The Effect of Relative Valuation between Countries on Acquirer’s abnormal returns in cross-border mergers and acquisitions

(1)

FACULTY OF ECONOMICS AND BUSINESS

UNIVERSITY OF AMSTERDAM

MSc Business Economics, Finance track

Master’s Thesis

Mathew Kleine Punte 5732670

The Effect of Relative Valuation between Countries on Acquirer’s Abnormal

Returns in Cross-Border Mergers and Acquisitions

July, 2014

(2)

2

Acknowledgements

I dedicate this page to everybody who has guided me through the writing of this thesis.

First and foremost, I would like to thank my supervisor, Dr. J.E. Ligterink of the University of Amsterdam who has been provided guidance in choosing the research question and, further along the process, in providing expert views on what areas to focus on during my research.

Secondly, I would like to thank my professors during my Bachelor’s as well as Master’s studies at the University of Amsterdam for being able to share their knowledge of Economics and Business. Obviously, the broad base of their knowledge has provided me with the ability be able to write my own academic research thesis.

Special thanks go out to Dr. M.J.G. Bun who teaches Econometrics at the University of Amsterdam and who has taught me everything I needed to know to be able to perform such an extensive empirical research.

Furthermore, I would like to thank our colleagues from the Erasmus University of Rotterdam. Their research team has provided me with a modeled excel file which I was able to use to derive the dependent variable, the cumulative abnormal returns, in my empirical research. The creator of this modeled excel file is Arco van Oord, who is a PhD candidate at the Erasmus University and is currently employed at the Dutch central bank, De Nederlandsche Bank.

Finally, I would like to thank my parents, Dierdre and Vincent, my sisters, Andrea and Bernadette and my partner, Rhea for being there for me in so many ways. Not only did they proofread my thesis, but they also were consistently there to help me at moments I encountered mental barriers.

(3)

3

Abstract

This thesis focuses on a sample of 6,967 cross-border mergers and acquisitions from around the globe to discover the effect of relative valuation between the country pairs of such deals on the acquiring companies short-term cumulative abnormal returns over a three-day event window. The research shows that there is indeed a statistically significant, negative effect. I interpret these findings as spill-over effects from other factors influencing investors’ reaction to the deal. The effect is relatively small and as such it does not take away the opportunities for acquiring companies cheaply in cross-border deals.

(4)

4

I. Introduction

During recent decades we have experienced a period of globalization. This is also explicit in the increasing number of cross-border mergers and acquisitions that have taken place. Companies have revealed their interest in expanding their borders beyond those of their home country and have increasingly done so by cross-border mergers and acquisitions (M&A). From 1998 to 2007 cross-border mergers have increased from 23% to 48% of total merger volume. The rationale behind it is similar to domestic mergers and acquisitions: two firms will merge when their combined value exceeds the value of the separate entities in view of the managers of the acquiring firm (Erel et al., 2012). However, with the country borders come various advantages and disadvantages, such as opening up to new markets on the one hand and cultural barriers on the other. Also the influence of relative valuation on the probability of a cross-border merger occurring has been an area of interest in recent studies. This study focuses on the effect of relative valuation of stocks between countries. However alternatively to other studies it focuses on the effect it has on the abnormal returns acquirers experience from these transactions. Moreover, does a difference in relative valuation of stock prices of the two countries of interest have an impact on the stock price performance of the acquiring company? The focus of relative valuation in this research will be on price-earnings (PE) ratios and market-to-book (MTB) ratios. The research question is as follows:

What is the effect of a difference in relative valuation between countries, more specifically differences in PE and MTB ratios of stocks in bidder and target countries on abnormal returns of the acquiring company’s stock in cross-border mergers and acquisitions?

Very little has been written on this topic although it could have huge implications for the choice of merger and acquisition decisions, and specifically the choice of the country of interest. If a lower relative valuation of stocks for companies in the target’s country relative to the bidder’s country indeed has a positive effect, would it be enough to make companies seek buy-and-build opportunities in countries where stocks are valued lower? Or will this effect, if any, be overwhelmed by other factors influencing the choice of the country of the target company, such as regulations, labor wages, taxes, interest in end-markets et cetera? This study bears in mind above questions, while focusing on its main research question,

(7)

7

whether relative valuation difference of stocks in different countries affect the acquirer’s abnormal returns in cross-border mergers and acquisitions.

The hypothesis for the research is:

The positive differences in average PE ratios and MTB ratios have a positive effect on abnormal returns of the acquiring company’s stock price.

This thesis evaluates the main drivers behind this process and why the market has not yet corrected for this outcome. The main contribution of the thesis lies in the fact that although some literature already focuses on the effect of relative valuation between companies in acquisitions and its effect on acquirer’s abnormal returns, the effect of relative valuation between countries in cross-border mergers and acquisitions has been left fairly untouched. Erel et al. (2012) show that for their sample the market-to-book ratio of the acquirer’s country outperforms that of the target countries’ by 9,93% at the time of the merger. However, they do not discuss the effect this has on the acquirer’s abnormal returns. Hence, although they show evidence for an increase in probability of a merger occurring between these countries, the real question remains, can these relatively cheap mergers and acquisitions add value to acquirers or are they unable to translate this into excess value for their shareholders?

A further differentiation with existing literature is the use of price-earnings ratio for the method of valuation next to the more commonly used market-to-book ratio (MTB). This gives the research a more practical usage, because the PE ratio is more widely used by practitioners on a day-to-day basis.

II. Literature Review

Although there exists a vast amount of literature on mergers and acquisitions, this literature focuses mostly on domestic mergers and acquisitions, especially on the domestic public market in the United States. Even though a great deal can be extracted from this data when focusing on cross-border mergers and acquisitions there are some important distinctions as

(8)

8

well, such as geographic and cultural differences and factors influencing the economy and relative valuation in particular, which is the focus of this thesis.

A. Mergers and Acquisitions in General

Over the past century, empirical research has revealed many implications of why mergers and acquisitions take place. Mergers seem to create shareholder value mostly accruing to the target company’s shareholders. With target stock prices increasing with an average 15% from 1980 to 2005, while bidder´s stock prices reacted with a mere 1% increase (Eckbo, 2008). However, the specific characteristics driving these mergers have been widespread. This section examines the most important reasons to perform a merger or acquisition in general. The section thereafter focuses on additional reasons to perform a merger or acquisition overseas.

A1. Economies of Scale and Scope

Large firms could benefit from economies of scale, or a reduction of production costs that smaller companies do not have. The analogy behind this is that fixed costs are shared over the whole company instead of expensed over smaller companies separately, such as head office management, support services and accounting and financial control. Larger companies could also benefit from economies of scope, which could accrue from combining different types of products and sharing costs of for example marketing and distribution (Berk and DeMarzo, 2007).

A2.Vertical Integration

Closely related is vertical integration, where two companies merge which are in the same industry, but in a different part of the supply chain. The main benefit for both companies here is coordination. Management should be more able to guide both companies towards a common goal. Both companies are less affected by competition on the part of the supply of delivered by the company lower down the supply chain (Berk and DeMarzo, 2007).

(9)

9

A3. Expertise

Often a merger takes place because a company wants to be able to obtain the expertise another company has. Firms could try to establish this expertise themselves by hiring personnel with the required experience or doing a lot of research and development by themselves. However, sometimes a more efficient and faster way is to acquire a company which has already established a well-functioning unit with the required expertise (Berk and DeMarzo, 2007). This could save time and give the acquiring company the cutting edge in expertise it requires before losing market share and falling behind.

A4. Efficiency Gains

Mergers could act as a disciplinary force on the market. When management of the acquiring firm believes the target firm is managed inefficiently. A merger could be a way to replace the management. Although in theory investors could replace the board of directors by using their voting rights, this is often not done in this way. It requires alignment of a majority of shareholders and hence an investor is more likely to sell his shares and buy a better managed company. Therefore, a takeover is a good way to return efficiency to a company (Brealey et al., 2008).

A5. Surplus Funds

When a firm in a mature market, has excess cash, but not a lot of investment opportunities, management could choose to buy another firm’s shares which has more investment and growth opportunities, instead of buying back its own shares or paying out dividend. A firm that has these free cash flows and does not pay them out to shareholders or use it in a positive NPV takeover could engage in wasteful activities and hence could become a target of a takeover itself (Brealey et al., 2008).

A6. Monopolistic Gains

When a company combines with another major rival in its industry it is often said that competition is reduced in the industry and hence profits could be increased. Stand-alone

(10)

10

entities are not allowed to make price arrangements, but the combined entity is allowed to set its own prices. Often anti-trust regulations in a country set a hold to this type of mergers however (Brealey et al., 2008). It could also require a company to perform extra reorganizations before or after the merger, in order for regulators to approve the deal. Such as sell-downs of certain divisions to reduce capacity and increase competition.

B. Determinants of Cross-Border Mergers and Acquisitions

Exploring the case for cross-country mergers in particular, a new set of drivers should be included. With country borders come advantages and disadvantages in the case of mergers and acquisitions. Ahern et al. (2012) argue cultural differences could cause frictions when attempting to combine two firms from different countries. To benefit from synergy gains mergers require coordination between employees from both firms after the merger. However, if employees do not share similar cultural values, issues such as mistrust and individualism could result in those companies not being able to exploit these expected synergies to their full potential. Corporate government issues could also be of influence. Rossi and Volpin (2004) show that acquirers are generally from countries which promote better corporate governance, by maintaining better legal or accounting standards. Targets are typically from countries with poorer investor protectionism. Hence, when a takeover takes place additional synergies can be attributed to better investor protectionism for the target company by implying the bidder’s standards.

Next to earlier mentioned benefits in the general case of mergers and acquisitions, another case can be made for cross-country mergers in particular. Chari et al. (2009) show that developed countries often benefit from weaker contracting environments in non-developed countries. This is subscribed to the fact that non-non-developed countries are often less strictly regulated.

Another factor of influence is bilateral trade between the countries of interest. Firms are more likely to merge with firms from countries with which their own country has more bilateral trade. Erel et al. (2012) subscribe this to the fact that these countries are more likely to have more synergies because of a common cultural background. Furthermore, Erel et al. show that plain geographic proximity measured by the great circle distance between the capital cities of the countries of interest has a statistically significant effect on the likelihood of a merger taking place of companies within these countries.

(11)

11

C. Relative Valuation in Mergers and Acquisitions

In relative valuation the valuation of an asset is based on how similar assets are priced in the market. What is important is that these assets are priced by the market in a similar way. Furthermore, scaling down these prices to a common variable makes the prices comparable. The last step in comparing these variables is to adjust for differences across assets. When comparing stocks these differences originate from differences in fundamentals, such as growth, risk and cash flow generating potential.

Valuation plays an important role in the likelihood of cross-border mergers. Consider two countries of which one’s stock is valued lower than the stock of the other country. The country with the higher valued the stock would have an incentive to acquire a company from the country with the relatively lower valued stock. The companies from the country with the lower valued stock would on average be relatively inexpensive compared to companies from the higher valued company’s country. The same logic holds for a country of which the currency has appreciated relative to another country (Erel et al., 2012).

An important distinction to be made with these valuation differences is whether these valuation differences are considered to be permanent or temporary. Baker, Foley and Wurgler (2009) argue that foreign direct investment (FDI) arbitrage can occur for multinationals when valuation deviates from its normal values due to a change in investor optimism or risk aversion in either country. They discuss a cheap asset view in which the assets in another country are cheaper than in the domestic country making it cheaper for the acquirer to purchase the target across the border, because of a lower purchase price. However, they find more evidence confirming a cheap finance view in which financing becomes cheaper in the country with a higher relative valuation relative to the other country, making the acquirers more eligible to perform the FDI than bidders from the home country of the target. Shleifer and Vishny (2003) discuss a model in which overvalued firms use their equity to purchase undervalued or less overvalued firms. In their model they take mispricing as a given. They also assume that managers have private information about the valuation of their own stock, and of the stock of other companies. Especially the latter assumption obviously is less reliable. Once the transaction is made the firms’ valuations revert back to their normal levels and the acquiring firm will have arbitraged its relative overvaluation by locking in the profits from the valuation difference. This obviously is related to market timing theory which entails that firms will time the market to issue or buy back equity. Hence when takeovers are financed with equity, according to market timing theory this could be seen as a signal of the

(12)

12

acquirer’s stock being overvalued. Since financing a merger by equity is often a merger and stock issue in one.

If these valuation differences are considered to be permanent, Froot and Stein (2001) discuss a model in which the cost of capital decreases for the acquiring company once its stock value increases relative to that of the target’s country. They subscribe this to a decrease in information asymmetry when raising capital.

Furthermore there could potentially be role for the market timing theory in M&A. This theory is built upon the assumption that firms will issue equity when they are highly valued and repurchase when their stock is valued low, thereby reaping in an arbitrage profit (Baker and Wurgler, 2005). In M&A this could be executed by stock offers when the stock of the acquirer is valued relatively high compared to the target firm. Hence, making use of the overvaluation of its stock before the market corrects for this overvaluation and the stock price reverts to its correct value. Contrarily, cash offers would be preferred when bidder’s stock price is undervalued.

D. Price-Earnings Ratio

To be able to value companies on a relative basis, I standardize the equity values. One way to do this is by calculating price-earnings (PE) ratios. Since I will be using the PE ratio as one of the measures of the level of valuation, it is important to fully grasp the essence of this ratio and its drivers. The PE ratio is defined as follows:

The derivation of the price per share and earnings per share is discussed in section 3.

The main forces driving the PE ratio are growth, risk, and cash-flow generating potential. To control for differences in growth it is common to control for glamour versus value firms. Glamour firms being high growth firms and value firms being more mature, often larger, low growth companies. Risk could potentially be controlled for by adding a company’s beta. However this is less practically done when comparing relative valuation between nations, since the beta is already a determining factor of the dependent variable, abnormal returns. Although cash flow generating potential is of influence on a company’s PE

(13)

13

ratio it is not possible to control for in a reliable way in this study, which could potentially influence the results.

Because earnings can also be negative there will be an upward bias in the PE ratio, since negative values are dropped. This creates a bound on the downside of zero, whereas the upside is potentially unbounded. The upward bias can be seen in the difference between the average and the median. The former being higher than the latter. This contributes to the fact that the median of PE ratios is a better representation of the level of valuation than its average (Damodaran, 2006, p. 241). Figure 1 helps explain this fact, showing a skewness towards the higher PE ratios.

Figure 1. PE ratios – US Stocks in January 2006, source: Damodaran (2006), p.260. This figure presents the distribution MTB ratios of listed US firms in 2006.

The most important reason for using the PE ratio as a measure of valuation is its common use in valuation. Also it gives a representation of the value of the company relative to its cash generating potential. Furthermore, PE ratios are dependent on earnings instead of book value of assets, which allow for a better comparability across industries, because in different industries different relative amounts of assets are needed.

0 100 200 300 400 500 600 700 800 Num b er of F irm s PE ratio Trailing P/E

(14)

14

E. Market-to-Book Ratio

The market-to-book ratio is defined as follows:

The market capitalization is defined as the market price per stock times the amount of stocks outstanding. The net asset value is defined as the total firm book value minus the book value of the liabilities.

Figure 2 shows the distribution of MTB ratios of the US stock market in 2006. The same analogy counts for the MTB ratio as for the PE ratio with respect to its skewness. Because it is capped on the left hand side at zero and it is potentially unlimited on the right hand-side, the average of the MTB ratios will be higher than its median (Damodaran, 2006). The spikes in the figure at higher MTB ratios are solely caused by an increase in the MTB around those spikes. In reality the figure is thus completely smoothly skewed to the right.

Figure 2. MTB ratios – US Stocks in January 2006, source: Damodaran (2006), p.264. This

figure presents the distribution MTB ratios of listed US firms in 2006.

Another important factor influencing market-to-book ratios is that firms can have negative book values of equity, which makes it impossible to calculate the MTB ratio. However in practice this happens less often than PE ratios not being able to be calculated due to negative earnings per share. Furthermore, what could be of influence in deriving MTB ratios is accounting procedures in different regions and differing between individual firms. For instance goodwill is often not reported as book value of a firm. However, after a firm is

0 100 200 300 400 500 600 700 800 Num b er of F irm s MTB Ratio MTB Ratios

(15)

15

taken over, goodwill is reported on the balance sheet. This will obviously be of influence on the MTB ratio (Damodaran, 2006). However, there will be no influence of earnings as is the case for PE ratios and therefore MTB ratio should have less influence by having to drop firms with negative earnings. This should result in a limitation of upward bias in the average and median values.

F. Glamour vs. Value Firms

Considering this growth driver, Rau and Vermaelen (1998) and Sudarsanam and Mahate (2003) show that firms with high market to book ratios, defined as glamour firms, are the main cause of underperformance in long-run abnormal returns of bidders. Especially when they pay with stock, investors perceive this as an indication of them being overvalued and hence the market corrects their prices downwards after the acquisition. An alternative are value firms, with the lowest market-to-book ratios. The intuition behind them is the reverse of the glamour firms. They are firms with low MTB ratios and they are thus more likely to finance their acquisitions with cash, because of the lower valuation. Value firms are often large, slower growing firms. Conn et al. (2003) however, find that the method of payment in cross-border mergers has less of an impact. One explanation they give for this is that an increase in due diligence and monitoring in cross-border mergers takes away the effect of private information of valuation. This supports the hypothesis of valuation differences being permanent and not just bidders trying to arbitrage their stocks being overvalued, but being able to take advantage of cheaper financing.

III. Methodology

To derive the correct methodology to come up with the effect of the PE and the MTB ratio on abnormal returns of stocks in cross-border mergers and acquisitions one has to consider a few important factors. We need to make a clear concept of how PE and MTB ratios are derived. The effect of the magnitude of PE and MTB ratios on abnormal returns of the bidder have to be considered, and with it the implicit effects of a company being either a glamour or a value firm. Furthermore, a clear definition of these abnormal returns is defined. This chapter focuses on these issues and comes up with a method to account for these factors.

(16)

16

Furthermore, a model is derived to account for other factors that may be of influence on the abnormal returns during a cross-border merger or acquisition.

A. Derivation of Variables

Before diving into the econometric model any further, this subchapter gives a clear definition of some of the most important variables of interest. To promote uniformity as much as possible these guidelines are strictly followed in the computation and derivation of subsequent variables.

A1. Price-Earnings Ratio

Given the formula for the PE ratio above, the PE ratio is defined as follows. For the price per share I use its price at the time of the last earnings announcement. To give a more accurate presentation of earnings I use a trailing PE ratio, which means I take the sum of the earnings in the past four quarters. This is also because financial years differ in different regions. Using a trailing PE ratio hence gives a consistent representation of recent earnings. Furthermore, earnings per share are derived by the following formula:

For the most common stock exchanges a set of the largest companies with respect to market capitalization is taken. Hence, for the United Kingdom the PE ratios are derived from a list of a hundred companies all listing on the FTSE100 index. All of these companies report quarterly PE ratios. From these figures median quarterly PE ratios are derived for the United Kingdom.

For less common stock exchanges, such as the Bahrain Stock Exchange (BSE), DataStream does not report a commonly referred to sub-index and hence PE ratios are derived from all listed companies on the BSE.

Obviously, this leads to a bias in having more mature companies in more developed companies with larger and more developed stock markets. It should result in having more value firms in these countries. However, a firm being connoted a value versus a glamour firm

(17)

17

is a relative distinction and as such the distinction can still be made between these more mature firms within one stock exchange.

The downside of using PE ratios is that they can be negative when a firm has negative earnings. When this is the case its values will be dropped, which has an effect on the outcome, but it is the only way to handle these values and still being able to work with PE ratios. Another way to account for this upward bias in PE ratios is by using median values next to average values. Median values will be lower for PE ratios and hence experience less of an upward bias because of these missing values.

Furthermore, PE ratios are derived quarterly for every company, resulting in the most up-to-date representation of valuation.

A2. Market-to-Book Ratio

Given the formula for the MTB ratio as stated earlier, the MTB ratio is derived as follows. The market price is the current market price on the announcement date. The book value follows the same analogy and is the book value of the firm’s assets on the announcement date of the merger (the reported quarter before the announcement). For the MTB ratio, as for the PE ratio, quarterly data is derived. With respect to the stock exchanges from which the data is derived, the MTB ratio follows the same analogy as the PE ratio, also taking median values from the quarterly data to derive at a country’s valuation. This accounts partly for large outliers as the MTB ratio cannot be negative.

A3. Glamour vs. Value

To account for growth rates affecting the PE and MTB ratios, I divide the acquiring firms in value firms and glamour firms. Per countries quintiles are taken of the level of PE ratios and MTB ratios, defining the lowest quintile as value firm, and the highest quintile as glamour firm. Quintiles 2 through 4 are defined as neutral. The glamour and value variable are added as dummy variables to isolate these effect. The neutral variable is left out to avoid the dummy variable trap.

A4. Cumulative Abnormal Returns

I follow MacKinlay’s (1997) market model, using a three-day event window to retrieve cumulative abnormal returns (CARs) for the acquiring company. The abnormal returns are

(18)

18

estimated using the market-adjusted model, with the main stock index of the country of incorporation as a benchmark. The normal market return is estimated based on a 122-day estimation period and is estimated based on the stock’s ‘normal’ return with respect to its index within these 122 days.

Riτ is defined as the return of security i at time τ and Rmτ is defined as the return of the

market at time τ. From these two values the abnormal return of security i at time τ is calculated as follows:

Where αi and βi are OLS values from the estimation period.

Given N events, the aggregated abnormal return for security i at time τ is:

From this the cumulative abnormal return of security i is derived as follows:

With τ1 being the first time variable of the event window and τ2 being the last.

B. Event Window

As mentioned, a three-day event window is examined. The three-day event includes the day before the announcement date, the date of the announcement, and the day after the announcement. The day after the announcement is included because the announcement could be post trading, or at the end of the trading day. This way the full price effect of the announcement is captured. This assumes that the market can react to the announcement within one day post the announcement.

Only working days are considered, because on working days exchanges are open and stocks will show a return. There is no correction for non-western holidays and when observations are incomplete the deals are dropped from the sample.

The reason for solely focusing on short term returns is to isolate as much as possible the effect of the merger and the influence the valuation has on this. Obviously there is less

(19)

19

time for the market to correct for mispricing, but this is not of interest. The area of focus is the market reaction to the transaction and the value at which the market prices it assuming the market can distinguish the difference in valuation but is not efficient enough to correct for it on its own. Hence, only after the merger of acquisition this will be corrected.

C. Regression

For the abnormal return I use the following regressions. Returns in both cases are defined as cumulative abnormal returns (CARs). The first regression checks the effect of PE ratios. The regression is defined as follows:

Returns = β0 + β1PEAT + β2 Log Firm Size (acquirer) + β3 Same Industry + β4 Glamour +

β5Value + ε

The dependent variable Returns is defined as CAR. PEAT is the difference between average

PE ratios in country A of the acquirer and country T of the target. Firm size is the book value of assets at the time of the announcement. Same Industry is a dummy variable indicating 1 if the companies are in the same industry and 0 otherwise. The Glamour and Value dummies are defined as mentioned above. In this regression the Neutral variable is dropped due to the dummy variable trap.

The second regression is similar. Only here PE ratio is replaced by MTB ratio. It shows whether there is a difference between the two ratios and what determines this effect. It also counts as a robustness check for the regression. The MTB regression is defined as follows:

Returns = β0 + β1MTBAT + β2 Log Firm Size (acquirer) + β3 Same Industry + β4 Glamour +

β5Value + ε

The dependent variable Returns is defined as CAR. MTBAT is the difference between average

MTB ratios in country 1 of the acquirer and country 2 of the target. Firm size is the book value of assets at the time of the announcement. Same Industry is a dummy variable indicating 1 if the companies are in the same industry and 0 otherwise. The Glamour and

(20)

20

Value dummies are defined as mentioned above. In this regression the Neutral variable is dropped due to the dummy variable trap.

D. Heteroskedasticity

The regression assumes the ordinary least squares assumption; however there is potential heteroskedasticity in the error terms, which means that the error terms could have different variabilities. This is very much possible because independent CARs of companies in different indices could differ. Also, investor appetite could be different within different countries. To account for this possibility of heteroskedastic errors I make use of heteroskedasticity robust standard errors. Further reference to this problem is given in chapter IV, Data and Descriptive Statistics.

E. Potential Bias

Looking at the regression from part C of this chapter, there is a potentially inconsistent estimation caused by omitted variable bias. This is mainly because some variables are very difficult to derive for all these companies and countries. As Rossi and Volpin (2004) state in their research there is a potential influence of difference of accounting standards between companies and countries. These, however are not clearly stated for all companies and are not the main focus of this research and hence are not included in this research to be able to keep the sample size large and to keep the amount of variables included in the regression limited, since there is always a pay-off when including variables between incorrect estimation of the coefficient and an increase in variance. Furthermore one could argue the real relation between PE and MTB ratios and the accounting standards being used, as they are mostly a regulatory issue. There could however be a difference in reporting that could influence the ratios, such as a difference in balance sheets due to different reporting of goodwill.

The same analogy holds for including a proxy for governance standards. There is no consistent estimator of these standards for the countries of interest. Hence this variable has been left out of the regression.

Bilateral trade, circle distance between countries and a dummy variable for countries having the same language have been left out. The reason for this is that although this is expected to have an influence on the abnormal returns, there is no reason to believe it is

(21)

21

correlated with the relative valuation of the countries. As such exclusion will not result in omitted variable bias.

Another factor that could be of influence is that there could be an effect of the difference between valuations of the companies of interest. However, since the targets are not all listed companies it is not possible to calculate corresponding PE and MTB ratios. Therefore this could result in omitted variable bias, since valuation differences of the individual companies are most likely to be correlated with the valuation differences of the countries in which they are incorporated. Both variables are hypothetically correlated with the dependent variable, cumulative abnormal returns. The reason to not only include listed companies as targets is to create a relevant data set, not only for mergers and acquisitions between companies that are listed on stock exchanges, but also ensures validity for deals between listed companies and private targets.

Potentially, omitting a variable indicating the percentage of equity with which a company financed a deal could result in omitted variable bias. The reason for not including this variable is that in DataStream for most of the deals this is not given. The reason that this could be of influence is as explained earlier. When a company is valued relatively higher than a target, the company could choose to use its own higher valued equity to finance the deal. This could have an effect on abnormal returns of the acquirer, because of the diluting effect this has on its equity before the deal. Omitting this variable should thus have a negative effect on the coefficient and could result in a failure to reject the null hypothesis, even though there is indeed a positive effect of relative valuation between countries and the dependent variable, abnormal returns.

A variable for geographical proximity, which could be used as a proxy for cultural similarities, as mentioned by Erel et al. (2012) is not accounted for in the regression. The reason for this is that although this could potentially be of influence on the abnormal return, i.e. being cultural similar there could be a lower chance of mistrust and individualism post the merger, there is no reason to believe that this factor is correlated with the independent variable of interest, the relative valuation difference between the two countries. Hence this variable is not accounted for in the two main regressions.

(22)

22

IV. Data and Descriptive Statistics

This chapter gives a description of the data sources used and an extensive description of the variables included in the regression and their corresponding statistics.

A. Data Source

The time period for the data is 2003 until 2013. Although this contains several market downturns this should not be a problem as it should be a good representation of the business cycle in general. The most important reason for including the credit crunch, although being a quiet period with respect to merger and acquisition transactions, is that the aim of the research is to get a very up-to-date representation of the effect of relative valuation on cross-border takeovers. The time span of 11 years should prove to be a large enough time span to be statistically solid.

Cross-border deals are derived from Thomson One and include all cross-border deals which are completed between 2003 and 2013 and are between countries which have a national stock exchange market. This is to ensure that stock prices and valuations for these companies can be derived from DataStream. The acquirer should be a listed company. Deals are corrected for industry as well, dropping all deals with either the bidder or target belonging to the financial services of real estate industry. This is mainly due to heavy regulations on these sectors with respect to earnings, balance sheets and growth. Also because earnings have an extreme reliance on balance sheets in these industries. To be able to create the most homogeneous dataset these industries are dropped. After action to account for extreme outliers, as discussed in the section to com, this leaves remainder of 6,967 cross-border deals.

B. Descriptive Statistics

B1. Cumulative Abnormal Returns

The CARs, derived as described in the methodology section. The variables, however, contain some extreme outliers which are accounted for by using the Winsor command in Stata at the 1% level. This cuts of extreme outliers at the top 0.5% and bottom 0.5% end of the range of the variable. A graphical representation of the above is given in figure 3 and 4.

(23)

23

Figure 3. Cumulative abnormal returns. This figure presents the distribution of cumulative

abnormal returns of the acquiring company within the three-day event window.

Figure 4. Cumulative abnormal returns – Winsorized at 1%. This figure presents the distribution

of cumulative abnormal returns of the acquiring company within the three-day event window. The observations are cut off at 0.5% on either side of the range to prevent outliers.

0 .2 .4 .6 .8 F ra ct io n -5 0 5 10 CAR 0 .0 1 .0 2 .0 3 .0 4 .0 5 F ra ct io n -.2 -.1 0 .1 .2

(24)

24

Table I

Cumulative abnormal returns – Winsorized at 1%

This table presents an overview of the statistics for the dependent variable, cumulative abnormal returns for the acquiring company for the three-day event window surrounding the deal.

Percentiles Value Smallest Other Metrics

1% -0.125 -0.125 5% -0.062 10% -0.036 Observations 6967 25% -0.013 Sum of weight 6967 50% 0.005 Mean 0.010 Standard Deviation 0.048 Value Largest 75% 0.028 0.195 90% 0.065 Variance 0.002 95% 0.097 Skewness 0.778 99% 0.195 Kurtosis 5.918

Table I shows the distribution of the winsorized CARs. The mean is 0.010, meaning that on average there is a positive abnormal return of one per cent in a three-day window surrounding a cross-border merger or acquisition. The variable is a bit skewed to the positive side, implying there are more or larger positive outliers than negative outliers.

B2. Difference in PE ratios between countries

The difference in PE ratios between countries also contains some large outliers. This is mainly due to the fact that earnings can be relatively high or relatively low between countries. To account for these outliers the figures are winsorized at the 1% level. Figure 5 and 6 give a graphical representation of the above. As discussed in the Methodology section because there is no negative PE ratio there is some bias in this variable. However, there still are plenty of observations left and since the differences between the acquiring and target countries these effects cancel out for the largest degree.

(25)

25

Figure 5. Difference in PE ratios between countries.This figure presents the distribution of the difference in price-earnings ratios between the company of the acquiring firm and the country of the target company.

Figure 6. Difference in PE ratios between countries – Winsorized at 1%. This figure presents the

distribution of the difference in price-earnings ratios between the company of the acquiring firm and the country of the target company. The observations are cut off at 0.5% on either side of the range to prevent outliers. 0 .2 .4 .6 .8 F ra ct io n 0 500 1000 1500 PE DIF countries 0 .0 05 .0 1 .0 15 .0 2 .0 25 F ra ct io n -10 -5 0 5 10

(26)

26

Table II

Difference in PE ratios between countries – Winsorized at 1%

This table presents an overview of the statistics for the independent variable, difference in price-earnings ratios between the acquiring company’s country of incorporation and the target company’s country of incorporation.

1% -9.6 -9.6 5% -6.3 10% -4.6 Observations 6967 25% -2.6 Sum of weight 6967 50% 0.3 Mean 0.243771 Standard Deviation 3.938741 Value Largest 75% 2.8 10.7 90% 5.3 Variance 15.51368 95% 6.8 Skewness 0.076739 99% 10.7 Kurtosis 2.921536

Table II summarizes the statistics of the variable. The variable shows no significant skewness. The variable ranges between -9.6 and 10.7 with a variance of 15.51, which shows there is a large variance of relative valuation between countries, which makes sense. Also, the mean being slightly higher than zero suggest that deals are more often from relatively higher valued countries than lower valued countries. This makes sense as well, intuitively.

B3. Difference in MTB ratios between countries

As shown in figure 7 and table III, the difference between MTB ratios between countries is less affected by large outliers. Hence, there is no need for winsorizing. Table III shows no significant skewness in the variable. The mean is centered around zero and the variable ranges between -4 and 4.4. This makes sense considering there is a good change that acquiring companies are more often from relatively higher valued countries than lower valued countries.

(27)

27

Figure 7. Difference in MTB ratios between countries. This figure presents the distribution of the

difference in market to book values ratios the company of the acquiring firm and the country of the target company.

Table III

Difference in MTB ratios between countries

This table presents an overview of the statistics for the independent variable, difference in market-to-book ratios between the acquiring company’s country of incorporation and the target company’s country of incorporation.

1% -2.045 -4 5% -1.350 10% -0.990 Observations 6967 25% -0.495 Sum of weight 6967 50% Mean 0.045 Standard Deviation 0.813 Value Largest 5% 0.605 4.395 90% 1.050 Variance 0.662 5% 1.270 Skewness -0.311 9% 1.790 Kurtosis 3.603 0 .0 1 .0 2 .0 3 .0 4 .0 5 F ra ct io n -4 -2 0 2 4 MTBV Dif countries

(28)

28

B4. Book Value of Total Assets of the Acquiring Company

The book value of total assets of the acquiring company gives a good representation of the size of the acquiring company. This should have an effect on the CAR, because the stock price of a larger company will be less influenced by a merger or an acquisition than a smaller company. Furthermore, there is a possible correlation between the PE ratio and the MTB ratio of the country is which the acquiring resides. The value is retrieved for every acquiring firm in US dollars.

The book value is capped off at zero, since asset value cannot be negative. Hence there should be a skewness in the variable towards the positive side, which is shown in figure 8. Figure 8 also displays the presence of large outliers. To correct for this we winsorize the variable at the 1% level, this is represented in figure 9.

Intuitively, it makes even more sense to take the logarithm of the winsorized value of these book values. This gives a relative distribution of book value of assets of the acquiring company and the results is shown in figure 10.

Figure 8. Book Value of Total Assets of the Acquiring Company. This figure presents the total

distribution of the book value of total assets of the acquiring company.

0 .2 .4 .6 .8 1 F ra ct io n

0 5.000e+10 1.000e+11 1.500e+11 2.000e+11

(29)

29

Figure 9. Book Value of Total Assets of the Acquiring Company – Winsorized at 1%. This

figure presents the distribution of the book value of total assets of the acquiring company. The observations are cut off at 0.5% on either side of the range to prevent outliers.

Figure 10. Log of Book Value of Total Assets of the Acquiring Company – Winsorized at 1%.

This figure presents the distribution of the log of book value of total assets of the acquiring company. The observations are cut off at 0.5% on either side of the range to prevent outliers.

0 .2 .4 .6 .8 1 F ra ct io n

0 2.000e+09 4.000e+09 6.000e+09 8.000e+09 bookacq, Winsorized fraction .01

0 .0 0 5 .0 1 .0 1 5 .0 2 .0 2 5 F ra ct io n 10 15 20 25 bookacqwin_log

(30)

30

Table IV

Log of Book value of Total Assets of the Acquiring Company – Winsorized at 1%

This table presents an overview of the statistics for the independent variable, log of book value of assets for the acquiring company, which is used as a proxy for size of the company.

1% 9.5711 9.5711 5% 10.9555 10% 11.6344 Observations 6967 25% 12.9461 Sum of weight 6967 50% Mean 14.8275 Standard Deviation 2.6632 Value Largest 75% 16.50132 22.7701 90% 18.26088 Variance 7.0927 95% 19.66025 Skewness 0.5686 99% 22.7701 Kurtosis 3.2667

Table IV shows a large positive mean, which makes sense intuitively because acquiring companies should all have assets. There is still some relative skewness, but far less than when taking the plain book value.

B5. Announcement Year

A variable which is not used directly in the regression, but which is used while preforming robustness checks is the announcement year. Announcement year is the year in which the announcement was made a merger of acquisition. Therefore this could be earlier than 2003, the first year at which a deal was effective in this data set. This explains the smallest value of 2001 in table V.

(31)

31

Table V

Announcement year

This table presents an overview of the variable announcement year, which is the year in which the deal announcement takes place.

1% 2003 2001 5% 2003 10% 2004 Observations 6967 25% 2005 Sum of weight 6967 50% Mean 2007.583 Standard Deviation 2.902012 Value Largest 75% 2010 2013 90% 2012 Variance 8.421674 95% 2012 Skewness 0.095919 99% 2013 Kurtosis 1.949488

At 2007.6 the mean is along the expectation, because of the earlier announcements than effective dates. One could also argue that this is due the decrease in activity surrounding the credit crunch, starting in 2007. Figure 11 and the skewness in table V show that this effect is less present.

V. Results

In order to see whether a difference in relative valuation between the country of the acquiring and target company has an effect on CARs of the acquiring company the research focuses on two measures of relative valuation, PE ratios and MTB ratios. This chapter is subdivided into two chapters. Firstly the PE ratio is used as a measure of relative valuation and in the subsequent subchapter the MTB ratio is as the main independent variable.

A. PE ratios

A1. The Equation

(32)

32

Returns = β0 + β1PEAT + β2 Log Firm Size (acquirer) + β3 Same Industry + β4 Glamour +

β5Value + ε

The equation is subdivided into separate equation to test whether results differ extensively when regressors are added. Table VI shows uniform results between separate regressions and 1 through 4 and combined regressions of interest 5 and 6. Regression 2 and 6 show no significant impact of independent variable Same Industry. Theory suggests that firms could merge because they seek to add value through vertical integration (Berk and DeMarzo, 2007) or to create efficiency gains (Brealey et al. 2008) as described in the Literature Review of this research. However, there is no reason to assume that simply acquiring a firm from the same industry has an effect on the cumulative abnormal returns of the stock price when the deal has taken place.

The other variables, difference in PE ratios, log of book value of the acquiring company, the distinction into glamour and value company determined by quintile of PE ratio of the acquiring company do show small but significant impacts on the dependent variable, CAR. The proxy for relative valuation between countries, PE ratio, and the proxy for fast growing companies, Glamour Company, are both statistically significant at 5%. The proxy for the size of the company is statistically significant at the 1% level and the proxy for slower growing firms, Value Company, is statistically significant at the 10% level.

This gives the final and most important regression, number 5, where the Same Industry variable is excluded due to insufficient proof of significance. Also, adding the value in the regression, as in regression 6, does not influence the coefficients or the t-statistics of the other variables of interest and as such there is no added value in including this statistic in the main equation.

A2. Interpretation and Relevance

Focusing on the main independent variable of interest, difference in PE ratios, regression 5 tells us that when the difference in median PE ratios between the acquirer’s country and target’s country increases by 1 this has a statistically significant effect of -0.03 per cent on the acquiring company’s cumulative abnormal returns. That is, when the relative valuation of the acquiring company is higher than the target country, this will decrease the percentage of cumulative abnormal returns from the merger with 0.03 percentage point.

(33)

33

Table VI

Analysis of Relative Valuation (PE ratios) Between Countries on Cumulative Abnormal Returns

This table presents estimates of regressions of cross-border mergers and acquisitions and their country pairs. The dependent variable is the cumulative abnormal returns of the acquiring company of cross-border deals between 2003 and 2013. Columns 1 through 4 examine the separate effect of independent variables. Column 5 and 6 examine the combined effect. Column 5 is used as main regression of interest. Refer to the Appendix for variable definitions. Standard errors are heteroskedasticity robust and associated t-statistics are in parentheses. The symbols ∗∗∗, ∗∗, and ∗ denote statistical significance at the 1%, 5%, and 10% level, respectively.

Dependent Variable: Cumulative Abnormal Returns, Winsorized (p 0.01)

1 2 3 4 5 6

PE Dif (Win p 0.01) -0.000302** -0.000319** -0.000319**

(-2.104) (-2.203) (-2.203)

Same Industry 0.001243 0.001180

(1.068) (1.018)

Log Book Acquirer (Win p 0.01) -0.001488*** -0.001442*** -0.001442***

(-6.611) (-6.410) (-6.410) Glamour 0.004063** 0.003700** 0.003700** (2.477) (2.242) (2.242) Value 0.003066* 0.002690* 0.002690* (1.926) (1.693) (1.693) Constant 0.010085*** 0.009283*** 0.032070*** 0.008602*** 0.030212*** 0.030212*** (17.355) (10.661) (9.004) (12.781) (8.454) (8.454) Observations 6.967 6.967 6.967 6.967 6.967 6.967 Adjusted R2 0.000 0.000 0.007 0.001 0.008 0.008

(34)

34

The difference can be positive as well as negative, which makes it more difficult for the coefficient to be statistically significant if there was no correlation and hence no sign of causal effect. It does however explain why the effect is so small and hence close to zero. The difference between the median PE ratios of the companies within the most important stock exchanges is unlikely to be so large that the expected effect on CARs would be very large. As seen in table II, the difference in PE ratios between the two countries in the data set, after correcting these values for the smallest and largest 0.5 per cent, for these countries varies between -9.6 and 10.7. This would suggest a range between -0.32 and 0.28. Since there are corrections for the size and growth of the company in the regression this should be the isolated effect of the difference in relative valuation.

Hence, although the regression suggests a statistically significant effect of relative valuation difference on the CARs it does not seem to be of extreme relevance for the acquiring company due to its lack of magnitude. On the other hand, when a company is very large we could still be talking about a substantial amount of value that is being destroyed by the deal when a deal should in fact be adding value to the firm.

The outcome for the proxy for size, the book value of the acquiring company makes sense intuitively. The larger a company is the smaller is the positive effect of the cross-border merger or acquisition on its abnormal returns from the deal. When a listed company is large, it will have a larger market capitalization. Assuming an equal size target, the impact should be less on the larger company. Hence, the sign of the coefficient makes sense.

The other control variables controlling for its valuation at the time and its growth rate, glamour and value companies, show a higher coefficient for glamour companies than for value companies. This could be intuitively explained as follows. When a company is a glamour company and thus has a higher relative valuation, the market values this company higher than other companies. The market shows appetite for this company and rewards the company for exploring growth opportunities. An interesting variable to add when this effect is of interest would be measure of payment. When this is considered one could isolate the real effect of market appetite and on the other hand, through the added variable, the effect of paying with the company’s equity. As the market could realize that the company values its own equity as overvalued relative to the target.

(35)

35

B. MTB ratios

B1. The Equation

For MTB ratios the formula from the methodology section is tested as described below. Returns = β0 + β1MTBAT + β2 Log Firm Size (acquirer) + β3 Same Industry + β4 Glamour +

β5Value + ε

The equation is subdivided into separate equation to test whether results differ extensively when regressors are added. Table VI shows uniform results between separate regressions and 1 through 4 and combined regressions of interest 5 and 6. Regression 2 and 6 show no significant impact of independent variable Same Industry. Theory suggests that firms could merge because they seek to add value through vertical integration (Berk and DeMarzo, 2007) or to create efficiency gains (Brealey et al. 2008) as described in the Literature Review of this research. However, there is no reason to assume that simply acquiring a firm from the same industry has an effect on the cumulative abnormal returns of the stock price when the deal has taken place.

The other variables, difference in MTB ratios, log of book value of the acquiring company, the distinction into glamour and value company determined by quintile of MTB ratio of the acquiring company do show small but significant impacts on the dependent variable, CAR. The proxy for relative valuation between countries, MTB ratio, and the proxy for fast growing companies, Glamour Company, are both statistically significant at 5%. The proxy for the size of the company is statistically significant at the 1% level and the proxy for slower growing firms, Value Company, is statistically significant at the 10% level.

This gives the final and most important regression, number 5, where the Same Industry variable is excluded due to insufficient proof of significance. Also, adding the value in the regression, as in regression 6, does not influence the coefficients or the t-statistics of the other variables of interest and as such there is no added value in including this statistic in the main equation.

(36)

36

Table VII

Analysis of Relative Valuation (MTB ratios) Between Countries on Cumulative Abnormal Returns

This table presents estimates of regressions of cross-border mergers and acquisitions and their country pairs. The dependent variable is the cumulative abnormal returns of the acquiring company of cross-border deals between 2003 and 2013. Columns 1 through 4 examine the separate effect of independent variables. Column 5 and 6 examine the combined effect. Column 5 is used as main regression of interest. Refer to the Appendix for variable definitions. Standard errors are heteroskedasticity robust and associated t-statistics are in parentheses. The symbols ∗∗∗, ∗∗, and ∗ denote statistical significance at the 1%, 5%, and 10% level, respectively.

1 2 3 4 5 6

MTB Dif -0.001542** -0.002492*** -0.002481***

(-2.147) (-3.374) (-3.358)

Same Industry 0.001243 0.001186

-1,068 -1,024

Log Book Acquirer (Win p 0.01) -0.001488*** -0.001592*** -0.001593***

(-6.611) (-6.878) (-6.883) Glamour 0.004063** 0.003382** 0.003347** (2.477) (2.071) (2.050) Value 0.003066* 0.002486 0.002499 (1.926) (1.563) (1.572) Constant 0.010081*** 0.009283*** 0.010139*** 0.008602*** 0.032573*** 0.031893*** (17.360) (10,661) (17.140) (12.781) (8.842) (8.518) Observations 6.967 6.967 6.967 6.967 6.967 6.967 Adjusted R2 0.001 0.000 0.000 0.001 0.009 0.009

(37)

37

B2. Interpretation and Relevance

Focusing on the main independent variable of interest, difference in MTB ratios, regression 5 tells us that when the difference in median MTB ratios between the acquirer’s country and target’s country at the time of the deal announcement increases by 1 this has a statistically significant effect of -0.25 per cent on the acquiring company’s cumulative abnormal returns. That is, when the relative valuation of the acquiring company is higher than the target country, this will decrease the percentage of cumulative abnormal returns from the merger with 0.25 percentage point. The difference can be positive as well as negative, which makes it more difficult for the coefficient to be statistically significant if there was no correlation and hence no sign of causal effect The difference between the median MTB ratios of the companies within the most important stock exchanges is unlikely to be so large that the expected effect on CARs would be very large. As seen in table III, the difference in MTB ratios between the two countries in the data set, after correcting these values for the smallest and largest 0.5 per cent, for these countries varies between -4 and 4.4. This would suggest a range between -1.1 and 1. Since there are corrections for the size and growth of the company in the regression this should be the isolated effect of the difference in relative valuation.

Hence, the regression suggests a statistically significant effect of relative valuation difference on the CARs. Compared with the other measure of relative valuation this could be quite substantial as well.

The outcome for the proxy for size, the book value of the acquiring company makes sense intuitively. The larger a company is the smaller is the positive effect on the cross-border merger or acquisition on its abnormal returns from the deal. When a listed company is large, it will have a larger market capitalization. Assuming an equal size target, the impact should be less on the larger company. Hence, the sign of the coefficient makes sense.

The other control variables controlling for its valuation at the time and its growth rate, the distinction between glamour and value companies, show a higher coefficient for glamour companies than for value companies. This could be intuitively explained as follows. When a company is a glamour company and thus has a higher relative valuation, the market values this company higher than other companies. The market shows appetite for this company and rewards the company for exploring growth opportunities. An interesting variable to add when this effect is of interest would be measure of payment. When this is considered one could isolate the real effect of market appetite and on the other hand, through the added variable, the effect of paying with the company’s equity. As the market could realize that the company

(38)

38

values its own equity as overvalued relative to the target. In regression 5, the coefficient for value company is no longer statistically significant.

VI. Robustness Checks

For the PE ratios regression as well as the MTB focused regression the error terms show some signs of heterskedasticity, as shown in figure 11 and 12, and hence during the regression heteroskedasticity robust standard errors are used. This allows for heteroskedasticity without a bias in the coefficients and t-statistics.

Figure 9. Standard errors of regression 5 Table VI (PE) without heteroskedasticity robust standard errors. This figure presents the distribution of standard errors. The figure shows a greater

dispersion around 0.1 of the fitted values. However, there are also more observations around that point. -. 2 -. 1 0 .1 .2 R e si d u a ls -.005 0 .005 .01 .015 .02 Fitted values

(39)

39

Figure 10. Standard errors of regression 5 Table VII (MTB) without heteroskedasticity robust standard errors. This figure presents the distribution of standard errors. The figure shows a greater

dispersion around 0.13 of the fitted values. However, there are also more observations around that point.

Intuitively, this heteroskedasticity could be a result of clustering between years and acquiring nation as well. Table XI however shows that when clustering error terms in this manner this does not affect the coefficients in any way. There is a difference to be seen in the t-statistics and hence the significance of the coefficients. However, a better way to account for the effect of years and nations, if any, would be to add them straight into the regression.

Since there are so many different countries in the data set this is unlikely to add real value to the regression. For years however, this is more likely to retrieve worthy results if there indeed is an effect. To check for this effect I have created dummy variables for the announcement years 2002 through 2013. These dummies are added to the regression separately and as an interaction terms with the main independent variable of both regressions, PE ratios and MTB ratios. The outcome is summarized in table X.

-. 2 -. 1 0 .1 .2 R e si d u a ls -.01 0 .01 .02 .03 Fitted values

(40)

40

Table IX

Clustering of Error Terms

This table presents estimates of regressions of cross-border mergers and acquisitions and their country pairs. The dependent variable is the cumulative abnormal returns of the acquiring company of cross-border deals between 2003 and 2013. In regressions 1 and 4, the error terms are clustered per country of incorporation of the acquiring company. In regressions 2 the 5, the error terms are clustered per year in which the announcement is made. Regressions 3 and 6 the error terms are hetereskedasticity robust error terms such as presented earlier. Refer to the Appendix for variable definitions. Standard errors are heteroskedasticity robust and associated t-statistics are in parentheses. The symbols ∗∗∗, ∗∗, and ∗ denote statistical significance at the 1%, 5%, and 10% level, respectively.

1 2 3 4 5 6

Standard Errors Clustered by

Acquiring Nation Clustered by Year Heteroskedasticity Robust Clustered by Acquiring Nation Clustered by Year Heteroskedasticity Robust PE Dif (Win p 0.01) -0.000319* -0.000319* -0.000319** (-1.874) (-2.101) (-2.203) MTB Dif -0.002492** -0.002492** -0.002492*** (-2.091) (-2.804) (-3.374) Log Book Acquirer (Win. p 0.01) -0.001442*** -0.001442*** -0.001442*** -0.001592*** -0.001592*** -0.001592***

(-4.652) (-6.470) (-6.410) (-5.303) (-6.193) (-6.878) Glamour 0.003700* 0.003700** 0.003700** 0.003382 0.003382** 0.003382** (1.728) (2.980) (2.242) (1,619) (2,926) (2,071) Value 0.002690** 0.002690* 0.002690* 0.002486** 0.002486 0.002486 (2.269) (1.857) (1.693) (2,113) (1,642) (1,563) Constant 0.030212*** 0.030212*** 0.030212*** 0.032573*** 0.032573*** 0.032573*** (5.759) (8.255) (8.454) (6,692) (7,585) (8,842) Observations 6.967 6.967 6.967 6.967 6.967 6.967 Adjusted R2 0.008 0.008 0.008 0.009 0.009 0.009

The Effect of Relative Valuation between Countries on Acquirer’s abnormal returns in cross-border mergers and acquisitions

FACULTY OF ECONOMICS AND BUSINESS

UNIVERSITY OF AMSTERDAM

Master’s Thesis

The Effect of Relative Valuation between Countries on Acquirer’s Abnormal

Returns in Cross-Border Mergers and Acquisitions

Acknowledgements

Abstract

TABLE OF CONTENTS

I. Introduction

II. Literature Review

III. Methodology

IV. Data and Descriptive Statistics

V. Results

Analysis of Relative Valuation (PE ratios) Between Countries on Cumulative Abnormal Returns

Analysis of Relative Valuation (MTB ratios) Between Countries on Cumulative Abnormal Returns

VI. Robustness Checks

Clustering of Error Terms