits for the target company are higher in takeovers with single, preemptive offers, compared to targets in multiple-bidder takeover contests.

Fishman (1988) also shows that expected profits for first bidders acquiring a target, are higher in a single-bid takeover than in multiple-bidder takeover contests. If this was not the case, the first bidder would never make a preemptive bid (since it is assumed that the management teams act to maximize the expected wealth of their shareholders). The effect of the higher premium paid is more than offset by the effect that bidders with ‘high’ valuation of the target make preemptive bids.

Since initial offers in cash are expected to result in more than offers in shares, Fishman (1989) shows that initial offers in cash will result in higher profits than initial offers in shares. This is the result for target firms, as well as successfully first bidders.

As already noted, Fishman (1988) shows that the probability of single-bid takeovers is positively related to the costs of acquiring information of the second bidder. Therefore, the higher c , the higher the expected profits for the target company and the successfully 2 first bidder.

2.2 Alternative models of multiple-bidder takeover contests

This section presents some alternative theories with respect to multiple-bidder takeover contests.

(12)

12 management has private information about the value of the target company. Baron (1983) shows that the effect on the target returns of the rejection of the first bid, depends on the reason why the bid was rejected. Investors can obtain the reason by analyzing the information revealed at the day of the acquisition termination. If target management has a preference for control, the market value of the target firm will decline. If target management rejects the offer because it is too low given the true firm value, or if management wants competition to emerge, the market value of the target firm will increase.

In Giammarino and Heinkel (1986), a bidder can conduct only one bid; the second bidder can also make an offer if the bid from the first bidder is not rejected by the target firm management. In this model, there is one bidder with an informational advantage over both the target and the other bidding firm. The informational disadvantage of the other bidder is compensated by a tactical advantage. The uninformed bidder is allowed to make the final bid, and the informed bidder can only choose bids from a limited set of options. Because of this tactical advantage of the uninformed bidder, the informed bidder will make a high, preemptive bid to prevent the uninformed bidder to make the final, winning, bid.

In Khanna (1997) bidders arrive sequentially and are allowed to make any number of bids. After a bid is made, the tender offer bids must remain open for a minimum period (Jensen and Ruback, 1983). With a high, preemptive bid, a bidder can end the takeover process. This result is obtained, since bidders will not enter the contest when the present value of the synergy gains is smaller than the price necessary to acquire the target. The first bidder is always expected to conduct a preemptive offer. Khanna (1997) shows that this strategy is superior to other strategies (e.g. conduct a low offer and raise it over and over again).

(13)

more firms are bidding to acquire the company, the higher is the probability of overbidding. Moreover, if there is competition between bidders to acquire the target, winning the contest could become more important than acquiring the target, because the reputation of the manager depends on the outcome of the merger contest (at least in the eyes of the manager himself). This reasoning is in line with the hypothesis that management will not always act to maximize shareholder’s value and have a preference for control (Jensen and Meckling, 1976). The winner’s curse hypothesis predicts lower abnormal returns to bidding firms acquiring a target in multiple-bidder takeover contests than in single-bid takeovers. Opposite result are predicted for the target company.

The most important argument to use the preemptive bidding theory (Fishman, 1988 and Fishman, 1989) in this paper is the presence of information acquisition costs. In a numerical example, Fishman (1988) proves that ,,even with very small costs of acquiring information, preemptive bidding can be an economically important phenomenon”. Furthermore, the model has realistic assumptions with respect to the number of bids a bidder can make (unlimited). Finally, the model allows testing the difference between two strategies: offering a high initial premium to deter the second bidder, or offering a low premium which will cause a second bidder to make a competing bid.

2.3 Empirical evidence

2.3.1 Variables explaining the emergence of multiple-bidder takeover contests

(14)

14 relationship with the initial premium. However, Michel and Shaked (1988) found no significant differences in cumulative returns between single-bid takeovers and multiple-bidder takeover contests, up to the date of the announcement of the first bid. This result indicates the probability of observing multiple-bidder takeover contests is not related to the initial premium, which is in contradiction to the preemptive bidding theory (Michel and Shaked, 1988).

Jennings and Mazzeo (1993) found a significantly positive relationship between the costs of acquiring information and the emergence of multiple-bidder takeover contests. Jennings and Mazzeo (1993) used two different proxies for the costs of acquiring information: the number of analysts following the target and the exchange-listing status of the target firm’s common shares (quoted or unquoted). The proxy used did not have impact on the results. The results correspond to the preemptive bidding.

The results in Jennings and Mazzeo (1993) with respect to the method of payment do not correspond to Fishman’s (1988) prediction. Jennings and Mazzeo (1993) found a significantly positive relationship between initial cash offers and the emergence of multiple-bidder takeover contests. This relationship is insignificant in a cross-sectional analysis using the costs of acquiring information and the premium in addition to the method of payment. Because Fishman (1989) assumed that the target firm has serious impact on cash flows following the merger, Jennings and Mazzeo (1993) re-estimated their equation using large targets only, but results remained unchanged.

2.3.2 Abnormal returns in single-bid takeovers versus multiple-bidder takeover contests

(15)

correspond to higher profits. By assuming that the emergence of multiple-bidder takeover contests is independent from the pre-merger value of the target firms and the bidding firms, empirical papers (e.g. Bradley et al., 1988) justify the use of abnormal returns. This assumption should be taken into account in analyzing the empirical results.

Bradley et al. (1988), using a sample of 163 single-bid takeovers and 73 multiple-bidder takeover contest in the 1963-1984 period, found that target abnormal returns in multiple-bidder takeover contests are significantly higher than target abnormal returns in single-bid takeovers. This result is contradictory to the preemptive single-bidding theory. The result corresponds to the winner’s curse hypothesis (Varaiya, 1988). Moreover, Bradley et al. (1988) explain that the average time needed to acquire the target shares is longer in multiple-bidder takeover contests than in single-bid takeovers. If not all the target shares are acquired in a single-bid takeover, the remaining shares show negative abnormal returns (Bradley et al, 1988). Since the time period under study in Bradley et al. (1988) is until 5 days after the announcement of the first bid, this explains a part of the difference between abnormal returns in single-bid takeovers and multiple-bidder takeover contests.

(16)

De Sankar et al. (1996) did not find significant differences in abnormal returns between first bidders (both successful and unsuccessful) in multiple-bidder takeover contests and successfully bidders in single-bid takeovers (using the same period as Bradley et al., 1988). This indicates that it does not matter whether a high or a low premium is offered. Again, not the whole contest period is considered here. De Sankar et al. (1996) did find significantly lower returns for successful bidders in multiple-bidder takeover contests than for successful bidders in single-bid takeovers. Also second bidders are included in this analysis. This result is in line with the winner’s curse hypothesis (Varaiya, 1988). However, Bradley et al. (1983) found significantly higher returns for bidders winning the contest, compared to bidders losing the contests. Bradley et al. (1983) argued that the acquisition is expected to result in a competitive advantage for the rival bidder (the synergy hypothesis).

To test the effect of a change in the costs of acquiring information on the abnormal returns, the passage of the Williams Act in 1968 is often used (e.g. Bradley et al, 1988; De Sankar et al, 1996). The Williams Act requires bidding firms to provide detailed information about how the tender offer will be financed and what changes in the operations of the target will be made when the offer is successful. Furthermore, the tender offer should remain open for a number of days, before the bidder can acquire the shares of the target company. The Williams Act is expected to lower the costs of acquiring information. Bradley et al. (1988) found that, after the passage of the Williams Act, target companies earned significantly higher abnormal returns. Bidding firms earned significantly lower abnormal returns after the passage of Williams Act. However, Bradley et al. (1988) did not distinguish between first bidders and second bidders. De Sankar et al. (1996) used first bidders only and also found significant differences in abnormal returns between the period before and after the passage of Williams Act. These results support the preemptive bidding theory.

(17)

when the offer is in shares only, compared to offers in cash. This result is in line with the preemptive bidding theory. Among others, Travlos (1987) also found lower abnormal returns for the bidding firm when the initial offer is in shares, compared to offers consisting of cash. Travlos (1987) argues that the bidder will only offer shares if its shares are overpriced in the market. This signal will cause the share prices of the bidder to decline. Target firm returns are analyzed by Wansley et al. (1983). They found significantly higher abnormal returns for target firm shareholders in cash offers, compared to offers in shares. An explanation is that cash offers are immediately taxable; therefore a premium is paid to the target firm shareholders if the method of payment is cash (Wansley et al., 1983).

3. Data

In this study, I selected mergers and acquisitions in the period 1997-2006. First, I selected multiple-bid takeovers and multiple-bidder takeover contests. Second, single-bid takeovers were selected. The way in which these samples were selected is described in section 3.1. After that, I determined the initial method of payment (cash, shares, or a combination) in all the deals and the country in which the target company is located. Than, to compute the premium initially offered, information with respect to the target firm share price and the offer price is collected. Finally, to determine the abnormal returns, I collected data on share prices, exchange rates and market indices (section 3.2).

3.1 Sample selection

First, all mergers and acquisitions in the period from 1st January 1997 to 31st December 2006 are obtained from the ZEPHYR database. Both accepted and rejected offers are included. The following selection criteria were set on the initial sample of bids:

(18)

18 which is necessary to implement a more profitable operating strategy than the current one.

• The bidding firm(s) and the target firm are all quoted on a stock exchange. This restriction is made, because it is not possible to calculate abnormal returns for unquoted companies.

• The minimum deal value is 250 million U.S. dollars. This value is arbitrary, but it is assumed that from this value the mergers and acquisitions are large enough to have serious impact on bidder and target abnormal returns. Other papers (e.g. Michel and Shaked, 1988) use the book value of the total assets of the target company to determine whether the acquisition has enough impact on returns. I believe the deal value is a better alternative, since the deal value is in terms of the market value of the target company. Furthermore, expectations about future profitability of the acquisition are included in the deal value, which is also important for the impact the acquisition will have on abnormal returns.

After these criteria, a sample of 2750 bids to acquire a target company remained. As already noted, also rejected bids are included in this sample. This is necessary to identify multiple-bid takeovers and multiple-bidder takeover contests. Finally, two additional selection criteria were made on these two samples:

• The total number of days between the first bid and the final bid does not exceed 250 trading days (circa one calendar year). The reason for this selection criterion is explained in section 4.1.1.

• For multiple-bidder takeover contests, the total number of companies conducting an offer is two. This number corresponds to the preemptive bidding theory (Fishman, 1988).

The final sample consists of 53 multiple-bid takeovers and 54 multiple-bidder takeover contests (consisting of 107 bidding firms and 54 target firms).

(19)

multiple-bid takeovers and multiple-bidder takeover contests. The advantage of equal deal values across the samples is that the impact of the acquisition on bidder and target returns is equal across the samples. The sample of single-bid takeovers consists of 104 mergers and acquisitions.

3.2 Collection of data

After the selection of the mergers and acquisitions, share price information is obtained from DataStream. If the share price is denoted in a currency other than the U.S. dollar, the price is converted to the U.S. dollar using the corresponding exchange rate provided by DataStream. Returns are calculated using the total return index. In addition to the returns on the shares, returns on the world market index (MSCI World), the domestic market index (MSCI country indices), and the exchange rate (relative to the U.S. dollar) are obtained from DataStream to determine the abnormal returns (see section 4 for details).

For countries in the Euro zone, the MSCI EMU1 is used as the relevant domestic market model. This is done, because the common currency is assumed to achieve market integration in the Euro zone. Evidence for this assumption is provided by Hardouvelis et al. (2006). They show that the countries adopting the Euro converged to full market integration in the late 1990s. A country not adopting the Euro, the UK, showed no increased integration in the European stock market. The Euro was introduced at 1st January 1999, while the sample period in this study starts at 1st January 1997. To calculate the U.S. dollar returns in the period 1997-1999, the local currency exchange rate is used.

When there is no MSCI country index available for a specific country, a different country index is used2. If data is not available in the period under study (e.g. if the firm is not quoted yet or if there is no trade in their shares), the firm is deleted from the sample. In

1_{Only countries that adopted the Euro are included in the MSCI EMU: Austria, Belgium, Finland, France,}

Germany, Greece, Ireland, Italy, Netherlands, Portugal, and Spain.

(20)

20 the sample of single-bid takeovers, 20 target firms and 30 bidding firms were removed for this reason. In the sample of multiple-bid takeovers, 8 target firms and 12 bidding firms. 12 target firms and 30 bidding firms were removed in the sample of multiple-bidder takeover contests.3

An important aspect in analyzing mergers and acquisitions worldwide is lack of synchronism in trading hours. The stock markets in the United States and Asia do not have overlapping trading hours. Therefore, investors can not simultaneously respond to the announcement of the acquisition. To solve this problem, the stock returns and local market returns for companies located in Asia and Australia are lagged by one trading day (Park, 2004).

3.3 Descriptive statistics

After the selection described above, the total sample consist of 363 companies. Details are in table 3.1.

Table 3.1. The number of target firms and bidding firms in the data sample.

!umber of firms in the sample Bidding firms

Single-bid takeover 74

Multiple-bid takeover 41

Multiple-bidder takeover contest 77

Total number of bidders 192 Target firms

Single-bid takeover 84

Multiple-bid takeover 46

Multiple-bidder takeover contest 41

Total number of targets 171

Total sample 363

The sample of bidding firms in multiple-bidder takeover contests could be divided in first bidders and second bidders. In order to test the predictions of the preemptive bidding

3_{In order to analyze the emergence of multiple-bid takeovers and multiple-bidder takeover contests,}

(21)

theory, it is also important to distinguish between successful bidders (bidders that acquire the target after the contest) and unsuccessful bidders. Statistics on the number of companies in these sub-samples is in table 3.2.

Table 3.2. Details with respect to bidding firms in multiple-bidder takeover contests.

!umber of firms in the sample

First bidders 38 Successful 12 Unsuccessful 26 Second bidders 39 Successful 26 Unsuccessful 13 Total 77

The number of days between the first bid and the final bid differs when there are more bids needed to acquire the target company. Descriptive statistics with respect to the number of days between the first and the final bid in bid takeovers and multiple-bidder takeover contests are in table 3.3.

Table 3.3. Descriptive statistics with respect to the number of days between the first bid and the final bid (leading to the acquisition).

Multiple-bid takeovers Multiple-bidder takeover contests

Mean 58 52

Standard Deviation 42 32

Minimum 5 3

Maximum 172 155

(22)

22

Table 3.4. Overview of the countries in which the target company is located.

Country Single-bid takeovers Multiple-bid takeovers Multiple-bidder takeover contests Argentina 1 (1,19%) Australia 2 (2,38%) 7 (15,22%) 5 (12,20%) Austria 2 (2,38%) Belgium 1 (2,17%) 1 (2,44%) Canada 5 (5,95%) 4 (8,70%) 6 (14,63%) Croatia 1 (2,44%) Denmark 1 (2,17%) Finland 1 (1,19%) France 2 (4,35%) 2 (4,88%) Germany 3 (3,57%) 2 (4,35%) Greece 1 (2,44%) Israel 2 (2,38%) Italy 3 (3,57%) 2 (4,35%) 1 (2,44%) Japan 4 (4,76%) 1 (2,44%) Luxembourg 1 (2,17%) Netherlands 2 (2,38%) 1 (2,17%) Norway 1 (1,19%) 1 (2,17%) Spain 2 (2,38%) 1 (2,44%) Sweden 2 (2,38%) 2 (4,35%) Switzerland 1 (1,19%) 2 (4,88%) South Africa 1 (2,17%) United Kingdom 13 (15,48%) 7 (15,22%) 7 (17,07%) United States 40 (47,62%) 14 (30,43%) 13 (31,71%) Total 84 46 41

Table 3.5. The number of deals in which the method of payment offered initially was cash, shares or a combination of cash and shares (mixed).

Method of payment Single-bid takeovers Multiple-bid takeovers Multiple-bidder takeover contests Cash 32 (38,10%) 25 (54,35%) 23 (56,10%) Shares 31 (36,90%) 10 (21,74%) 9 (21,95%) Mixed 21 (25,00%) 11 (23,91%) 9 (21,95%) Total 84 46 41

(23)

Table 3.6. Descriptive statistics with respect to the deal values (in millions USD). Single-bid takeovers Multiple-bid takeovers Multiple-bidder takeover contests Mean 7 194 7 325 6 697 Median 2 300 1 957 2 476 Standard Deviation 11 988 12 528 8 568 Minimum 268 273 268 Maximum 69 635 55 482 38 908 Jarque-Bera 583,7 163,3 44,8 N 84 46 41 4. Methodology

4.1 The emergence of multiple-bidder takeover contests

This section presents the methodology that will be used to test the hypotheses of the preemptive bidding theory with respect to the variables that influences the emergence of single-bid takeovers, multiple-bid takeovers, and multiple-bidder takeover contests.

4.1.1 The initial premium

The premium initially offered (PR) is defined as follows:

j t j t j P P O PR − − − = (1)

(24)

24 anticipate the official offer (information about the offer is available before the official offer is made). The 10 days period to take into account information leakages also corresponds to the event window used in this study (which starts 10 days prior to the first offer, see section 4.2.1). To test whether the 10-days premium significantly differs from the 20-days, 30-days, and 40-days premium, a t-test assuming 2 dependent samples is conducted (equation 2). If there is a significant difference, this is evidence that there is information leakage more than 10 days before the first bid.

(

)

      + − + − + − − = 2 1 2 1 2 2 2 2 1 1 2 1 1 1 2 ) 1 ( 1 0 0 0 0 s 0 s 0 PR PR T j j (2) Where, 2 i

s is the unbiased estimator of the variance of the premium in the sample and 0 _i is the number of firms in the sample.

The difference between the initial premium in single-bid takeovers, multiple-bid takeovers, and multiple-bidder takeover contests is tested using the t-test assuming two independent samples: 0 s PR T d d = (3)

Where PR is the average difference between the two different j-days premiums and _d d

s the average standard deviation. 4.1.2 The costs of acquiring information

(25)

since these disclosures are public information (Bradley et al., 1988). Therefore, differences in legislation all over the world can cause differences in information acquisition costs. Among others, Pagano and Volpin (2001) show that takeover legislation is most developed and transparent in the United States, Canada and the United Kingdom. If a bidder wants to acquire a target company in these countries, the disclosure requirements are considered to be high, which will lead to lower information costs for a potential second bidder. If a target company is located in the rest of the world, the costs are assumed to be high.

A chi-square test is conducted to determine whether there are significantly more targets located in a country with low information costs when there is a multiple-bid takeover or a multiple-bidder takeover contest, compared to single-bid takeovers.

(

)

∑∑

− = i j ij ij ij E E O 2 2 χ (4)

WhereO is the observed number of cases and _ij E the number of expected cases if there _ij is no relationship between the target firm’s location and the emergence of multiple-bid takeovers and multiple-bidder takeover contests.

4.1.3 The method of payment

A chi-square test is also conducted to determine whether an initial offer in securities will more often lead to multiple-bidder takeover contests.

4.1.4 Cross sectional analysis

(26)

26 i

i i

i

i premium cash shares location

COMP =λ₀ +λ₁ +λ₂ +λ₃ +λ₄ +ε (5)

When the emergence of multiple-bidder takeover contests is analyzed, COMP equals 1 if there is a multiple-bidder takeover contest and zero otherwise if there is a single-bid takeover. The premium variable is the 10-days premium that is offered initially by the bidder. The variable cash equals 1 if initially cash was offered and zero otherwise. The variable shares equals 1 if initially shares were offered and zero otherwise. The variable location equals 1 if the target company is located in the US, UK or Canada (the low information cost countries) and zero otherwise.

The joint effect of the variables on the emergence of multiple-bid takeovers, is also tested using a logit transformation. Than, COMP equals 1 if there is a multiple-bid takeover contest and zero if there is a single-bid takeover.

4.2 The event study methodology

The methodology used in this paper to analyze the shareholder wealth effects is the event study methodology. Using financial market data, the effect of a specific event on the value of the firm is determined (MacKinlay, 1997). This effect is determined using a specific period of interest: the event window. To asses the effect of an event in the event window, it is necessary to determine what the returns of the firm under study are without the event taking place (the normal returns). To estimate these normal returns, a period preceding the event window is used, this period is called the estimation period. Finally, to compute the effect of the event, in each day of the event window, the normal return is subtracted from the actual return.

(27)

4.2.1 The event and the event window

The event in this study is the official announcement of a bid by a company to acquire an other company. In multiple-bid takeovers and multiple-bidder takeover contests, more bids are necessary to acquire the target; there will be more ‘events’. To determine the effects of the events on the shareholder returns, the whole period between the first bid and the final bid is used. The event window in this study is from 10 days before the first bid to 10 days after the final bid. This period is chosen to catch rumor before and after the bids. Since there is only one bid in single-bid takeovers, the event window is from 10 days before the bid to 10 days after the bid in single-bid takeovers.

The number of days used to catch information leakages is always arbitrary to some extend. Since multiple-bid takeovers and multiple-bidder takeover contests have different event windows, it is very difficult to accurately analyze when the announcement rumors start and disappear. I believe 10 trading days is a reasonable number of days to catch information leakages.

(28)

28 4.2.2 The normal returns

To estimate the normal returns in the event window, an estimation period of 250 trading days is used. This is roughly one calendar year and is often used as estimation period (e.g. MacKinlay, 1997). It is normal practice to have the estimation period before the event window. The closer the estimation period is to the event window, the higher the probability that the normal returns estimates are influenced by the event. Therefore, the estimation period is from 270 trading days before the announcement of the first bid of the first bidder to 20 trading days before the announcement of the first bid of the first bidder.

Normal returns in event studies are often estimated using the market model. The market model is a statistical model which relates the return of any given security to the return of the market portfolio (MacKinlay, 1997):

it mt i i it R R =α +β +ε (6) mt

R is the return on the market index. If the study focuses on a specific country, it is possible to use a country based index, like the S&P500 or the CRSP equally weighted index (e.g. Brown and Warner, 1985). Studies focusing on events in different countries often use a single world index, such as the MSCI world or the FTSE World index (e.g. Beckers et al., 1996). The study in this paper also has events in different countries. Several studies use a world market index as well as a domestic market index to estimate normal returns. For example, Agmon and Lessard (1977) showed that their multi-factor model (equation 7) is stronger in explaining returns in an international context than models using one market index.

it wt i dt i i it R R R =α +β +γ +ε (7) dt

(29)

To estimate normal returns, Park (2004) used the change in the exchange rate in addition to the return on the world market index and the return on the domestic market index. The model used in this study to estimate normal returns is based on Park (2004) and is as follows: it t i wt i dt i i it R R X R =α +β ~ +γ +δ ~ +ε (8) dt

R andR are defined as before andwt X is the change in the exchange rate of the t currency (of the country in which the company is located) relative to the U.S. dollar. The tildes indicate that the variable is determined using a statistical procedure (which is explained below and in appendix A).

In this paper, it is assumed that financial markets are integrated, but not perfectly. So, in addition to the world market index, pure domestic variables are expected to influence returns. Since the domestic market index also reflects the impact of world market variables on the domestic market, a statistical procedure is used to determine the pure domestic market index (i.e. the impact of world market variables on the domestic market index is removed). The statistical procedure is presented in appendix A.

(30)

30 Some papers, e.g. MacKinlay (1997), concluded that the gains from using multi-factor models (like the model in equation 8, which is used in this paper) to estimate normal returns are limited. To test whether this is true empirically, using the sample in this study, I determined the coefficients (using an OLS) and the adjusted R of the models in 2 equation 6 (using the MSCI country index and the MSCI world index), 7 (using Rdt

~ and wt

R ) and 8, for each firm in the sample. The mean values of the coefficient and the adjusted 2

R are in table 4.1. The market model used in this study is, on average, the strongest model in explaining the returns in the estimation period. The mean value of the adjusted 2

R is significantly higher than the mean value in a model using both the world market index and the domestic market index (The t-statistic of the difference is 4,46; this is significant at the 1% significance level).

Table 4.1. Average estimated coefficients and adjusted R2 for different models to explain normal returns.

Variables α β γ δ Adjusted R2

World index 0,001 0,7008 0,0837

Domestic index 0,001 0,6546 0,1343

World index + domestic index 0,0009 0,623 0,6748 0,1371 World index + domestic index +

exchange rate

0,0009 0,6239 0,6825 -0,5323 0,1883

4.2.3. The standardized cumulative abnormal returns

When the normal returns are determined, using the model in equation 9, it is possible to calculate the abnormal return of each firm on each day in the event window:

) ( _i _i _wt _i _dt _i _t it it R a bR g R d X AR = − + + + (9) it

R is the actual return of firm’s i share on day t. a,b,g, and d are the estimated α,β,γand

δfrom equation 8.

(31)

∑

= = = i i d t d t it i AR CAR 2 1 (10) i

d₁ is 10 days prior to the official announcement of the first bid; this is the first day in the event window. d2iis 10 days after to official announcement of the final bid; this is the last day in the event window. In single-bid takeovers, the first bid and the final bid are on the same day, since there is only one bid.

Since the event window differs across securities, it is difficult to economically interpret the CAR in equation 10 and to compare the average CAR between samples. Therefore, the CARs in the sample of multiple-bid takeovers and multiple-bidder takeover contests are harmonized to 21-days CAR (HCAR). 21 days is the event window in all single-bid takeovers: 21 1 1 2 × + − = i i i i d d CAR HCAR (11)

(32)

32 The cumulative abnormal returns are standardized as follows4:

1 1 2 − +       = i i i i i d d S CAR SCAR (12) i

S is the estimated standard deviation of the residuals from the estimation period of the market model. Among others, Patell (1976) adjustsS for variance in the market return _i outside the estimation period. However, MacKinlay (1997) explains that S is an _i appropriate estimation of the standard deviation of the abnormal returns if a long estimation period is used. Following MacKinlay (1997), an estimation period of 250 days is long enough to use the unadjusted S . _i

Equation 13 is used to determine the average SCAR in the samples under study. These are samples of first bidding firms and target firms in single-bid takeovers, multiple-bid takeovers, and multiple-bidder takeover contests.

∑

= = 0 i i SCAR 0 ASCAR 1 1 (13) Where, 0 is the number of companies in the sample.

Assuming that the ASCARs are independent across securities, the test statistic Z will be distributed unit normal for the assumed unit normal ASCAR (Dodd and Warner, 1983). The Z-statistic to test whether ASCAR is statistically different from zero is:

0 ASCAR

Z = (14)

4_{The harmonized cumulative abnormal returns are standardized in the same way. The denominator in}

(33)

To test the difference in ASCAR between different samples, assuming independent samples, the test statistic presented in Tehranian et al. (1987) is used:

2 1 2 1 1 1 0 0 ASCAR ASCAR Z + − = (15)

To test the joint effect of single-bid takeovers, multiple-bid takeovers, multiple-bidder takeover contests, the initial premium, the method of payment, and the costs of acquiring information on the HCAR, an OLS is conducted (equation 16). Abnormal returns of the target companies, the first bidding companies, and the first bidding companies which are successful in acquiring the target are analyzed.

i location shares cash premium contest multiple ingle HCAR=λ1s +λ2 +λ3 +λ4 +λ5 +λ6 +λ7 +ε (16)

Single equals 1 if the company is involved in a single-bid takeover and zero otherwise. Multiple equals 1 in bid takeovers, zero otherwise. Contest equals 1 in multiple-bidder takeover contests, zero otherwise. Premium equals the initial 10-days premium. Cash equals 1 when the initial offer is in cash, zero otherwise. Shares is 1 in initial offers in shares, zero otherwise. Location equals 1 when the target company is located in a country with low information costs, zero otherwise.

5. Results

5.1 The emergence of multiple-bid takeovers and multiple-bidder takeover contests

(34)

34 of multiple-bid takeovers and multiple-bidder takeover contests, are presented in the following subsections.

5.1.1 The initial premium

Table 5.1 shows the mean premiums offered by the bidding firm to acquire the target. Details are in appendix B.

Table 5.1. Premiums initially offered in mergers and acquisitions. The t-value of the difference is between brackets. Single-bid takeovers (1) Multiple-bid takeovers (2) Multiple-bidder takeover contests (3) Difference (1) and (2) Difference (1) and (3) 10-days premium 0,243 0,238 0,207 0,005 (0,132) 0,036 (0,727) 20-days premium 0,259 0,283 0,225 -0,023 (-0,553) 0,034 (0,699) 30-days premium 0,287 0,341 0,253 -0,054 (-0,928) 0,034 (0,633) 40-days premium 0,302 0,332 0,274 -0,03 (-0,564) 0,028 (0,513) Difference

10-days premium and highest premium -0,059 (-10,325)*** 0,103 (-2,625)** -0,067 (-2,096)** *,**, and *** indicate significance at the 10%, 5%, and 1% significance level

The premium initially offered in single-bid takeovers does not statistically differ from premiums offered in multiple-bid takeovers or multiple-bidder takeover contests. This result is robust when different values for j are used to calculate the premium. This result is in contradiction to the preemptive bidding theory (Fishman, 1988) and the results found in Jennings and Mazzeo (1993), but in line with the results (implicitly) found in Bradley et al. (1988) and Michel and Shaked (1988).

(35)

explaining harmonized cumulative abnormal returns. Using an OLS regression with the HCAR as the dependent variable and a constant and the premium as the independent variables, the adjusted R2 is 0,235 when the 10-days premium is used and 0,057 when the 40-days premium is used. Both premiums are used in the cross-sectional analysis in section 5.1.4.

The test results in table 5.2 show that in multiple-bid takeovers and multiple-bidder takeover contests, significantly more initial offers are in cash than in shares, compared to single-bid takeovers. This means that cash offers are more often followed by new bids (either by the same bidder or by an other bidder) than initial offers consisting of shares. The result does not support the preemptive bidding theory. Jennings and Mazzeo (1993) also found a positive relationship between cash payments and the degree of competition. A reason for this relationship is that hostile bids are more often financed with cash and more often resisted by the target firm management (Huang and Walkling, 1987). Resisted offers are more often followed by multiple-bid takeovers and multiple-bidder takeover contests (Lefanowicz and Robinson, 2000). Unfortunately, the ZEPHYR database in this study does not provide reliable information about management resistance. Therefore, it is not tested whether cash offers are more often resisted and followed by multiple-bid takeovers and multiple-bidder takeover contests.

Table 5.2. Chi-square test results with respect to the method of payment initially offered.

Method of payment Single-bid takeover (1) Multiple-bid takeover (2) Multiple-bidder takeover contest (3) 2 χ difference (1) and (2) 2 χ difference (1) and (3) Cash 32 (41,0%) 23 (29,5%) 23 (29,5%) Shares 31 (62%) 10 (20%) 9 (18%) 3,695* 4,445**

(36)

36 5.1.3 The costs of acquiring information

The conclusion from table 5.3 is that in single-bid takeovers, relatively more target firms are located in countries with low costs of acquiring information, compared to multiple-bid takeovers (significant at 10%). Compared to multiple-multiple-bidder takeover contests, no significant difference was found. The hypothesis of the preemptive bidding theory (Fisman, 1988) is rejected here. This result indicates that there are differences in the analysis using multiple-bid takeovers in stead of multiple-bidder takeover contests. It is important to notice that the predictions of the preemptive bidding theory apply to multiple –bidder takeover contests. The result in table 5.3 is also in contradiction to Jennings and Mazzeo (1993). However, they used a different proxy for the costs of acquiring information (number of analysts following the target).

An explanation for the result found in this paper is the sample selection method used (section 3.1). First, multiple-bid takeovers and multiple-bidder takeover contests were selected. After that, the single-bid takeovers were selected using a matched deal value. In general, more merger and acquisition activity takes place in the countries with low information costs (US, UK, and to a lesser extent Canada). Therefore, the probability that a single-bid takeover is selected from these countries is higher.

Table 5.3. Chi-square test results with respect to the number of target firms located in a low or high information cost country.

Single-bid takeovers (1) Multiple-bid takeovers (2) Multiple-bidder takeover contests (3) 2 χ difference (1) and (2) 2 χ difference (1) and (3) Low information costs 57 (53,8%) 23 (21,7%) 26 (24,5%) High information costs 26 (41,9%) 21 (33,9%) 15 (24,2%) 3,318* 0,343

(37)

This section shows the results of the logit transformation to test the joint effect of the variables in this study on the emergence of multiple-bid takeovers and multiple-bidder takeover contests5. In appendix C, the correlation tables are presented. There is no evidence found for multicollinearity between the variables.

Table 5.4 shows that the variable location is negatively significant at the 10% significance level. This indicates that less multiple-bid takeovers are observed if the target is located in a low information cost country. This is in line with the results found in section 5.3. The impact of the variables cash and shares is insignificant. An initial offer in shares will not result in more multiple-bid takeovers, which is in contradiction to the preemptive bidding theory (Fishman, 1988). The impact of the variable premium is also insignificant on the emergence of multiple-bid takeovers.

Table 5.4. Logit transformation results for single-bid takeovers versus multiple-bid takeovers

Variable Coefficient estimation Constant -0,112

Premium -0,120

Cash -0,324

Shares -0,579 Location -0,774*

*,**, and *** indicate significance at the 10%, 5%, and 1% significance level

Table 5.5 shows the impact of the variables in this study on the emergence of multiple-bidder takeover contests. Here, no variables significantly influence the emergence of multiple-bidder takeover contests.

5_{The coefficients are re-estimated using the 40-days premium rather than the 10-days premium (for which}

(38)

38

Table 5.5. Logit transformation for single-bid takeovers versus multiple-bidder takeover contests

Variable Coefficient estimation Constant -0,504 Premium -0,626 Cash 0,396 Shares -0,527 Location -0,14 5.2 Abnormal performance

This section presents the results with respect to the abnormal performance in single-bid takeovers, multiple-bid takeovers and multiple-bidder takeovers contests. First, the effects on target firms are analyzed. Second, the effects on first bidding firms are determined. Furthermore, the impact of the initial method of payment and the costs of acquiring information are analyzed. Finally, the joint effects of all the variables on abnormal returns are analyzed using an OLS.

Table 5.6. shows the CAR and the HCAR of the target firm and the first bidding firm6. Moreover, the differences between single-bid takeovers, multiple-bid takeovers, and multiple-bidder takeover contests are shown.

(39)

Table 5.6. Mean CAR and HCAR. The Z-statistic is between brackets. Single-bid takeovers (1) Multiple-bid takeovers (2) Multiple-bidder takeover contests (3) Difference (1) and (2) Difference (1) and (3) Difference (2) and (3) Target firm CAR 0,108 0,247 0,217 -0,143 -0,111 -0,318 (11,27) *** (11,419) *** (7,962) *** (-2,475) ** (-0,15) 1,964** HCAR 0,108 0,087 0,078 0,021 0,03 0,009 (11,27) *** (6,85) *** (4,78) *** (1,196) (2,466) ** (1,176) First bidding firm CAR -0,045 -0,073 -0,068 0,029 0,024 0,049 (-3,82) *** (-1,492) (-2,869) *** (-1,087) (0,104) (1,032) HCAR -0,045 -0,022 -0,02 -0,023 -0,025 -0,002 (-3,82) *** (-1,56) (-1,61) (-0,958) (-0,915) (0,017) First bidder successful in contest CAR -0,02 -0,025 -0,053 (-0,53) (-0,937) (-0,244) HCAR -0,005 -0,04 -0,017 (-0,28) (-1,171) (-0,542) First bidder unsuccessful in contest CAR -0,091 0,046 0,018 (-3,108) *** (0,723) (1,502) HCAR -0,028 -0,017 0,006 (-1,76) * (-0,434) (0,349)

5.2.1 Abnormal performance of the target firm

(40)

40 are compared to multiple-bid takeovers, the significantly higher CAR in multiple-bid takeovers becomes insignificant when HCAR is used.

When values of CAR are not harmonized, the higher returns observed in multiple-bidder takeover contests and multiple-bid takeovers are caused by a positive relationship between the CAR and the length of the event window.

The comparison of multiple-bid takeovers and multiple-bidder takeover contests shows that there is no significant difference in HCAR. This is in line with the results in De Sankar et al. (1996); they used normal CAR values. Since there are differences observed when multiple-bid takeovers and multiple-bidder takeover contests are compared to single-bid takeovers, it is good to distinguish between multiple-bid takeovers and multiple-bidder takeover contests in order to tests the predictions of the preemptive bidding theory.

5.2.2. Abnormal performance of the first bidder

Table 5.6 shows that the CAR of first bidding firms in single-bid takeovers does not statistically differ from the CAR in multiple-bid takeovers and multiple-bidder takeover contests. This result remains unchanged when HCAR is used and is line with De Sankar et al. (1996).

(41)

If the sample of successful first bidders in multiple-bidder takeovers is compared to first bidders in single-bid takeovers, no significant differences were found. This is in contradiction to the preemptive bidding theory (Fishman, 1988). In order to acquire a company, it does not matter whether the first bidder makes a high, preemptive bid or a low initial bid (that will result in a multiple-bidder takeover contest).

If the initial offer is in cash, target firms earn significantly higher (harmonized) cumulative abnormal returns than with an offer in shares (Table 5.7). This result is in line with the preemptive bidding theory (Fishman, 1989). However, in the preemptive bidding theory, the reason for higher returns in cash offers is, that more single-bid takeovers are observed than with initial offers in shares. As shown in section 5.1.2, initial offers in cash will not lead to more single-bid takeovers than multiple-bidder takeover contests in this study. Therefore, the reason why target firms earn higher returns in cash offers is not in line with the preemptive bidding theory. An alternative explanation for the result in table 5.7. is the tax hypothesis (Wansley et al., 1983).

(42)

42

Table 5.7. Mean values of CAR and HCAR for different methods of payment. The Z-statistic is between brackets.

Cash (1) Shares (2) Mixed (3) (1) – (2)

Target firm CAR 0,239 0,034 0,208 0,205

(15,537) *** (2,103) ** (12,693) *** (8,282) ***

HCAR 0,131 0,014 0,116 0,117

(12,020) *** (1,410) (10,283) *** (6,719) ***

First bidding firm CAR -0,038 -0,102 -0,04 0,064 (-2,43) ** (-3,194) *** (-2,953) *** (0,882) HCAR -0,015 -0,06 -0,028 0,09 (-2,051) ** (-3,019) *** (-2,436) ** (0,993) Successful first

bidding firm CAR

-0,035 -0,096 -0,025 0,061

(-1,90) * (-2,673) *** (-2,349) ** (0,801)

HCAR -0,015 -0,063 -0,028 0,048

(-1,816) * (-2,834) *** (-2,345) ** (0,958)

5.2.4 The costs of acquiring information

Table 5.8 shows that for target companies, the (H)CAR in low information cost countries is significantly higher than in high information cost countries. For firms making the first bid, the (H)CAR is significantly lower when the costs of acquiring information are low, compared to cases in which these costs are high. These results are in line with the preemptive bidding theory and empirical results in De Sankar et al. (1996).

(43)

Table 5.8. Mean CAR and HCAR in different information cost countries. The Z-value is between brackets. Low information costs (1) High information costs (2) Difference (1) and (2)

Target firm CAR 0,178 0,161 0,017

(15,517) *** (8,87) *** (2,301) **

HCAR 0,109 0,071 0,038

(12,57) *** (6,271) *** (2,593) ***

First bidding firm CAR -0,077 -0,022 -0,055 (-5,183) *** (-1,141) (-2,128) ** HCAR -0,045 -0,01 -0,035 (-4,532) *** (-1,114) (-1,767) * Successful first bidding firm CAR

-0,068 -0,02 -0,048

(-4,157) *** (-0,968) (-1,664)

HCAR -0,046 -0,01 -0,036

(-4,089) *** (-1,085) (-1,518)

. *,**, and *** indicate significance at the 10%, 5%, and 1% significance level

The results of the OLS regressions with respect to the target firms are presented in table 5.9.

Table 5.9. OLS regression results (target firms)

Variable Coefficient estimation Single 0,071** Multiple 0,037 Contest 0,034 Premium 0,338*** Cash 0,011 Shares -0,106*** Location -0,027 Adjusted R 2 0,314

(44)

44 to re-estimate the coefficients. This did not change the results in table 5.9. (test result is not shown).

Table 5.9. shows that the coefficient of the variable single is significantly positive. This indicates that HCAR is significantly higher in single-bid takeovers than in multiple-bid takeovers and multiple-bidder takeover contests, which is in line with previous results (section 5.2.1). The significantly positive value of the variable premium shows that there is a significantly positive relationship between the initial premium and the HCAR of the target firm. Since shares has a significantly negative coefficient, an initial offer in shares only will lead lower returns to the target firm shareholders than an initial offer in cash or a combination of cash and shares. This was also the result found in section 5.2.3. The effect of the costs of acquiring information on the HCAR of the target firm disappears in a cross-sectional analysis, since the variable location does not statistically differs from zero.

The same OLS is carried out using the first bidder HCAR. The results are in table 5.10. Again, a correlation table is in appendix C (table C6) and no evidence for multicollinearity between the variables was found. The White test was conducted to test for heteroskedasticity, which was not found (test result is not shown).

Table 5.10. OLS regression results (first bidding firms)

Variable Coefficient estimation Single -0,029 Multiple -0,027 Contest -0,026 Premium -0,032 Cash 0,013 Shares -0,030 Location 0,027* Adjusted R 2 0,033

(45)

different from zero; this means the HCAR of the first bidder is higher if the costs of acquiring information are low. This was also the result in 5.2.4.. The insignificant values of the other coefficients are also in line with the results found in the previous sections.

From table 5.11 can be concluded that the regression results do not change when only successful first bidders are considered. For the bidding firms, adjusted 2

R are very low. This is evidence that other variables, not included in this study, explain the harmonized cumulative abnormal returns found in this study.

Table 5.11. OLS regression results (successful first bidding firms)

Variable Coefficient estimation

Single -0,03 Multiple -0,027 Contest -0,029 Premium -0,029 Cash 0,016 Shares -0,03 Location 0,022* Adjusted R 2 0,019

6. Conclusions

(46)

46

Table 6.1. Overview of the hypotheses tested in this paper.

Hypothesis Test result

Emergence of multiple-bidder takeover contests

The initial premium offered in multiple-bidder takeover contests is lower than in single-bid takeovers.

Rejected

Multiple-bidder takeover contests are more often observed when the initial offer is in shares than in cash.

Rejected

Multiple-bidder takeover contests are more often observed if the costs of acquiring information are low.

Rejected

Shareholder wealth effects

Target firm payoffs are higher in single-bidder takeover contests than in multiple-bidder takeover contests

Not rejected

First bidder payoffs are higher in single-bid than in multiple-bidder takeover contests

Rejected Target firm payoffs are lower if the initial

offer is in shares than in non-shares offers

Not rejected First bidder payoffs are lower if the initial

offer is in shares than in non-shares offers

Rejected Target firm payoffs are higher if the cost of

acquiring information are lower

Not rejected First bidder payoffs are lower if the costs

of acquiring information are lower.

Rejected

When single-bid takeovers are compared with multiple-bid takeovers, the hypothesis that target firm payoffs are higher in single-bid takeovers is rejected. The hypothesis that a first bidder will have lower payoffs if the costs of acquiring information are low, is not rejected.

(47)

Extensions to this paper are possible by using other time periods. This paper used a very recent time period (1997-2006). To compare the results in this paper with the other papers in this field (e.g. De Sankar et al., 1996; Bradley et al., 1988; Jennings and Mazzeo, 1993), data from earlier time periods could be used. Furthermore, this paper can be extended using a more extensive database, including mergers and acquisitions with a deal value below 250 million US dollars and unquoted companies.

The hypotheses of the preemptive bidding theory with respect to the emergence of multiple-bidder takeovers are clearly rejected in this paper. What factors do have impact on the emergence of multiple-bid takeovers should be further researched. The most widely known factor is resistance of the management firm (Lefanowicz and Robinson, 2000). With the gaining interest in behavioral finance, it is possible that new hypotheses with respect to multiple-bidder takeover contests will emerge in this field

(48)

48 Appendix A

Statistical procedure to remove the effects of the world market index on domestic market index returns

First, α'and β' are estimated using an ordinary least squared regression (OLS).

' ' ' β ε α + + = _w d R R (A1) Than, R~_dt is determined: wt dt dt R R R~ ₌ ₋α' ₋β' (A2)

Now, R~_dt is the pure domestic impact on shareholder returns.

Statistical procedure to remove the effects of the domestic and world market index on the change of the exchange rate relative to the U.S. dollar

First, αˆ , βˆ , and γˆ are estimated conducting an OLS.

i w d i R R X =αˆ+βˆ +γˆ +ε (A3) Than,X~_iis determined: w i d i i i i X R R X~ = −αˆ −βˆ −γˆ (A4)