The case of acquirers from China, India, Brazil and South Africa

(1)

Shareholders’ Wealth:

The case of acquirers from China, India, Brazil and South

Africa

Author: M. de Groot

University of Groningen

Faculty of Economics & Business

Supervisors:

Prof. Dr. C.L.M. Hermes (1

st

supervisor)

Geir Gunnlaugsson (2

nd

supervisor)

(2)

The Case of Acquirers from China, Brazil, India and South Africa

2 Abstract

This paper analyses the announcement effect of acquisitions on the wealth of shareholders of the acquiring companies in China, India, Brazil and South Africa. This study is based on a sample of 426 acquisitions done between the beginning of 2000 until the end of 2007. For the entire sample there is a significant positive announcement effect, however the effect differs per country. It appears there is a strong positive relation between the return for shareholders and acquisitions with a domestic target company. The relation between cross-border acquisitions and the announcement effect is insignificant. Also when looking at cross-border acquisitions with a target in either developed or developing countries.

Key words:

(3)

3 Preface

I have written this thesis to complete both my MSc in International Business and Management for the faculty of Economics and Business, Groningen University (the Netherlands) and my MSc in International Business and Economics for the faculty of Economics, Uppsala University (Sweden). I have chosen for “the announcement effect on shareholders’ wealth in China, India, Brazil and South Africa” as my topic because of my personal interest in mergers & acquisitions and developing economies.

I would like to thank my first supervisor Professor Dr. Niels Hermes (Groningen University) for the useful feedback and guidance through the process of writing the thesis. Also I would like to thank my second supervisor Geir Gunnlaugsson (Uppsala University) for his helpful remarks and comments.

(4)

4

1 Introduction

Given the substantial value of mergers and acquisitions during the last decade, as can be seen in figure 1, it is not a surprise that numerous of articles have been written about merger and acquisition (M&A) subjects. Especially the last six years the deal value per year has increased tremendously. Last year alone the worldwide deal value reached US$4.78 trillion (Wall Street Journal). Europe had the biggest stake of 43% in the deal value and the U.S. came in second with a 32% stake. 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Trillion US$ 19951996 1997 1998 1999 2000 2001 2002 2003 2004 2005 20062007 Years

Yearly M&A value worldwide since 1995

Figure 1: M&A yearly value since 1995 until 2007 (Source; Dealogic)

One of the M&A related subjects is; whether mergers and acquisitions increase shareholders’ wealth. Numerous studies have been performed on this topic (Thompson and Mullineaux, 1995; Walker, 2000; Moeller, Schlingemann and Stulz, 2003; Scholtens and de Wit, 2004). The majority of the studies are focusing on the banking sector.

The fact that mergers take place has many possible reasons. Studies have pointed out that the reasons can lie in for example increased market power (a firm’s ability to alter the market price of a good or service), economies of scale as well as scope (cost advantages due to expansion in production volume and scope), diversification (a broader product range) and so on (Andrade, 2001; Pillof, 1996). Although there are several reasons pointed out, it seems that there is an

(6)

6

overall consent that mergers should improve the acquiring company’s financial performance and therefore increase the shareholders wealth (Lubatkin, 1983).

In this thesis “shareholders’ wealth” is defined as the (personal) wealth shareholders derive from investing in a company. The change in shareholders’ wealth as from the moment of the investment can be both positive and negative in monetary terms. Shareholders can benefit in three different ways. Firstly, they can benefit through dividends paid by the company, which are a distribution of part of the company’s net profit to the shareholders. Secondly, from the issuance of new shares when the company offers these newly issued shares to its shareholders at a discount or even for free as a bonus for being a shareholder. Thirdly a shareholder can benefit from capital growth, which is the increase of the market value of the company’s shares. It usually reflects the growth of the company’s assets and profits, or better said it reflects the investors’ interpretation of these assets and profits (Watson & Head, 2007). The influence on the share price as a result of investors’ opinion on this price (and therefore the trading in shares), is the subject of this research. More precise it handles about the degree to which acquisition announcements impact the shareholders’ wealth.

To increase shareholders’ wealth through a merger, the target company and the acquiring company have to be worth more together than accumulated separately (Pillof, 1996). However, research has not been able to find convincing empirical evidence that mergers do indeed increase value. Although more and more mergers and acquisitions are taking place, most of them do not succeed to bring the expected or desired returns for the acquiring firm (Hitt et al.) Another research pointed out that over 60 percent of the acquisitions failed to have a greater return than the annual costs of the loan(s) taken to finance the acquisition (McKinsey, 1990).

This is a contradictory finding to the assumption that M&As add value to a company. This contradiction is supported by numerous studies. On the one hand mergers will in the end not prosper. On the other hand there is a consistent result in numerous of event studies that the mere announcement of an acquisition brings wealth for the shareholders (Fridolfsson & Stennek, 2005). It would imply that the shareholders’ wealth created at announcement gets lost later during the execution phase of the merger.

(7)

7

were responsible for 13 percent of all the cross-border deals done in 2005 (which is approximately US$90 billion).

Until now little research has been performed on the M&A deals done by companies located in developing countries. The main reason being that as a result of their recent entering the M&A market and the relatively small size of the deals the availability of data is limited. However, the developing countries are moving up, India for example, has done over 90 overseas transactions worth over US$4.5 billion in 2005 alone (Financial Times, 2005). This is an increase of more than 40 percent compared to 2004. For other emerging countries the growth of total value of acquisitions per year is also far above the developed country average. In particular the emergence of China, India, Brazil and South Africa (CIBS) is reshaping the global economic and political landscape. The prospects and implications of the rise of CIBS are of profound importance to national governments, business communities, and international governance. Because of the potential role CIBS have in the world economy, this research will focus on these four countries, which inhabit 40% of the world population.

Given the rising presence of CIBS in the global M&A market as well as in the world economy and combined with the interest in the topic “the announcement effect of acquisitions on shareholders’ wealth”, I have formulated the following main research question:

What is the effect of acquisition announcements in China, India, Brazil and South Africa on the wealth of the acquiring company’s shareholders, and what is the impact on the announcement effect of whether an acquisition target is located domestically or cross-border?

(8)

8

optimize their investment decisions. Both the trend of a growing M&A business and the fact this study can add value to the existing literature are reasons to perform this research.

I have chosen the countries China, India, Brazil and South Africa specifically because of their important role in the prosperity of developing countries and their rapid growth. Their rapid economic expansion may have an influence on the announcement effect in CIBS. It could be that the announcement effect differs in CIBS from developed countries.

(9)

9

2 Literature review

This chapter will discuss existing literature on topics relating to the content of this thesis. The following section is built on existing articles. This section will discuss numerous studies, which form the existing theoretical framework in the announcement effect literature area. An overview of the results of the performed studies will be displayed as well. Section 2.2 will explain why it is likely that there is a difference in the announcement effect in developing countries compared with developed countries. The last and third paragraph will elucidate the focus of this research on domestic and cross-border acquisitions.

2.1 Empirical evidence

(10)

Author Sample Size Sample Period Country / Region Outcome

Chang 281 1981-1992 United States Significant positve

1998 abnormal return

Conn et al. 3,637 1984-1998 United Kingdom Small positive abnormal return 2005

Draper & Paudyal 7,499 1981-2001 United Kingdom Small positive abnormal

2006 return*

Eckbo & Thorburn 1,261 1964-1982 Canada Significant positve abnormal

2000 returns

394 United States Indistinguishable from zero abnormal return

Faccio, McConnel & 3,694 1996-2001 Europe Small positve abnormal

Stolin return

2006

Healy et al. 50 1979-1984 United States Significant negative abnormal

1992 (large mergers) return

Mitchell & Stafford 2,421 1958-1993 United States Small negative abnormal

2000 return

Scholtens & 61 1990-2000 United States Small negative abnormal return

De Wit 20 Europe Small positive abnormal return

2006 total of 81

Schwert 2,346 1979-1996 United States Small positive abnormal return 2000

Walker 278 1980-1996 United States Small negative abnormal return 2000

Table 1: An overview of existing literature and its outcomes

This study will answer five different hypotheses in total in the end of the research. Because of the existing literature on the announcement effect in developed countries I expect that there will be

(11)

11

an announcement effect in CIBS as well. To test whether in CIBS an announcement effect exists I propose the following hypotheses.

H: When everything else is equal, there is a positive abnormal return for the shareholders of the acquiring firm in CIBS around an acquisition announcement.

Throughout this whole thesis I will deliberately look at the acquiring companies. This is because of the fact that not enough deals have been done in developing countries with listed target companies. Therefore it is not possible to measure the announcement effect on the shareholers’ wealth of the target companies with the event study performed in this thesis.

Although numerous articles have been written on the same subject with a sample of acquiring companies from developed countries, it does not mean the same can be assumed for developing countries like CIBS. The reason why it can not be assumed will be explained in the paragraph 2.3. I will explain the general chracteristics of the acquisitions that influence the announcement effect in the following paragraph.

2.2 The influence of general acquisition characteristics on the announcement effect

In this section it will be explained what general characteristics have an influence on the announcement effect of acquisitions.

(12)

12

shareholders of the target company (Conn et al., 2005). Because of the signal of uncertainty it does not mean there are no longer expected returns.

By looking at the acquirer and target industry it is possible to indicate if an acquiring company is either following a focus or diversifying strategy (Moeller & Schlingemann, 2004). Acquisitions made to diversify have a lower value than acquisitions made focusing on the core-business. The relative size of the company is simply a comparison of the acquirer and target size (Draper & Paudyal, 2006, Faccio et al, 2006). Research points out that the abnormal retruns increase when the target size decreases. So there is a negative relationship between the targetsize and the announcement effect.

An acquisition can be either friendly or hostile. In case of a hostile takeover it means the target does not want to be acquired by the acquiring party. A friendly acquisition means the target company is willing to sell to the acquirer (Schwert, 2000). Hostile takeovers trigger a substantially larger price reaction than a friendly bid.

The difference between the announcement effect when a target is listed or unlisted on the stock market is the so called listing effect. When a target is listed the returns caused by the announcement will be lower than an acquisition concerning an unlisted target (Draper & Paudyal, 2006).

Another characteristic is the time period the acquisition was done. The moment in time can have an influence on the announcement effect (Faccio et al, 2006). The country where the target is located is of importance as well. In several studies a clear difference in the announcement effect between acquisitions with a target in the same country as the acquirer and a target in a cross-border country has been measured.

Regarding the CIBS acquisitions I will test the sample used in this thesis on several characteristics. These characteristics are chosen because of their proven significant influence in other studies.

(13)

13

government intervention, capital restrictions, market for corporate control and cost of capital. The less restrictive the institutional environment is, the less rules and restrictions can intervene the prosperity of a company. It is assumed that when the target company has a less restrictive institutional environment, this will reduce agency problems and asymmetric information and higher bidder gains can be expected.

These are the possible underlying reasons for differences between the countries targets are located in. In earlier research it is shown that with exception of the UK as target country there is a negative relationship between the bidder returns and the target country’s economic restrictiveness (Moeller and Schlingemann, 2004).

2.3 Why difference is likely between developed countries and CIBS

This paragraph will explain why I expect there will be differences between the announcement effect in developed countries and CIBS. I expect there will be a difference between the announcement effect in developed countries and CIBS because of three possible reasons.

First of all, as stated in the previous paragraph, the general characteristics influence the announcement effect. It is possible that the characteristics have a different influence on the announcement effect in China, India, Brazil and South Africa. In numerous researches on developed countries it appears there are differences between the influences of general characteristics on the announcement effect in the United States and Europe. There are even differences between the United Kingdom and Continental Europe.

Secondly, it is possible that the characteristics of acquisitions are differently distributed and therefore the sample will show different results. For example the acquisition methods in the Middle East are different from those used in other markets such as Southeast Asia and Latin America (Metwalli and Tang, 2003). There is a different distribution of acquisitions in cash and shares. The method of payment can also be a reason for differences among developing countries as it is amongst developed countries. Existing literature shows differences exist in the methods of payment between developed economies situated in different continents. For example there is a significant difference between the United Kingdom and the United States. The developing countries accounted for in this thesis are situated in different continents as well.

(14)

14

is the increase in value of goods and services produced by an economy. In economics it refers to potential output and can be measured as the percentage of growth in gross domestic product (GDP). The GDP growth over 2006 was a mere 3 percent for developed countries and a 7.3 percent for developing countries (World Development Indicators database, 2007). Another factor can be the macroeconomic volatility, which causes more risk for investments like acquisitions. Volatility refers to the amount of uncertainty or risk about the size of changes in an economy. A higher volatility means that the economy can potentially be spread out over a larger range of values. This means that the price of the security can change dramatically over a short time period in either direction. A lower volatility means that a security's value does not fluctuate dramatically, but changes in value at a steady pace over a period of time. This factor will not so important for domestic acquisitions, because the targets are in the same economy as the acquirer. However it can be of influence on cross-border acquisitions, because then the target company is situated in another economy. This research can provide actual proof of these other studies and the assumption that these deal characteristics will influence the announcement effect. Therefore this study can really add to the existing literature.

2.4 Focus on Cross-border versus Domestic acquisitions

As mentioned before the empirical studies done on the topic of the effect of acquisition announcements on the returns for shareholder in developed countries state with consensus that target firms gain a significant wealth at the time of the announcement. However these studies can be categorized into domestic and cross-border studies (for domestic deals e.g. Jarrell & Poulsen, 1989; Servaes, 1991; Kaplan & Weisbach, 1992; for cross-border e.g. Harris & Ravenscraft, 1991; Cebenoyen et al. 1992 and Cheng & and Chan, 1995). The same categorization can be made for literature solely covering the acquiring firms. For domestic deals several studies pointed out a negative or an insignificant return (Walker, 2000; Michell & Stafford, 2000; Sirrower, 1997 and Healy, Palepu & Ruback, 1992). Studies performed on cross-border point out different results. A number of performed studies have shown that the stock price of acquiring companies increased around the announcement date (Morck & Yeung, 1992; Markides & Ittner, 1994; Harris & Ravenscraft, 1991). Another group of studies claims the acquisition announcements have a negative or insignificant impact (Datta & Puia, 1995; Shimizu et al., 2004).

(15)

15

Previous research on domestic and cross-border deals is mostly focused on the UK and the U.S. According to several studies a significant return for shareholders can be realized with cross-border deals. Similar to the cross-cross-border deals, domestic ones result in significant returns as well (Harris & Ravenscraft, 1991; Cebenoyan, 1992; Cheng & Chan, 1995). A comparison done by two studies (Conn & Connel, 1990; Feils, 1993), conclude that the percentage of returns from cross-border acquisitions between the UK and US differ. The return for US shareholders is significantly higher than the return for the UK shareholders (40% vs. 18%). A study on the difference of stock return between domestic and cross-border deals proved a significant difference between the shareholder returns (Moeller & Slingemann, 2004). Authors of researches concerning this topic suggest the difference lies in the differences in bid characteristics (Wansley et al., 1983; Dewenter, 1995; Danbolt, 2002). The bid characteristics are aspects of the bid, which can make a bid more or less attractive (e.g. payment in shares versus cash). This means the characteristics of the sample of the deals done by UK companies and companies from the United States differ.

(16)

16

(Shaked et al., 1991). Lastly, doing cross-border acquisitions is an activity which has rapidly developed the last decade and therefore numerous of companies may still be learning from these ventures, implying they are not yet efficient in acquiring and picking out the right targets (Aw & Chatterjee, 2004).

What would the difference of a domestic or a cross-border acquisition mean for the acquirer shareholders returns in CIBS? To test the previous question this study comes to its second hypothesis which is split up in two.

H2a: When everything else is equal, the abnormal return is positively significant when the acquisition involves a domestic target.

H2b: When everything else is equal, the abnormal return of domestic acquisitions differ significantly from cross-border acquisitions.

As can been seen by all the earlier mentioned studies, numerous researches have been done primarily on the US and European continent. Several of these studies have pointed out that there are significant differences between the US and Europe in the stock return characteristics after an acquisition announcement. The characteristics of acquisitions done by CIBS can differ per country or continent and therefore influencing the result on the announcement effect. Therefore, when doing a research which uses a sample that includes multiple continents, it can be useful to compare the outcomes between the continents. If I would only investigate the entire sample of acquisitions in CIBS, I would be ignoring the findings of other researches on different continents. The results for the entire sample can be far from representative for a specific country. In order to test the data in case of differences in the announcement effect between CIBS I have formulated the following third hypothesis.

H3: When everything else is equal, there is a difference in the abnormal return between CIBS.

(17)

17

(same country as the acquirer or another country). In cross-border research performed on the developed markets such as the U.S. and the U.K. market, results show there is a difference in the announcement effect between cross-border acquisitions made in developed and developing geographical areas. Because of the smaller risk in developed countries and their stable economic performance I expect a larger return will be generated when a cross-border acquisition concerns a target from a developed country. To the extent of this study these results arouse interests in the effects cross-border acquisitions in developed and developing countries have on the announcement effect in CIBS. Therefore the fourth hypothesis will test the cross-border target effect.

H4: When everything else is equal, returns for acquirer shareholders are larger when a cross-border acquisition has a target located in a developed country

instead of a target in a developing country.

The previous hypotheses and literature show the focus of this thesis.

(18)

18

3 Data

This part will explain the origin of the data sample used as well as the characteristics of the data which are put into the sample. The first paragraph will point out the data resources and how the data is constructed. The second paragraph will discuss the data characteristics.

3.1 Data resources and the sample selection

For this research a sample of acquisitions done between the first of January 2000 and December 31, 2007 has been used. Source is the database called Zephyr of “Bureau van Dijk Electronic Publishing”. The sample group will exist of acquisitions done by China, India, Brazil & South Africa and targeting companies in China, India, Brazil & South Africa as well as the rest of the world. The target countries in the rest of the world can be divided in developed and developing countries using the list of developed and developing countries created by the Worldbank. All acquiring companies are listed and the targets are private companies. The deals picked for this research are acquisitions which are both domestic (within the acquirer’s country) as well as cross-border (outside the acquirer’s country). The minimum deal size is 10 million US dollars, because the information on smaller acquisitions is often insufficient. The Zephyr database comes up with a sample of over 800 acquisitions with the previous mentioned restrictions, however because of other restrictions and missing data in this sample the eventual sample exists of 426 acquisitions. Table 2 shows the construction of the sample from the beginning until the final sample.

(19)

which lack disclosing the needed relevant information, which is most often the case when retracting information on these relatively small deals from the Zephyr database. Another restriction is whether the deal is paid for in cash or shares. According to several articles there is a significant difference between acquisitions paid for by cash and shares. The last mentioned restriction in table 2 is valid data. After having collected all acquisitions I deleted those acquisitions, which had data missing or which comprised data which seemed invalid. The missing data was caused by the date a company got listed or was de-listed. So it could be there was a streak of data missing in the estimation period or just after the announcement. There were some return indexes which had gaps in the data as well. Several return index lists also appeared to be invalid. Some had the same return in 30 days. Those acquisitions were also excluded from of the sample.

Number of

Restriction Specified options or values acquisitions left Difference

Country Acquirer China, Brazil, India and South Africa 24,558 Target All countries

Time Period From 1/1/2000 to 31/12/2007 20,182 4,376 Announced and completed

Deal Type Acquisition 8,084 12,098

Quoted companies Acquirer - Quoted 2,055 6,029 Target - Unquoted

Vendor - All

Deal Value Minimum = 10 million US$ 785 1,270

Method of Payment Cash and Shares 611 174

Valid Data All fields 426 185

Table 2: Sample construction

To gather the needed stock information I have used the “Datastream” database. Datastream is a historical financial numerical database, which provides a range of charting and reporting tools that enable users to manipulate and display, or simply download data. In order to gather the data required, I have extracted the daily total return indexes, from which the daily return can be calculated (RI) and the market return (MR) from the same database. To calculate the difference between the returns measured after an announcement and the expected return without an

(20)

announcement (abnormal return), I have used the daily market returns. The market information is based on the indices from the provider of the Standard & Poor indices.

3.2 The sample characteristics

This section will display characteristics of the sample. I will show how the acquisitions are distributed over the years, over the countries and over several acquisition characteristics.

3.2.1 Distribution over CIBS

Table 3 displays combination of an overview of the distribution of the acquisitions over the years and the distribution of acquisitions over the different sample countries (CIBS).

No. % No. % No. % No. % Total per year

2000 0 0% 0 0% 4 40% 6 60% 10 2001 6 35% 2 12% 4 24% 5 29% 17 2002 3 33% 1 11% 3 33% 2 22% 9 2003 14 36% 11 28% 10 26% 4 10% 39 2004 43 64% 7 10% 9 13% 8 12% 67 2005 47 77% 6 10% 4 7% 4 7% 61 2006 80 73% 11 10% 5 5% 14 13% 110 2007 87 77% 10 9% 8 7% 8 7% 113 Total 280 48 47 51 Total 2000-2007 426 South Africa

China India Brazil

Table 3: Overview distribution acquisitions over CIBS and years

As can be seen in table 3, the companies in China realize significantly more acquisitions than the companies in the other countries from 2004 onwards. Therefore the sample is biased towards China. This table clearly shows an acquisition boom starting in 2003 for CIBS. The first four years of the sample differ significantly from the second four years, when looking at the quantity

(21)

of acquisitions. In the appendix a clear graph is displaying the acquisition cycle in CIBS. In the last three years of the sample it appears the distribution of acquisitions has levelled out and shows very small yearly changes.

3.2.2. Distribution over domestic versus cross-border

The main acquisition characteristic in this research is whether the acquisition is cross-border or domestic. Table 4 illustrates the dispersion of this characteristic in this sample. As you can see far more than the majority of the sample comprises domestic deals. Amongst the cross-border deals, developing targets are apparently more interesting for companies from CIBS. It is remarkable to detect how few cross-border acquisitions Chinese companies relatively execute. China is responsible for three quarters of the total domestic deals and for 15 percent of the cross-border acquisitions. The explanation lies in the fact the Chinese government restricts the investment in foreign companies by Chinese companies and the other way around.

% of total acquisitions

Country No. % No. % No. % No. %

China 270 75,21% 10 14,93% 9 18,37% 1 5,56% India 22 6,13% 26 38,81% 22 44,90% 4 22,22% Brazil 32 8,91% 15 22,39% 7 14,29% 8 44,44% South Africa 35 9,75% 16 23,88% 11 22,45% 5 27,78% Total 359 100% 67 100% 49 100% 18 100% (cross-border) (cross-border) 84,27% 15,73% 11,50% 4,23% Cross-border

Domestic Developed Target Undeveloped Target

Table 4: Dispersion of acquisitions over domestic and cross-border

3.2.3 Distribution over other characteristics

To test the influence of other characteristics I have gathered information about the sample acquisitions as well. The other characteristics taken into account are the period of acquiring and the method of payment. In table 5 both characteristics are combined.

(22)

22

Total Method of payment No. % No. % Paid in Cash 62 96.88% 234 75.00% 296 Paid in Shares 2 3.13% 78 25.00% 80 Total 64 100% 312 100% 376 2000-2003 2004-2007 Period in years

Table 5: Dispersion of acquisitions over periods and the method of payment

(23)

4 Methodology

In the chapter “methodology” I will explain the method used to perform the event study. The methodology of this research resembles the method elucidated by MacKinlay in his article “Event studies in economics and finance”. After explaining the event study I will clarify how the data will be test on validity and how I will perform an Ordinary Least Squares (OLS) regression.

4.1 Event Study

The essence of this research is finding out what kind of effect an event has on the value of firms. To perform this research a measure can be constructed using an event study. An event study is a statistical study that examines how the release of information affects the return of stock at a particular time. Figure 2 displays the relevant relations between the different elements in a conceptual model for this event study. The event study is a research method developed over 30 years ago. In the 1980’s the method gained widespread use and acceptance as an important research tool in the field of finance (Wells, 2004). As the result of Brown and Warner’s (1980, 1985) researches on event studies, empirical studies in finance and accounting started to make extensive use of this methodology. Using financial market data, an event study measures the normal returns of stock (MacKinlay, 1997). Event studies attempt to measure the abnormal returns of the listed companies caused by an event such as the announcement of an acquisition. The abnormal return is the difference in return between the actual and the normal return. The normal return can be calculated by two different models. These models are the mean adjusted model and the market model. Because the stock market is a form of auction market, the

Deal characteristics: - cross-border vs domestic - payment method - period - country Acquisition announcement

Total return index of acquiring firm

Figure 2: The conceptual model

(24)

supply and demand should react on the shared information. Therefore an acquisition announcement can have impact on the total return index (Wells, 2004).

How exactly to perform a proper event study, will be described in the following sub-paragraphs in a chronological order.

4.1.1 Defining the event and setting the event period

The initial task of conducting an event study is defining the event (MacKinley, 1997), which in this case is the acquisition announcement. The measure of the value of the firm is in this case the total return index which indicates the daily return shareholders receive. The total return index of the firm is actually the result of all fluctuations in shareholders’ wealth. Because the total return index is only available for listed companies this research will only include listed companies as acquirer.

After defining the event it is important to set the event period over which to evaluate the stock returns. To calculate the effect of the announcement on the acquirers’ returns, the announcement date is denominated as day 0 and an event window around the announcement date is defined. The event window is used to control leakage of information (rumors) before the announcement date and a delayed effect on the acquirers’ returns after the announcement date. In this research a multiple event windows varying from one day before and one after the announcement to twenty days before and twenty days after, based on MacKinley (1997) will be used and corresponds to Morck et al (1990), Servaes (1991), Goergen and Renneboog (2004) and Martynova and Renneboog (2006). An estimation window ending before the event window is used to estimate the acquirers their normal (expected) returns (NRs). In this research I will use an estimation window of 180 trading days before the event window. The event window is the period from twenty days before the announcement until twenty days after. This means the estimation period starts at 200 days before the announcement. This is based on the average of other studies like MacKinley (1997), which recommends 120 trading days, Servaes (1991), 210 trading days and Martynova and Renneboog (2006), 240 trading days before the event window. Figure 3 illustrates the construction of the time windows used to conduct the event study.

τ = 0 Event window Estimation Window

τ = -200 τ = -20 τ = +20

(25)

Figure 3: Graphical display of the estimation and event window

The next step is identifying which companies will be affected by the event. Only the acquiring companies are of concern in this study, because the targets are not listed. For companies which are not listed it is not possible to get the daily total return indices. It would be more interesting to perform a research on the targets as well, because then I would also have results for the announcement effect for target companies. However, the available sample of listed targets is too small. This research will look at the returns for the shareholders of the acquiring firm. This way I can see if, as mentioned earlier, if the acquisitions have an increasing effect on the shareholders wealth. Following up on identifying the relevant companies is the calculation of the abnormal returns.

4.1.2 Abnormal return

To calculate the impact of the event, the abnormal return has to be calculated. The abnormal return is the actual return of the security over the event window (the period from 20 days before the announcement until 20 days after) minus the normal (expected) return of the firm over the event window. For firm i and event date τ the abnormal return is;

)

( _it _t

it

it R E R X

AR = − (1)

where

AR

_it ,

R

_it and E(R_it X_t

)

are the abnormal, actual, and expected (normal) returns respectively for time period

t (Mackinley, 1997)

. The conditioning information for the normal

return model is

X

_t. The returns are calculated with the following formula.

1 1 − − − = t t t it I I I R (2)

In this formula is equal to the total return of the share of firm I on time t. and are equal to the total return index of the share, on respectively t =1 and t = -1.

it

R

I

_t

I

_t₋₁

The normal (expected) return ( )) can be specified by conducting a market model. This will calculate the expected returns of the stock.

it

R

E(

(26)

4.1.3 Market model

For this study the market model will be used, because this model is the most accurate when calculating the expected (normal) return. The reason lies in the fact that the model takes into account the volatility of the markets where the stocks are listed. This way the return of the stock around the moment of the announcement will be compared with the relative return of the market portfolio the concerning stock is in. The model uses the market return of the specific market the acquiring company is listed on and the company’s alpha and beta, which reflects the company’s behavior towards the stock market. The model removes the variation related portion of return; therefore the variation of the abnormal return is reduced. This is the reason why the market model is better than other methods. This means that in case of volatility on the market, the abnormal returns of the specific stock of a company will be compared with the market movements the stock is in. Therefore the calculated abnormal return in this event study is more accurate than an event study which does not use the market model. For example, the case of the crash of stocks (a large decrease of value of the stocks) which occurred after the 9th_{of September 2001 in the U.S. would}

be taken into account when calculating the abnormal returns for a stock in that specific period. Therefore the abnormal return displays more accurately the relative abnormal return of the stock compared to the market it is in. For any security the market model can be expressed by the following formula. The formula is used to calculate the actual return during the estimation period and will then be used to determine the normal (expected) return for the event window.

it mt i i it

R

=

α

₁

+

β

+

ε

(3)

where and are the period-t returns on security I and the market portfolio, respectively, and

it

R

_mt

it

ε

is the zero mean disturbance term. The parameters of the market model are

α

_i₁ and

β

_i. These parameters are estimated trough a line-estimation from the regression on over an estimation period. After running the market model regression for all the values in the sample during the estimation period (t = -200 to -21), I will use the parameters to forecast the expected return ( )). To be clear, the market model determines how the specific return of a company and the return of the market relatively perform in order to estimate the expected return during the event window. Practically this means I will extract all the market index values from the Datastream database, which specify the market return per day for the complete estimation

it

R

_mt

it

R

E(

(27)

window. These values for the estimation period will then be used to forecast the expected returns of the specific stock during the event window.

4.1.4 Average abnormal returns and cumulative average abnormal returns

When using the market model I can determine the abnormal and average abnormal returns. The formula for the abnormal return is the following.

− = − − = _it _i _i _mt _it it R R R AR

α

ˆ

β

ˆ

E

(

R

_it

)

(4) where

)

(

)

(

)

(

R

it

E

i

E

t

R

mt

E

=

α

+

β

(5)

To calculate the daily average abnormal return for the sample during the estimation window, you have to accumulate all of the daily abnormal returns first. It is important not to include the returns during the estimation window, so I will only use the 41 days in the event window. After aggregating all these abnormal returns, the outcome has to be divided by the number of your sample to amount to the average. In formula written the average abnormal return is the following quotation.

∑

=

N i it t

AR

N

AR

1

(6)

where ARt stands for the average abnormal return for period t.

So because we can now calculate theARt

,

we know what the average abnormal (actual minus

expected) return amounts to per day. To be clear, the average abnormal return per day is the average of all the abnormal returns of the stocks included in the sample. This abnormal return is the actual return compensated with the market model and explains the effect caused by the event; the acquisition announcement. So when the abnormal returns differ significantly from zero for the

(28)

period that is looked at, this will mean the acquisition announcements had an influence on the stock price.

With all the daily average abnormal returns at hand, the cumulative abnormal return (CAR) can be calculated. The CAR is aggregating daily average abnormal returns and therefore the CAR can be used to look at the influence of the event in certain intervals, for example (-1,+1). It means the CAR for the interval (-1,+1) is the average abnormal return for day -1 plus the average abnormal return for day 0 plus the average abnormal return for day +1. This is the formulation of the CAR:

∑

=

2 2 , 1 1 t t t it t t

AR

CAR

₍₈₎

So CARt1,t₂ is the cumulative abnormal return for a specific period. In chapter five with the result

CAR tables will display the helpfulness of the CAR.

4.2 Statistical tests

The aggregated abnormal return for period t has to be tested to verify the announcement effect on the value of the firms. To find out if the outcomes differ from zero are valid I will have to see whether the abnormal returns are normally distributed (Brooks, 2002). Without assuming a normal distribution, it could interfere with the validity of the parametric test (T-test). To test the normality I will use the Jarque-Bera test. This test is based on the skewness and kurtosis of the sample. The Jarque-Bera value will be calculated in Eviews (statistical software program). If needed for this study I will also perform a non-parametric rank test for abnormal performance in event studies (Corrado, 1989) to test the robustness of the results (Chang, 1998). This is in case the sample is not normally distributed. In that case the T-test is no longer valid.

(29)

5 Results

Following up on the previous chapter on the used methodology to gather the desired data in order to analyze the announcement effect in CIBS, I will now discuss the results from the event study. The structure of this chapter follows the same sequence as the hypotheses stated in the second chapter. The first paragraph will discuss hypothesis 1 and 2 and then each following paragraph will discuss the succeeding hypothesis.

5.1 Announcement effect for acquirers from CIBS

This first paragraph will discuss the results for the total sample. These results will answer the following hypothesis.

H: When everything else is equal, there is a positive abnormal return for the shareholders of the acquiring firm in CIBS around an acquisition announcement.

The results I found confirm the first hypothesis. On the day of the announcement it appears an average abnormal return is measured of 1.10%. Information on the figures of average abnormal returns can be seen in the Appendix B. Cumulative abnormal returns are stated in Appendix C. Figure 3 clearly shows the average abnormal returns (AAR) over the event window.

N = 426 -0.40% -0.20% 0.00% 0.20% 0.40% 0.60% 0.80% 1.00% 1.20% -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 Days

Total CIBS sample

Figure 3: Average Abnormal Return of the complete data sample

(30)

There is a large peak at t =1, just after the acquisitions announcement. Before and after the peak at t = 1 the average abnormal return oscillates randomly. So it appears the announcement does not have an effect before of after the significant peak.

Figure 4 displays the cumulative abnormal return (CAR) over the event window. This figure also shows the jump in CAR which represent the peak in at t =0 in the AAR.

N = 426 -1.00% 0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00% -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 Days

Total CIBS sample

Figure 4: Cumulative Abnormal Return All acquisitions

So the results authenticate what was stated by other researches mentioned in chapter 2. The announcements of acquisitions affect the average abnormal returns of the entire sample including China, India, Brazil and South Africa; the announcement effect theory holds.

As mentioned in chapter four, looking at several intervals will help to gain better insights into the AAR behaviour. By looking at different intervals within the event window we actually look at the announcement effect for different periods. It is possible there is a strong announcement effect in a short period around the announcement, even when there is not a significant impact for the complete event period (-20,+20). Table 6 shows the CARs for the 8 different intervals chosen to display various intervals for the announcement effect. In this case it is clear to see there is a strong significance. Each interval reached a significance position 1 and all have high T-stat levels.

(31)

Interval CAR T-stat Sig. (-20, +20) 4,59% 7,18 sig 1 pos (-10, +10) 4,20% 9,18 sig 1 pos (-5, +5) 3,18% 9,60 sig 1 pos (-2, +2) 2,26% 10,10 sig 1 pos (-1, +1) 1,81% 10,45 sig 1 pos (-1, 0) 1,19% 8,41 sig 1 pos (-10, -1) 1,09% 3,46 sig 1 pos (+1, +10) 2,01% 6,38 sig 1 pos

Table 6: CARs for the entire sample

In the appendices you can find the list with the abnormal average returns on each day of the event window. The cumulative abnormal return is used to determine the significance of returns during a certain period. For example all the returns from day minus 5 to plus 5 accumulated is 3.18%, which is the CAR for interval (-5, +5).

To be able to say more about the effect characteristics of the acquisition have on the announcement effect, I will discuss the other hypotheses with the same structure as just done with the total sample. For the other characteristics the AAR tables and CAR tables can be found in the appendices as well.

5.2 Cross-border versus Domestic

This paragraph will proof if the second hypothesis is either right or wrong. The hypothesis is the following:

H2a: When everything else is equal, the abnormal return is positively significant when the acquisition involves a domestic target.

H2b: When everything else is equal, the abnormal return of domestic acquisitions differs significantly from cross-border acquisitions.

The results of the gathered data clearly confirm the main characteristic of this research to have an influence on the announcement effect. Both the “domestic-line” and the “cross-border-line” have a peak at around the announcement day. However in case of the announcements of acquisitions with cross-border targets a large part of the abnormal returns are negative (see figure 5).

(32)

N = 426 -1.00% -0.50% 0.00% 0.50% 1.00% 1.50% 2.00% -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 Days Cross-border (n=67) Domestic (n=359)

Figure 5: Average Abnormal Return Cross-border and Domestic Acquisitions

It appears that acquisitions with a domestic target have a significant positive abnormal return unlike the acquisitions with a cross-border target, when looking at the event window period (figure 6). The average abnormal returns for domestic acquisitions peak around t = 1 and then the volatility in the abnormal returns slowly extinguish. The cross-border AAR however peaks even more around the announcement date, but then drops to negative returns. After day 8 the AAR stays negative throughout the rest of the event window to an extent of 2.46% in the red. It looks like the actual announcement has an impact even a bit larger than the domestic one.

In the appendix you can find all the actual values per day of the AAR and the CAR, on which previous graphs are based. These results confirm hypothesis 2a as well as 2b to be accurate and support other outcomes of some studies in developed countries, even more strongly so than other studies.

(33)

N = 426 -3.00% -2.00% -1.00% 0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00% 7.00% -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 Days Cross-border (n=67) Domestic (n=359)

Figure 6: CAR Cross-border and domestic acquisitions

When we compare the outcomes for the domestic and cross-border acquisitions to the entire sample it seems clear the domestic and cross-borders acquisitions differ and the domestic acquisitions have the largest impact on the entire sample. This means that the percentage of domestic acquisitions is more substantial than that of the cross-border acquisitions. The appendix clearly displays the outcomes compared to the entire sample.

The statistical difference between the domestic and the cross-border acquisitions can be seen in table 7. For the entire event window there is a significant difference between domestic and cross-border acquisitions. On the day before the announcement there is a significant difference as well. The delta CARs are the cumulative abnormal returns of cross-border acquisitions minus the domestic CAR for the same period. So the day before the announcement the cross-border acquisitions have a significantly positive return compared to the domestic acquisitions.

Interval Delta CAR T-stat Sig. (-20, +20) -8,32% -2,37 sig 5 neg (-10, +10) -5,75% -2,28 sig 5 neg (-5, +5) -1,99% -1,09 0 (-2, +2) 0,86% 0,70 0 (-1, +1) 0,96% 1,01 0 (-1, 0) 1,31% 1,69 sig 10 pos (-10, -1) -2,71% -1,56 0 (+1, +10) -3,43% -1,98 sig 5 neg

Table 7: Delta CARs for domestic and cross-border acquisitions

(34)

34

The cause of the results can lie in various explanations. If the acquirer can secure firm lever economies of scale (Scherer & Ross, 1990), acquisitions with cross-border targets would demonstrate larger returns (as stated in chapter 2). This is not the case of CIBS, because the CAR with interval (-20, +20), the event window period, of acquisitions with a cross-border target is negative. A negative CAR means that the total of average abnormal returns is negative. Therefore the announcement did not result in more wealth for the shareholders. It seems that other characteristics of domestic and cross-border deals are more important in the case of CIBS. The loss of control (explained in chapter 2) can be seen as a downside by investors (Ravenscraft & Scherer, 1987). The shareholders’ attitude towards foreign acquisitions can play a role, because investing in foreign targets, which are less familiar, can harm the investor’s confidence. Another reason can be the extent to which governments interfere with the economy. Less interference will result in higher bidder returns (Moeller and Schlingemann, 2004). Other possible explanations for lower returns, as mentioned in chapter 2, can lie in cross-border differences as language, accounting and politics. These factors can result in negative abnormal returns (and accumulated it results in a negative CAR). The explained factors can cause acquirers to be unaware of certain elements of the acquisitions. It can cause an acquirer paying too much for the target, the so called ‘winner’s curse’ (Eckbo, 1992). Another explanation is the fact that acquirers might be overpaying because they pay a premium for gaining foreign market share, which other companies in the target country would not pay (Shaked et al., 1991). Therefore the company pays a premium which is actually to high for the target acquired. This can result in a lower stock price, which results in a negative CAR.

5.3 Difference in announcement effect between CIBS

The following section will discuss the result on the differences in the announcement effect between CIBS. The hypothesis to test the difference stated below will be answered.

H3: When everything else is equal, there is a difference in the abnormal return between CIBS.

(35)

-1,50% -1,00% -0,50% 0,00% 0,50% 1,00% 1,50% 2,00% 2,50% -20 -18 -16 -14 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 Days China Brazil India South Africa

Figure 7: Average abnormal return per Country

From the results several conclusions can be drawn. We can see that the largest announcement effect is noticeable in India with an abnormal return of just over 2 percent just after the announcement. China has just over a percent of AAR, just like South Africa. Brazil however has a small negative AAR on the announcement day. The AARs of CIBS result in the CAR graph displayed in figure 8. This graph reveals the significance of the outcomes as well as a clear difference in the announcement effect within the developing countries CIBS. China has the highest event window CAR, followed by India. India’s CAR goes up rapidly around the moment of announcement. After five days the stock price declines again. Brazil and South Africa however show different returns. Brazil and South Africa their CARs decline instead of going up like China and India. South Africa its CAR however does rise on the announcement day unlike Brazil’s CAR. This might as well be a coincidence though, because of the volatile movement of South Africa’s returns throughout the entire event window.

(36)

-4,00% -2,00% 0,00% 2,00% 4,00% 6,00% 8,00% -20 -17 -14 -11 -8 -5 -2 1 4 7 10 13 16 19 Days China Brazil India South Africa

Figure 8: Cumulative abnormal return per country

As said before the CAR graph reveals how significant the outcomes are. As stated in table 7, significance is present in the results of China and India and not in Brazil and South Africa. As stated in paragraph 5.2, it matters if an acquisition has a domestic or a cross-border target. Therefore a bias towards more acquisitions with a domestic or cross-border target in a sample concerning one country can affect the outcome of the announcement effect for that specific country. Less than 4 percent (see table 8) of the acquisitions done by Chinese companies involved a cross-border target, hence the large positive significant outcome for the entire Chinese sample.

% of total acquisitions

Country No. % No. %

China 270 75,21% 10 14,93% 100,00% India 22 6,13% 26 38,81% 100,00% Brazil 32 8,91% 15 22,39% 100,00% South Africa 35 9,75% 16 23,88% 100,00% Total 359 100% 67 100% 3,57% 54,17% 31,91% 31,37% cross-border targets 84,27% 15,73% target

target domestic targets

Domestic Cross-border Percentage of Percentage of

96,43% 45,83% 68,09% 68,63%

Table 8: Percentage of domestic & cross-border targets of the samples per country

India has done a few more acquisitions with a foreign target than a domestic target. India does however have a large positive significant return for the complete country sample. This is not in

(37)

line with the earlier findings. Brazil and South Africa have a negative total CAR for the event window, which could be because approximately one third of their acquisitions have a cross-border target.

China has a significant outcome in each interval (see table 9). The outcomes for India show that

Interval CAR T-stat Sig. Interval CAR T-stat Sig. (-20, +20) 7,06% 5,27 sig 1 pos (-20, +20) -2,17% -0,49 0 (-10, +10) 6,39% 6,67 sig 1 pos (-10, +10) -0,67% -0,21 0 (-5, +5) 4,81% 6,94 sig 1 pos (-5, +5) -1,44% -0,63 0 (-2, +2) 3,04% 6,51 sig 1 pos (-2, +2) -0,97% -0,63 0 (-1, +1) 2,17% 6,00 sig 1 pos (-1, +1) -0,57% -0,47 0 (-1, 0) 1,25% 4,23 sig 1 pos (-1, 0) -0,15% -0,15 0 (-10, -1) 1,78% 2,69 sig 1 pos (-10, -1) -0,61% -0,28 0 (+1, +10) 3,46% 5,24 sig 1 pos (+1, +10) 0,06% 0,03 0

Interval CAR T-stat Sig. Interval CAR T-stat Sig. (-20, +20) 2,22% 0,89 0 (-20, +20) -0,49% -0,17 0 (-10, +10) 2,21% 1,23 0 (-10, +10) -1,42% -0,71 0 (-5, +5) 3,88% 2,99 sig 1 pos (-5, +5) -2,16% -1,50 0 (-2, +2) 3,46% 3,95 sig 1 pos (-2, +2) -0,23% -0,24 0 (-1, +1) 3,47% 5,13 sig 1 pos (-1, +1) 0,42% 0,56 0 (-1, 0) 2,83% 5,12 sig 1 pos (-1, 0) 0,53% 0,86 0 (-10, -1) 0,73% 0,59 0 (-10, -1) -0,77% -0,56 0 (+1, +10) -0,62% -0,50 0 (+1, +10) -1,67% -1,21 0 China Brazil

India South Africa

Table 9: Interval CARs and significance CIBS

there are significant outcomes up to 5 days around the announcement. Over the entire event window Brazil has a negative return. The only positive return is in interval (+1, +10), but the return is not significant. South Africa has a negative return over the event window as well, but it is less negative than Brazil. The day before the announcement and for the interval (-1, +1) however, South Africa has a positive return. It looks convincing that it is caused by the announcement or the rumors just before the announcement. Unfortunately the result is not significant.

In order to get a good view on the statistical differences between the announcement effect in China, India, Brazil and South Africa, the difference between the average abnormal returns have been compared. Table 10 explains theses differences graphically. For the entire event window

(38)

there is a significant difference between China and the other countries. Between Brazil, India and South Africa there is no significant difference for the event window, however there are significant differences for shorter periods. For the interval (-5, +5) there are significant differences between Brazil & India and India & South Africa.

Interval Delta CAR T-stat Sig. Interval Delta CAR T-stat Sig. (-20, +20) 9,23% 2,00 sig 5 pos (-20, +20) 4,83% 1,70 sig 10 pos (-10, +10) 7,06% 2,13 sig 5 pos (-10, +10) 4,18% 2,06 sig 5 pos (-5, +5) 6,25% 2,61 sig 1 pos (-5, +5) 0,93% 0,63 0 (-2, +2) 4,02% 2,49 sig 5 pos (-2, +2) -0,41% -0,42 0 (-1, +1) 2,74% 2,19 sig 5 pos (-1, +1) -1,30% -1,69 sig 10 neg (-1, 0) 1,40% 1,37 0 (-1, 0) -1,58% -2,52 sig 5 neg

(-10, -1) 2,39% 1,05 0 (-10, -1) 1,05% 0,75 0

(+1, +10) 3,40% 1,49 0 (+1, +10) 4,08% 2,91 sig 1 pos

Interval Delta CAR T-stat Sig. Interval Delta CAR T-stat Sig. (-20, +20) 7,54% 2,44 sig 5 pos (-20, +20) -4,40% -0,86 0 (-10, +10) 7,81% 3,53 sig 1 pos (-10, +10) -2,88% -0,79 0 (-5, +5) 6,97% 4,35 sig 1 pos (-5, +5) -5,32% -2,02 sig 5 neg (-2, +2) 3,28% 3,04 sig 1 pos (-2, +2) -4,43% -2,49 sig 5 neg (-1, +1) 1,75% 2,10 sig 5 pos (-1, +1) -4,04% -2,94 sig 1 neg (-1, 0) 0,72% 1,06 0 (-1, 0) -2,98% -2,65 sig 1 neg (-10, -1) 2,55% 1,67 sig 10 pos (-10, -1) -1,34% -0,53 0 (+1, +10) 5,13% 3,36 sig 1 pos (+1, +10) 0,68% 0,27 0

Interval Delta CAR T-stat Sig. Interval Delta CAR T-stat Sig.

(-20, +20) -1,69% -0,32 0 (-20, +20) 2,71% 0,72 0 (-10, +10) 0,75% 0,20 0 (-10, +10) 3,63% 1,35 0 (-5, +5) 0,72% 0,27 0 (-5, +5) 6,04% 3,11 sig 1 pos (-2, +2) -0,74% -0,40 0 (-2, +2) 3,69% 2,82 sig 1 pos (-1, +1) -0,98% -0,70 0 (-1, +1) 3,05% 3,01 sig 1 pos (-1, 0) -0,68% -0,59 0 (-1, 0) 2,30% 2,78 sig 1 pos (-10, -1) 0,16% 0,06 0 (-10, -1) 1,50% 0,81 0 (+1, +10) 1,73% 0,67 0 (+1, +10) 1,05% 0,57 0

Brazil vs South Africa India vs South Africa

China vs Brazil China vs India

China vs South Africa Brazil vs India

Table 10: Delta CARS for all country comparisons

As discussed in chapter two there are different reasons which can explain the difference between the announcement effect in China, India, Brazil and South Africa. The difference can lie in the

(39)

influence of general characteristics on the announcement effect, the distribution of characteristics in the samples and the difference can be of macro-economic nature.

China’s acquisitions have the highest relative number of domestic deals, which would mean that indeed the domestic acquisitions have a positive effect. However China’s acquisitions are paid for by cash mostly, although payment by means of shares should have a more positive effect according to existing literature.

China’s macro-economic growth is relatively high and so is it’s CAR. However the existing literature also states that more government interference should lead to lower abnormal returns. Apparently the government of China its interference does not lead to low abnormal returns and therefore a low CAR.

5.4 Difference in announcement effect between developed and developing targets

In this paragraph I will discuss the outcomes of a sample of cross-border acquisitions divided in developed and developing targets. As discussed in chapter two the following hypothesis is used to test on the effect of a developed or developing country target.

H4: When everything else is equal, returns for acquirer shareholders are larger when a cross-border acquisition has a target located in a developed country

instead of a target in a developing country.

In total there are 65 valid cross-border acquisitions of which 48 have a target in a developed country and 17 have a target in a developing country. The AAR results show (in figure 9 and in

-1,50% -1,00% -0,50% 0,00% 0,50% 1,00% 1,50% 2,00% 2,50% -2 0 -1 8 -1 6 -1 4 -1 2 -1 0 _-8 _-6 _-4 _-2 0 2 4 6 8 10 12 14 16 18 20 Days Developed target Developing target

(40)

Figure 9: AARs for acquisitions with developed and developing targets

the appendix B) that the acquisitions with the target in a developed country have a very high return on the announcement day, unlike acquisitions with a developing target country. Both have a positive return the day before the acquisition, which might be reactions on rumours.

In figure 10 we can see the CAR results for both categories. It is clear that the companies which

-6,00% -5,00% -4,00% -3,00% -2,00% -1,00% 0,00% 1,00% 2,00% 3,00% 4,00% -20 -17 -14 -11 -8 -5 -2 1 4 7 10 13 16 19 Days Developed target Developing target

Figure 10: CAR for acquisitions concerning targets in developed and developing countries acquire a target in a developed country have a significant positive return around the announcement date. For the return of companies which acquire targets in developing countries there is no significance. It appears that there is no announcement effect traceable when discussing acquisitions of targets in developing countries done by acquirers from CIBS. The acquiring of targets in developed countries however, has significant outcomes from two days before the announcement date until two days after and in all other intervals within the (-2, +2) interval (see table 11).

(41)

Interval CAR T-stat Sig. Interval CAR T-stat Sig. (-20, +20) -1,56% -0,36 0 (-20, +20) -5,03% -1,38 0 (-10, +10) -0,03% -0,01 0 (-10, +10) -2,47% -0,95 0 (-5, +5) 2,34% 1,04 0 (-5, +5) -0,91% -0,48 0 (-2, +2) 4,33% 2,85 sig 1 pos (-2, +2) -0,79% -0,62 0 (-1, +1) 3,64% 3,09 sig 1 pos (-1, +1) -0,27% -0,27 0 (-1, 0) 3,07% 3,20 sig 1 pos (-1, 0) 0,11% 0,14 0 (-10, -1) -0,84% -0,39 0 (-10, -1) -2,23% -1,24 0 (+1, +10) -1,27% -0,59 0 (+1, +10) 0,19% 0,11 0

Developed Targets Developing Targets

Table 11: CARs for cross-border acquisitions with a target in a developed or developing country

5.5 Statistical Tests

To find out if the sample used in this research is normally divided I have used the Jarque-Bera test. This is needed because otherwise the T-test used would be invalid. As can been seen in figure 11, the sample is normally distributed. This graph is for the entire sample, but all the other

0 20 40 60 80 100 120 -1.0 -0.5 -0.0 0.5 1.0 Series: Residuals Sample 1 426 Observations 426 Mean -8.99e-18 Median -0.015932 Maximum 1.022308 Minimum -1.222746 Std. Dev. 0.242490 Skewness 0.680212 Kurtosis 7.719718 Jarque-Bera 428.2452 Probability 0.000000

Figure 11: Result for the normality of the entire data sample

samples are normally distributed as well. According to Corrado [1989] this means the T-test is allowed and the rank test is not needed. The entire sample has a Jarque-Bera value of 428 and a probability of zero.

(42)

42

6 Conclusion, limitations and recommendations

The first paragraph of this chapter will discuss the conclusion which can be drawn from the earlier presented results. The structure of the paragraph will follow the same chronology as the hypotheses. After the conclusion the second paragraph will discuss the limitations of this research. In the final paragraph I will give my recommendations for further research.

6.1 Conclusion

From all the results presented in the previous chapter I can draw conclusions for all the hypotheses used in this thesis. The first hypothesis is supported, which means there is a significant outcome when looking at the effect announcements of acquisitions have on the wealth of acquirer shareholders in China, India, Brazil and South Africa. In short it means an announcement in CIBS will lead to a positive return for shareholders of an acquiring firm. For each of the investigated intervals the cumulative abnormal returns are significant with a peak on around the announcement. The results however are biased towards the Chinese announcement effect, because 280 acquisitions are done by Chinese companies. The difference in the announcement effect will be discussed after the conclusion on domestic and cross-border acquisitions.

The second hypothesis tested the difference in the announcement effect between acquisitions with a target from the same country (domestic) and acquisitions with a target from another country (cross-border). The acquisitions with a domestic target had a significant positive outcome, unlike the cross-border acquisitions. Therefore the distribution of domestic and cross-border deals is important in determining the announcement effect per country. China did over 90 percent domestic deals and has the highest announcement effect of CIBS. Contradictory is that India is the only other country with a significant positive effect; despite the fact that over 50 percent of the acquisitions done by companies from India is cross-border. Around a third of the acquisitions done by companies from Brazil and South Africa were cross-border. So only the results for India are out of the line of expectance, when looking at the percentage of cross-border acquisitions done.

The case of acquirers from China, India, Brazil and South Africa

Shareholders’ Wealth: