The Effect of Corporate Sell Offs on Shareholder Wealth

(1)

The Effect of Corporate Sell Offs on

Shareholder Wealth

Michaël van Erp

(2)

______________________________________________________________________________ The Effect of Corporate Sell Offs on Shareholder Wealth

University of Groningen Faculty of Economics

Master of Science in Business Administration Specialization: Finance

Profile: Corporate Financial Management Supervisors:

Drs. J.A.M.J. Schipperijn Drs. M.M. Kramer

(3)

Acknowledgements

With the completion of this thesis comes an end to my years at the University of Groningen. After having started in September 2000 as a student of the faculty of Management and Organization, I chose, after one year, to go another direction. I started studying economics in September 2001 and received my Bachelor’s degree in Economics and Management in august 2006. This last year I have chosen, within the Master of Science in Business Administration, the specialization Finance. The knowledge I have gained during my last year is the foundation for the research done in this thesis. I chose this topic because it is a very up to date topic that is occurring in many financial markets all over the world. Furthermore it is a topic which was not yet been researched much by other students at this faculty. I found it very interesting to do research on this topic and to write a thesis about it.

There are some people I would like to thank. First of all my supervising professor Drs. J.A.M.J. Schipperijn. His remarks, advise, and positive feedback have helped me a lot during the making and completion of my thesis. Secondly, I would like to thank my parents for their support and care. Finally, I would like to thank my little brother and my older brother and older sister for their support as well.

After a year of many classes, many assignments, one Master Thesis, and hard work I have now finished my final year. I hope you enjoy reading this thesis just as much as I enjoyed writing it!

Michaël van Erp

(4)

______________________________________________________________________________

Abstract

While there has been done an abundance of empirical research on the subject of mergers and acquisitions, little research exists on a closely related topic, the corporate sell offs. The main research question of this thesis examines the effects on shareholder wealth of the announcement by management of an investment decision to sell part of its asset(s) to another firm in the United Kingdom (UK) and in the United States (US). The method used is the event study. The models used are the Mean Market Model, the Ordinary Least Squares Market Model, and the Market Return Model. Two more topics are discussed in this thesis: The Size Effect Hypothesis and the Financial Distress Hypothesis.

The results for the main research question are positively significant for the UK buyer, the UK seller, and the US seller. The US buyer results are, although positive, not significant. Furthermore the Size Effect is tested and the results from the US were significantly positive for the “High Value Group” and therefore in line with those of the Size Effect Hypothesis. Those of the UK were not significant. Finally, both the UK seller and US seller have significant results that are in line with the Financial Distress Hypothesis. These results indicate a period of poor economic performance before the announcement of the sell off.

(5)

Introduction

As a result of corporate restructuring many firms have decided to divest a part of the business. One recent example is ABN AMRO’s intended selling of the ABN AMRO North America Holding Company, which primarily consist of the LaSalle Bank Corporation, to the Bank of America for USD 21 billion in cash. LaSalle offers an important strategic opportunity to the Bank of America to strengthen its position in the US financial market.

In today’s economic environment divestments are a frequently discussed topic. Divestitures are no longer seen as a symbol of failure, but as a way to create and preserve shareholder wealth. Corporate restructuring can take place in many forms (e.g. equity carve out, management buy-out, sell-off and spin-offs). There exist certain reasons that motivate the companies management to do such a divestiture. Depending on which factor the decision is motivated, the shareholder in the divesting firm (the divestor) is likely to experience either an increase or a decrease in its wealth following such a divestment.

In this Master Thesis Finance, an event study will be done on the shareholder wealth effects of corporate divestments. Companies can divest a part of the company (e.g. division or business unit). If a company divests a part of its company, what are then the effects on the wealth of the shareholders? The seller gives up the cash flow associated with an asset in exchange for the cash flow from the buyer. If the exchange produces a positive net present value to the seller and the buyer, the wealth of its shareholders should increase; if the exchange produces a negative net present value, shareholders should see their wealth decline. If a divestiture has real economic value (either positive or negative), then the divestiture announcement should give significant information to the financial markets. One would assume that there would be no significant effects because this would be in contrast with the Efficient Market Hypothesis (EMH). The EMH presumes that all available information is directly processed into the prices of the stocks. An efficient market is described as a market that is efficient when it fully and correctly reflects all relevant information in determining the prices of securities (Fama; 1970). If there were effects than the investor could take advantage of these arbitrage opportunities. Hearth and Zaima (1984) argue that the market for real assets appears to lack some of the conditions of a ‘perfect’ market. They mention that real assets are not completely homogeneous nor are they perfectly divisible. Furthermore, the market for real assets may not be informationally efficient.

(7)

companies but also on the wealth effects of the acquiring companies in the UK and the US stock markets. This will be done by means of an event study.

The main research questions in this thesis will be the following:

1) Is there a relationship present between the divestment activities of a divesting firm and its value in the UK and in the US?

2) Is there a relationship present between the divestment activities of a acquiring firm and its value in the UK and in the US?

Along with these questions some sub questions are formed:

• What are the differences and the similarities in the outcomes of the tests done for the UK and US?

• Does the size of the divestment have an impact on the change in wealth? • Does the Financial Distress Hypothesis hold?

Apart from the central research question these three sub questions will be discussed. It is interesting to see what the differences are between the two countries in the outcomes of the central research question. Furthermore, according to Hearth and Zaima the market reaction on a divestment announcement may also depend on the size of the divestiture. The larger the divestment the larger the reaction of the market should be. Finally, the Financial Distress Hypothesis will be examined. Alexander et al. (1984), Jain (1985), and Kiymaz (2006) found that divestiture decisions tend to be perceived by a period of poor market performance.

These sub questions will be tested next to the main research question of this thesis.

Data on divestments will be gathered from Zephyr. Zephyr is a database for information on mergers and acquisitions. The top 50 most recent divestments will be investigated for the UK and the US market.

(8)

______________________________________________________________________________ divestiture can have an effect on the value of a company. The spread of an announcement can change the value of a stock more than it would have been ‘normally’ the case. This can be tested by calculating what would be the ‘normal’ return of the stock and compare this with the true value. Now an abnormal return can be calculated by subtracting the actual return from the ‘normal’ return. Brown and Warner and MacKinlay use the following three models: the Mean Return Model, the Market Return Model and the Ordinary Least Squares Market Model. The Mean Return Model and the Ordinary Least Squares Market Model will be used to test the null hypotheses. Eventually a t-test will decide if the null hypotheses should be accepted or rejected.

(9)

1. Theoretical Background and Literature Overview

In this chapter an overview will be given of the results of previous research done by other researchers on corporate divestitures. Here the research done on divestitures in the UK and the US will be discussed. Moreover an overview will be given of the theoretical background. First of all some more information on corporate divestitures in general will be given, secondly, some reasons for undertaking a divestiture are presented, thirdly, the theoretical background will be explained on which the hypotheses are based. Finally, an overview will be given of previous research on this topic.

1.1. Corporate Divestitures

As already mentioned in the introduction of this thesis, corporate restructuring can take place in many forms (e.g. equity carve out, management buy-out (MBO), sell-off and spin-offs). The equity carve out is sometimes known as a partial spin off, a carve out occurs when a parent company sells a minority (usually 20% or less) stake in a subsidiary for an Initial Public Offering or rights offering. In most cases the parent company will spin off the remaining interests to existing shareholders at a later date when the stock price is much higher. This is also known as a carve out or an equity carve out. The MBO occurs when the managers and/or executives of a company purchase controlling interest in a company from existing shareholders. In most cases, the management will buy out all the outstanding shareholders and then take the company private because it feels it has the expertise to grow the business better if it controls the ownership. Quite often, management will team up with a venture capitalist to acquire the business because it is a complicated process that requires significant capital.

The two most important corporate divestitures are a spin off and a sell off. In a spin off, the asset(s) divested form(s) a new independent firm with the shareholders of the divesting firm receiving shares in the new corporation. A sell off involves the sale, usually for cash, of the asset(s) to another corporation. The shareholders in this case do not retain a link to the divested asset(s).

According to Koller et al. (2005) divestitures come in waves, like merger and acquisitions. During the 1960s and 1970s many companies refocused their portfolios. These divestitures were generally sales to other companies or private buyout firms. During the 1990s another wave included many more transactions involving public ownership, such as spin offs, carve outs, and tracking stocks.

(10)

______________________________________________________________________________ 1.2. Reasons for Corporate Divestitures

In the literature different researchers give different possible (general) reasons for corporate divestitures.

Hearth and Zaima (1984 and 1986) discuss general motives for a sell off from a divesting and acquiring firms perspective. A firm might want to sell off a part of its firm to reduce the complexity of its operations. Furthermore a firm might use the resources obtained in the sale of the asset to invest in other areas of the firm. The acquiring firm also has a general motive to buy a part of another firm. The buyers do have the potential managerial and financial resources to take full advantage of the asset they bought.

According to Hearth and Zaima one of the main reasons for divesting is that the seller decides to “concentrate on the major operation”. In this case the announcement of a divestiture is significant information for the financial markets and significant positive price movements should occur around the announcement date. Reason for the divestment might be the lack of financial and managerial resources of the divesting firm to optimally exploit the growth opportunities of the asset(s).

Koller et al. (2005) and Hearth and Zaima (1984 and 1985) in short state that divestitures create value when the asset(s) to be sold are worth more to some other owner or in some other ownership structure. In the current structure there may exist unique costs to the parent or the asset(s) that the current owners have to bare which would not exist in another ownership structure.

One can also divest profitable and/or growing businesses. This alternative creates value because the subsidiary will become more competitive, because of the increased freedom for tailored financing and investment decisions, improved management incentives, or better focus. Divestitures may also create value when taking advantage of information asymmetries when fully informed managers take advantage of imperfectly informed managers or financial or strategic buyers. Reasons for this are that real assets are not completely homogeneous nor are they perfectly divisible. Furthermore the market for real assets may not be informationally efficient, some market participants may posses superior information concerning the real worth of the assets being bought and sold. Therefore they assume that in a less than perfect market an asset that is sold could be worth more to the buyer than to the seller.

Koller et al. state that for these potential benefits, companies should regularly divest businesses, even good, healthy ones, so that the corporation can grow stronger and the remaining businesses can reach the full potential.

(11)

the firm’s sell off announcement. Secondly, a sell off can be seen as a partial merger. If certain specific assets of the firm are worth more to outsiders than they are to the current owners of the firm it could be in the best interest of the shareholders to sell those assets. In this case the reaction of the stocks will be similar to those of a real merger, the target firm’s stock will have a significant positive reaction to the announcement of the sell off. Thirdly, sell offs can be compared to spin offs. With a spin off a firm distributes all the common shares of the subsidiary to the existing shareholders. Since spin off announcements are associated with an increase in the value of the common stock, this result is expected for the sell off as well. Finally, there is one more scenario possible, according to Jain. This scenario is based on the agency theory. Here sell offs might be implemented by the desire to transfer wealth. This can be done by paying dividends to the shareholders from the proceeds of the sale of the assets. This way a transfer of wealth takes place from the bondholders to the shareholders, because the money spent on the dividend payments is no longer available to the bondholders. This of course implies that the value of the firm is unchanged. In this scenario the seller also expects a positive stock price reaction.

1.3. Theoretical Background

In this research the wealth effects for the shareholders will be investigated. The main research question is whether there is a relationship present between the divestment activities of a divesting and/or acquiring firm and its value in the UK and in the US. Furthermore, the sub questions are discussed. These questions are all backed by theory from the literature. Now the theory behind the research questions will be explained.

The Efficient Capital Market

The price of stocks is determined by the information available on the stock market. Market efficiency is used to discuss how quickly and accurately new information is incorporated into the prices of the stocks on the capital market.

(12)

______________________________________________________________________________ Prices of stocks do react on new information. When all available information at a certain point in time is incorporated in the stock price and the price reflects this information correctly, one can speak of an efficient market.

In the article of Afshar et al. (1992) the Value Additivity Theory is discussed. This theory assumes strong form efficient capital markets. The value of the divested part is the same whether it is a stand-alone business or a subsidiary of another firm. Here it is assumed that the wealth transfer will not have any effect on the value of the firms. But reasons exist that value creation may arise when divestments take place. An asset may be worth more to another party then to its current owners, the sell off asset may be a loss making asset to its current owners with negative synergies. Its disposal eliminates a wealth destroying resource for the shareholders. Divestments may reduce the level of business activities, and thereby reducing management diseconomies and preserving valuable management resources. Finally, a sell off may be a signal that the company has chosen a carefully thought out strategic redirection, where it focuses on (new) high yielding activities instead of low yielding activities. Hearth and Zaima (1985) describe possible scenarios for the valuation consequences of the sell offs as well. One of these scenarios is that the asset has the same value to both the acquiring as well as the selling party. Therefore neither firms shareholder wealth will change. The announcement incorporates no information to the financial markets. This assumption concurs with the EMH.

In this research the expectation is that, according to the literature, the EMH will hold and that there will be no significant positive or negative abnormal returns.

Size of the Divestitures

The size of the valuation effect is likely to be positively related to the size of the operation divested. The reason for this effects lies in the fact that a larger divestment is assumed to lead to a larger impact.

According to Hearth and Zaima (1984 and 1985) and Borde et al. (1998) the market reaction of a divestment announcement may also depend on the size of the divestiture. So in general the market’s reaction to the announcement should be greater for transactions of larger divestment value.

Hearth and Zaima test three possible hypotheses. The third hypothesis tests whether the size of the divesting asset is of any influence on the return to the shareholders. They found larger sell offs are associated with larger positive price movements.

Borde et al. state that larger divestments are assumed to have a larger impact. In their study, Borde et al. conclude that large divestments have a greater wealth effect than do smaller ones.

(13)

Research will be done to see if for the dataset at hand this hypothesis holds. More on how to approach this research will be explained in the section on methodology.

Financial Distress Hypothesis

The financial condition of a firm will determine the probability of default for a firm. If a decrease in expected cash flow is expected, then the probability of default will increase. Divestitures are often undertaken by firms in financial distress. One way to deal with financial distress is to use the cash from the sale of the asset(s) to repay debt.

Alexander et al. (1984) and Jain (1985) found that divestiture decisions of divesting firms were perceived by a period of poor market performance. According to Alexander et al. for either kind of divestiture, the announcement of a such a divestment should result in an upward movement in the price of the common stock. However they found such a sell off takes place after a period of abnormally negative returns. This would indicate according to them that a divestment occurs when a firm is in financially rougher times. Thus would indicate that the firm was not doing to good at the time before the divestment announcement. The Financial Distress Hypothesis test for the divesting firms whether such a period of poor market performance exists before the announcement date.

1.4. Literature Overview

In this section an overview will be given of previous research done by other writers. The findings will be presented. The influence of divestitures on the wealth of shareholders has been examined extensively in the last years. Most studies done on this topic are done on firms from the US. Outside the US this topic has received little attention. Only six non-US-based divestiture studies are reported: Cao et al. (2006) with UK data; Cooney et al. (2004) with Australian data; Kaiser and Stouratis (1995) with French, German, Swedish, and UK data; Afshar et al. (1992) with UK data; Hamilton and Chow (1993) with New Zealand data; and Capon et al. (1987) with again Australian data.

One of the first studies on this topic were done by Hearth and Zaima (1984). Hearth and Zaima test three possible hypotheses. The first hypothesis tests whether a significant positive excess return exists. The second hypothesis tests whether a strong financial position of the seller signifies a larger return to the shareholders. The third hypothesis tests whether the size of the divesting asset relative to the divesting firm is of any influence on the return to the shareholders.

(14)

______________________________________________________________________________ Furthermore they present higher positive abnormal returns to the seller with the stronger financial position and the larger divestiture size.

Alexander et al. (1984) present positive abnormal returns for divesting firms and moreover state that voluntary sell offs take place after a period of abnormally negative returns.

Initial studies focused mainly on the effects of the sell off announcements from the perspective of the divesting firm only. A second group of studies investigated the influences of corporate sell offs for buyers as well as for sellers. In these studies they find both divesting and acquiring firms benefit from divesting equally.

Hearth and Zaima (1986) continue their previous research by examining voluntary sell offs. A sell off appears to be regarded as a positive influence on the wealth of shareholders of the divesting firm in their article from 1986. For the acquiring firms no significant price movements are observed for the preannouncement period. Not even in the days just before the announcement. These results suggest that the announcement of an impeding sell off has a little impact on the wealth of the shareholders of the acquiring firm.

Jain (1985) states that in a market with a large number of potential buyers, all gains from the expected synergy will flow to the seller of the unique asset and the buyers earn zero excess returns. This implies that a sell off decision is perceived to be a positive net present value transaction for the seller. The results from the study of Jain indicate that the sell off announcements are associated with statistically significant positive excess returns to the sellers. Although, Jain states that the excess returns are much smaller than those earned by target firms in actual mergers.

Jain concludes from his results that buyers are also better off because of the announcement but to a lesser extent than the sellers. Jain concludes the article by indicating that in his research voluntary sell offs have a positive effect on the shareholders of both the sellers and the buyers. But the sellers earn more from a sale than the buyers do.

On the other hand Hearth and Zaima (1985) find insignificant wealth gains for the acquirers. The results for the acquiring firms show some evidence of positive price movements, but although these price movements appear to be positive they are insignificant. The general conclusion from the research of Hearth and Zaima is that divesting firms generally gain from sell offs and that acquiring firms neither gain nor lose.

Kiymaz (2006) finds statistically significant positive wealth effects for both the divesting and the acquiring firms in the US.

A different approach to this topic is taken by researchers who investigated corporate divestitures internationally.

(15)

shareholders gains from such announcements are higher when the completion of a sell off is announced and the price is declared. And financially distressed firms experience statistically higher abnormal returns compared to those of financially healthy firms according to Afshar et al. When looking at the financial strength of the divestor they found this is negatively related to the event day excess return earned by stockholders. The size of the sell off is supposed to be positively connected to the increase in shareholder wealth according to Afshar et al.

Borde et al. (1998) argue that international acquisitions allow for unique opportunities, not characterized in domestic acquisitions, and therefore they could result in unique valuation effects. In the article Borde et al. try to analyze and determine whether foreign divestitures result in different wealth effects than the valuations of domestic divestitures. Borde et al. discover significant positive valuation effects for foreign divestitures of US firms. According to the research of Borde et al. valuation effects are positively related to the size of the divested asset.

In general, existing studies examine the divestitures that took place during the 1970s and 1980s and report statistically significant positive abnormal returns to the shareholders of divesting firms.

Cao et al. state that because of increasing integration of global capital markets firms expand in many countries all over the world. Along with this expansion come advantages but disadvantages as well. Organization and control of a firm that is operating in several different countries at the same time can cause problems and can lead to divestitures.

Morck et al. (1990), and Lang et al. (1995) come to these conclusions in their previous research. The announcement of a divestiture usually results in a positive market reaction for the seller of the asset(s), this conclusion is in line with the findings done by Alexander et al., Jain, and Afshar et al. Cao et al. state that the existing literature like Lang et al. and Chen et al. (1990) offer many explanations for the these abnormal returns such as the desire to concentrate on the core business areas and to eliminate units that are not closely aligned to the companies primary operations.

Cao et al. argues that none of the existing papers investigated the importance of the location of either the divested unit or the buying firm, and if this has any influence on the market reaction. Difficulty with interpretation of information received and perceived from overseas is one of the points that is mentioned by Cao et al. The evidence found by Cao et al. is that the abnormal return is significant and positive on the announcement date, but is larger for the divestitures in which the buyer is located within the UK. This indicates that market participants are more interested in transactions that take place in their domestic market.

(16)

______________________________________________________________________________ elsewhere. Returns resulting from a UK based sale are significantly greater than those for sales located outside the UK.

(17)

2. Data and Methodology

2.1. Data

The data and information on the divestitures are found with the help of Zephyr. Zephyr is a database for information on mergers and acquisitions. This database can be found by using the electronic library on the website of the University of Groningen1_{. With the help of this database the} information on the vendor, the acquirer, the target, the announcement date, and the value of the divestiture can be collected. To find the data necessary for the research the following search criteria have to be entered in the search program of Zephyr. Firstly, the search criteria locator for geography has to be set on the UK or the US for the acquirer as well as for the vendor for each separate dataset. Secondly, the vendor and the acquirer need to be set to be quoted companies. In this research only quoted companies will be used. Thirdly, the current deal status has to be completed. And finally, the company search criteria must be set on division/spin offs. After having set these criteria, Zephyr gives a list of respectively 227 and 898 divestitures for the UK and the US respectively2_.

The data are screened for completeness. Transactions with information missing on the seller, the acquirer, the divestments, the announcement date, and the value of the divestment retrieved from Zephyr are deleted from the sample. Furthermore, only divestitures between national firms are used. This means that only transaction between UK sellers and UK buyers are used (this holds for the US as well). Thereafter the data are ordered by date and the 50 most recent divestitures are selected for the research. Thereafter the stock prices for these 200 firms need to be retrieved.

The data for the prices of the stocks will be drawn from Datastream. Datastream is a statistical data base which consists of stock, macroeconomic, and financial data.

After having filtered the data set for completeness when they are retrieved from Zephyr some criteria apply to the selection of the dataset:

• Transactions from the same firm must be apart at least 6 months. • No overlapping time periods may exist for one and the same firm.

• Accurate stock data must be available for the entire period to be investigated. • Rumors are not incorporated in the research.

Daily returns will be gathered for the stocks from Datastream. The return index (RI) is used. This return index gives a good representation of the value of the stocks because it incorporates the value

(18)

______________________________________________________________________________ of the dividends. This gives a better representation of the value of the company on that day. For the market return index, an index will be used that represents the total market return of that country. These will be the FTSE all share index and the Dow Jones index for the UK and the US respectively. Appendix 2 and Appendix 3 give an overview of the 50 transactions per country. In total there are 100 transactions with 96 divesting firms and 97 acquiring firms. The period in which the divestments take place range from 2007 to 1997 for the UK and from 2007 to 2006 for the US.

The method to be used is the event study. Models to be used are from the articles of Brown and Warner (1980 and 1985) and MacKinlay (1997). The estimation window (EW), the event window, and the post-event window (P-EW) are set at 60, 40, and 30 days respectively. The event window timeframe is based on the timeframe used in one of the articles of a previous researcher and for the EW and P-EW the timeframe is chosen manually. Figure 1 shows the time line used in this research.

Figure 1: Timeline for the Event Study

-80 -21 -20 ‘Day 0’ +20 +21 +50

I---I I---I---I I---I

EW event window P-EW

The event day (‘day 0’) is specified as the first trading day the divestiture information reaches the market. It is important to determine exactly what day this announcement reaches the market, because stock market reactions can only be observed when the news is unexpected. If the date is set after the actual date of the information release, this information will already be incorporated into the price of the stock. It will then not be possible to investigate the real value of the excess return.

A period of twenty days before and after this period is determined as the event window. This period is chosen because it is not entirely certain that ‘day 0’ is chosen correctly. A few days prior to the event and after the event are included in this period. According to MacKinlay this period may take into account information leakage and/or slow market reaction as well as the end-of-the-trading day effects. In this research four different alternative specifications for the event window will be given. This way the changes in market reaction can be observed more precisely. These different time frames are constructed as follows: 41 days (-20,+20); 21 days (-10,+10); 11 days (-5,+5); 3 days

(-1,+1); and 2 days (-1,0 and 0,+1). These timeframes are chosen based on the articles of other researchers who have done research on cumulative abnormal return values and used similar timeframes.

(19)

of a normal control performance model. This model is then used to calculate abnormal stock returns over the event window. The estimation window does never include ‘day 0’. This to prevent the influence of the event window information on the parameters of the normal return model. The post-event window is set at thirty days. First, the absolute differences of each firms stock for the period of the firms event are calculated per day. Secondly, the average per day is calculated per day for all 50 events per country. Eventually a column of 101 days will be left. This column can be divided in an estimation window (60 days) and an event window (41 days).

In Table I an overview is given of the descriptive statistics of the raw data that will be used in the research. This data has not yet been adjusted or changed in any way.

Table I : Descriptive Statistics of the Average Returns3

50 events, 60 days

Estimation Window Average Return

Measures of

Location Measures of Spread Measures of shape Performance measure Mean Median Variance SD Range Min. Max. Skewness Kurtosis UK sellers -0,03% -0,02% 0,000014 0,003677 2,22% -1,14% 1,08% -0,005427 1,869226 UK buyers 0,15% 0,15% 0,000008 0,002791 1,79% -0,41% 1,38% 1,291805 5,246887 US sellers 0,07% 0,11% 0,000020 0,004421 2,03% -1,09% 0,94% -0,445659 0,115384 US buyers 0,13% 0,14% 0,000025 0,004985 2,28% -0,73% 1,54% 0,432473 0,189352 50 events, 41 days Event Period Average Return Measures of

Location Measures of Spread Measures of shape Performance measure Mean Median Variance SD Range Min. Max. Skewness Kurtosis UK sellers -0,02% -0,08% 0,000022 0,004713 2,51% -0,93% 1,58% 1,416610 3,803389 UK buyers 0,16% 0,06% 0,000013 0,003644 1,80% -0,34% 1,46% 1,221088 2,491289 US sellers 0,19% 0,08% 0,000028 0,005288 3,24% -0,61% 2,63% 2,653204 10,978541 US buyers 0,11% 0,04% 0,000019 0,004322 2,45% -0,57% 1,88% 1,780783 5,922615 50 events, 101 days Total Average Return Measures of

Location Measures of Spread Measures of shape Performance measure Mean Median Variance SD Range Min. Max. Skewness Kurtosis UK sellers -0,03% -0,05% 0,000017 0,004106 2,72% -1,14% 1,58% 0,835980 3,242721 UK buyers 0,15% 0,12% 0,000010 0,003148 1,87% -0,41% 1,46% 1,272415 3,640757 US sellers 0,12% 0,10% 0,000023 0,004806 3,73% -1,09% 2,63% 1,238464 6,933530 US buyers 0,12% 0,06% 0,000022 0,004706 2,61% -0,73% 1,88% 0,844109 1,643729

Source: Cooper and Schindler (2003)

(20)

______________________________________________________________________________ The corresponding Bera-Jarque (BJ) test values are presented in Table II. The P-values corresponding to the BJ statistics are 0,000000; 0,000000; 0,408277; 0,389163. This results in the fact that one is able to reject the null hypotheses at a significance levels of 1 and 5 percent for the UK buyer as well as for the UK seller. These sets of residuals are probably not normally distributed and therefore the skewness and the kurtosis probably differ significantly from the normal distribution values of 0 and 3, respectively. For the US buyer and US seller the set of residuals is probably normally distributed although the skewness and the kurtosis differ from the normal distribution values of 0 and 3 respectively, as can be seen in Table I. More on normality of a distribution in the section on methodology.

Table II: Bera-Jarque (BJ) Test Results for the Average Returns of the Raw Dataset

UK Buyer UK Seller US Buyer US Seller

Jarque-Bera 7,158625 6,550664 1,791618 1,887517 Probability 0,000000 0,000000 0,408277 0,389163

Observations 60 60 60 60

Source: Eviews

The data will now be used with the help of the models of Brown and Warner and MacKinlay to answer the central research question of this thesis. This will be further elaborated in the next section on methodology.

Furthermore, to see if the size of the divestitures has any influence on the size of the wealth effect the dataset will be split up into two groups per country. The top 25 firms, based on divestment value, are compared to the 25 firms with the lowest divestment value. The 25 firms with the highest divestment values will be named the “High Value Group” and the 25 firms with the lowest divestment values will be named the “Low Value Group”. In Appendix 4 an overview is given of the 25 high and low value divestitures per county.

(21)

Figure 2: Timeline for testing the Financial Distress Hypothesis -240 -91 -90 +1 ‘Day 0’ I---I I---I---I EW event window 2.2. Methodology 2.2.1. The Event Study

This research is done with the help of an event study. An event study looks at an unexpected event that can be of influence on the value of a firm. It concerns a sudden change that was not predicted by analysts. In the case of our research an announcement of a divestment will be the unexpected event. The day of the announcement can change the value of a stock more than it would have been ‘normally’ the case. This abnormal change in the value of a stock is called the abnormal return. The abnormal return can be determined by calculating what would be the ‘normal’ return of the stock and compare this with the true value at the same moment. Now an abnormal return can be calculated by subtracting the actual return from the ‘normal’ return.

Brown and Warner (1980 and 1985) use the event study on monthly and daily stock returns. According to Brown and Warner the first thing one needs to do in an event study is determine a benchmark. This benchmark is needed to establish a ‘normal’ return. This ‘normal’ return is generally used to see if an abnormal return takes place. Brown and Warner discuss in their articles the following three models: Mean Return Model, Market Return Model, and Ordinary Least Squares (OLS) Market Model. They discuss that an advantage of using daily stock data is that these data are easy to use with an event study. A big difference with monthly data is that daily returns have a substantial difference in outcome from normality which monthly data do not have. Moreover daily data are more sensitive than monthly data to the impact of an event, in this case a divestiture.

(22)

______________________________________________________________________________ 2.2.2. The Abnormal Return

Brown and Warner (1980 and 1985) focus on the returns of each firm around the time of the event. They look at the value of the stock on ‘day 0’ and they investigate if this value differs significant from the time period around ‘day 0’. Days -1 and +1 are considered to be the days around ‘day 0’.

The three models look at the abnormal return (Ai,t). This is the actual return (Ri,t) after the occurrence, subtracted from the calculated ‘normal’ return. The actual return is the value of share i on day t.

This calculated ‘normal’ return differs per model. The ‘normal’ return is defined as the return which would occur if the event would not have taken place. When calculating the ‘normal’ return the relative differences between the days are used, so the different events can be compared. The formula to calculate the relative differences is composed as follows: _

      − − − 1 , 1 , , t i t i t i R R R . 2.2.3. The Models

The abnormal returns in this research are estimated with the help of the Mean Market Model and the Ordinary Least Squares Market Model used by Brown and Warner (1980 and 1985) and MacKinlay (1997).

1. Mean Return Model:

The Mean Adjusted Return model assumes that the ex ante expected return for a given security i in time period t is equal to a constant Ki, which can differ across securities: E(Ri)=Ki

~

.

The predicted ex post return on security i in time period t is equal to Ki. The abnormal return εi,t is equal to the difference between the observed return, Ri,t, and the predicted return Ki: εi,t =Ri,t - Ki. The Mean Adjusted Return model is consistent with the Capital Asset Pricing Model; under the assumption that a security has constant systematic risk and that the efficient frontier is stationary, the Asset Pricing Model also predicts that a security’s expected return is constant (Brown and Warner). The Mean Return Model is specified as follows (Brown and Warner):

(23)

where Ri is the simple average value of stock i in the estimation window (in this research: -80, -21).

2. Ordinary Least Squares Market Model:

The OLS Market Model is a statistical model which relates the return of any given security to the return of the market portfolio (MacKinlay). This is the dominant specification of normal return models used for parameter estimation. It uses the market return as a single factor of a specific security return. The models linear specification follows from the assumed joint normality of asset returns. The OLS Market Model is specified as follows (MacKinlay):

t i t m i i i,t i,t R α β R A = − − , +

ε

, ∧ ∧ (2) where: 2 , , ) var( ) 0 ( i t i t i E ε

σ

ε

= =

where Ri,t and Rm,t are the period-t returns on security i and the market portfolio, respectively, and

t i,

ε

is the zero mean disturbance term. ∧

i α and

∧ i

β are the parameters of the market model from the estimation window of firm i.

2.2.4. The Cumulative Abnormal Return

According to MacKinlay (1997) the abnormal return observations must be aggregated in order to draw overall inferences for the event of interest. The concept of a cumulative abnormal return (CAR) is necessary to create a multi period event window. The CAR can be defined as the sample of cumulative abnormal return from day τ1 to τ2 where T1 ≤ τ1 ≤ τ2 ≤ T2 (T1 = day -20 and T2 = day +20). The CAR from τ1 to τ2 is the sum of the included abnormal returns. The CAR is formed in the following way:

∑

= = 2 1 , 2 1, ) ( τ τ τ τ

τ

_i i A CAR (3)

(24)

______________________________________________________________________________ 2.2.5. The Non-parametric Tests

Until now the methods that have been used are of a parametric nature. The assumptions that has been made is that the distribution of the abnormal returns is normal. Non-parametric tests are tests used when the abnormal returns are not normally distributed. These assumptions are free of specific assumptions concerning the distribution of the returns. According to MacKinlay (1997) the two most common non-parametric tests for event studies are the Sign Test and the Wilcoxon Signed Rank Test.

1. Sign Test

The Sign Test is based on the sign of the abnormal return. It requires that the abnormal returns (or more generally cumulative abnormal returns) are independent across securities and that the expected portion of positive abnormal returns under the null hypothesis is 0,5 (MacKinlay). In short, under the null hypothesis there exists an equally great chance that the CAR will be positive or negative.

MacKinlay uses the following method to calculate the sign test. First the number of cases where the abnormal return is positive, N+, and the total number of cases, N. Letting θ2 be the test statistic. The Sign Test is specified as follows:

) 1 , 0 ( ~ 5 , 0 5 , 0 N N N N       − = +

θ

(4)

This distributional result is asymptotic. For a test of size (1-α), Ho is rejected if

θ

>Φ−1(

α

).

2. Wilcoxon Signed Rank Test

A weakness of the Sign Test is that it may not be well specified if the distribution of abnormal returns is skewed as can be the case with daily data (MacKinlay). For this problem another non-parametric test for abnormal performance exists. The Wilcoxon Signed Rank Test assumes no abnormal return for event day zero.

The abnormal returns are ranked from one to L2. Kiτ is the rank of the abnormal return of security i for event time period τ. The range of τ goes from T1 + 1 to T2 (T1 = day -20 and T2 = day +20) and τ = 0 is the event day. The Wilcoxon Signed Rank Test takes advantage of the fact that the expected rank of the event day is (L2 + 1)/2 under the null hypothesis. The test statistic for the null hypothesis of no abnormal return on event day zero is (MacKinlay):

(25)

where: 2 1 1 2 2 2 1 2 1 1 1 ) (

∑

+ = =            + − = T T N t i L K N L K s τ τ

The test of the null hypothesis can be implemented using the result that the asymptotic null distribution of θ is standard normal.

These non-parametric tests are normally not used on a stand alone basis but in combination with the parametric tests. The non-parametric tests can function as a test of robustness of the results found with the parametric tests.

2.2.6. Testing for Normality

After the abnormal returns are calculated, the error terms of the three different models should be tested for non-normality by using the BJ test. The BJ test uses the property of a normally distributed random variable that the entire distribution is characterized by the first two moments, the mean and the variance. The standardized third and fourth moments of a distribution are known as its skewness and kurtosis. A normal distribution is not skewed and is defined to have a coefficient of kurtosis of 3. A normal distribution is symmetric and said to be mesokurtic. A normal distribution is said to be symmetric around its mean, whereas a skewed distribution will not be like that and have one tail longer than the other (Brooks; 2002). The test for normality will not be conducted in this thesis. The reasons for not testing the dataset for normality are that the literature discussed by other researchers does not discuss any form of testing for normality or the use of the Sign Test and Wilcoxon Signed Rank Test. Secondly, due to quantity restrictions the normality test is not performed. And finally, one can assume that the outcomes of the Sign Test and Wilcoxon Signed Rank Test will be in line with the results from the Mean Market Model and the OLS Market Model. Therefore only the parametric models will be used in this thesis.

2.2.7. The T-test

When the parametric models are used, a t-test will be preformed on the models. Here one continues with the abnormal returns of the models from the event window and estimation window using equation 1, 2, and 3.

(26)

______________________________________________________________________________

∑

= = t N i i,t t t A N A 1 1 (6)

Where, N is the number of events which will be studied. An average abnormal return is calculated for the event window (-20,+20) and the estimation window (-80,-21).

To estimate the t-value one has to divide the average abnormal return per day for all the events on ‘day 0’ by the standard deviation (

( )

_A

t S ∧

) of the abnormal return from the estimation window. The formula for calculating the t-value is:

( )

t t/S A

A

∧ (7) where:

( )

(

)

_            − =

∑

=− − = ∧ 59 21 80 2 / A S t t t t

A

and where:

∑

− = − = = 21 80 60 1 t t t A A

With the help of these formulas the t-value per model per country can be estimated. After that the t-values are tested to look at which level they are significant.

Based on the t-test one can see if the null hypothesis can be accepted or rejected.

2.2.8. The Size Effect

The size effect of the divestments is researched by using the models discussed above. Only now the country data are divided based on deal value. The “High Value Group” and the “Low Value Group” will be tested. The same procedure will be followed as above mentioned in the previous sections. Here any different effects between the two groups will be investigated. As already mentioned in the previous section, the articles from other researchers state that a higher deal value will result in a higher positive abnormal return.

2.2.9. The Financial Distress Theory

(27)

M is a linear combination of all securities, it follows that E(Rit)=E(Rm,t)=Kt ~

, ~

, for any security i. The ex post abnormal return on any security i is given by the difference between its return and that on the market portfolio. The Market Adjusted Returns model is also consistent with the Asset Pricing model if all securities have systematic risk of unity (Brown and Warner; 1980).

The Market Return Model is specified as follows:

m,t i,t

i,t R R

A = − (8)

where Rm,t is the return of the equally weighted all share total return index on day t.

During this test the time line will be changed according to the time line presented in Figure 2.

(28)

______________________________________________________________________________

3. Empirical Results

In this section the empirical results of the tests done, will be presented. First an explanation will be given of the testing for significance. Secondly, the results from the main research question will be presented. Thereafter the results of the Size Effect will be explained. The results of both these tests are performed with the help of The Mean Return Model and The OLS Market Model. The CAR results will also be presented.

Finally, the results for the Financial Distress Theory will be presented with the help of the Market Return Model.

3.1. Testing for Two-Sided Statistical Significance

A two-sided test is a statistical hypothesis test in which the values for which one can reject the null hypothesis, H0, are located in both tails of the probability distribution.

In other words, the critical region for a two-sided test is the set of values less than a first critical value (t*) of the test and the set of values greater than a second critical value (t*) of the test.

A two-sided test is also referred to as a two-tailed test of significance.

The choice between a one-sided test and a two-sided test is determined by the purpose of the investigation or prior reasons for using a one-sided test. Because in this thesis one looks at either no effect or an effect one has to test two-sided (e.g. H0 = 0; H1 ≠ 0). Figure 3 gives a presentation of a normal distribution and its critical tail regions.

Figure 3: Normal Distribution Graph

(29)

The null hypothesis will be rejected when the estimated t-value is located in the blue colored areas of the graph. All tests will be preformed at the 1 percent and 5 percent significance levels. For the tests, the tables with the critical t-values (t*) will be presented in the next sections.

In the next sections the various results will be presented of the tests done. For every test done the results for the entire event period (day -20,+20) are found in the corresponding appendices, as described in the text. Furthermore, in the main text an overview will be given in each section of the results of the returns and estimated t-values on ‘day 0’ for every country individually per model. Finally, the CAR results are also given for the different timeframes. For the calculation of the CAR values and CAR t-values information is found in sections 2.2.4. and 2.2.7. The results of these CAR calculations in the tables are not directly obtainable from the event period overviews in the appendices (except for the timeframe -20,+20, which is the same as that for the whole event period). The average abnormal return (AAR) values look at the significance at a certain specific day (e.g. day -5 or day 0) and the CAR values determine if a certain specific period of time has been significant (e.g. -10,+10 or -1, +1). With a CAR calculation one can create a multi period event window.

3.2. The Main Research Question

The first hypotheses to be tested are the ones computed from the main research question:

H0 : There is no relationship present between the divestment activities of a acquiring firm and its value in the UK and in the US.

H1 : There is a relationship present between the divestment activities of a acquiring firm and its value in the UK and in the US.

H0 : There is no relationship present between the divestment activities of a divesting firm and its value in the UK and in the US.

H1 : There is a relationship present between the divestment activities of a divesting firm and its value in the UK and in the US.

(30)

______________________________________________________________________________ Table III: Critical Values

df = 50

p 0,1 0,05 0,025 0,01 0,005 0,001 0,0005

t* 1,299 1,676 2,009 2,403 2,678 3,261 3,496

Confidence Level 80% 90% 95% 98% 99% 99,8% 99,9%

Source: Moore and McCabe (2006)

3.2.1. The Mean Return Model United Kingdom:

Appendix 6 gives an overview of the results of the t-test for the UK buyer and UK seller in the event period using The Mean Return Model. Table IV shows the results on ‘day 0’. Here one can see that the AAR t-value for the UK buyer and UK seller are 4,505492 and 4,388007 respectively. The t-test is done two-sided. The critical value of 1 percent is divided by two because 0,5 percent is located in both tails. The 5 percent level is divided by two as well because 2,5 percent is located in every tail region. The critical t-value for a confidence level of 95% is 2,009. The t-value corresponding to a confidence level of 99% is 2,678. The AAR t-value of the UK buyer is 4,505492. This t-value is higher than the critical t-value of 2,678. The AAR t-value of the UK buyer is positively significant at the 99% percent confidence level. The null hypothesis can be rejected on the 1 percent significance level and therefore at the 5 percent significance level as well. This assumes that there probably is a relationship present between the divestment activities of a acquiring firm and its value in the UK. For the UK seller a AAR t-value of 4,388007 is estimated. For this value the same reasoning is applicable as for the UK buyer just described. The t-value 4,388007 lies outside the 99% confidence interval (t* = 2,678) as well. The t-value is positively significant on a 99% confidence level. The null hypothesis for the UK seller can be rejected on the 1 percent and 5 percent significance levels, which indicates that there probably is a relationship present between the divestment activities of a divesting firm and its value in the UK.

Table IV: AAR Results of ‘Day 0’ of The Mean Return Model for the UK

‘Day 0’ AAR t-value AAR

UK Buyer 1,31% 4,505492**

UK Seller 1,61% 4,388007**

(31)

The CAR values of both the UK buyer and UK seller are presented in Table V. For the UK buyer there are no significant results found in the different CAR timeframes. For the UK seller, only the time windows that are close around ‘day 0’ have statistically significant values. For the UK seller the timeframes -1,0 and 0,+1 are the timeframes in which the t-values of the CAR are significant at the 5 percent significance level. In these timeframes the null hypothesis for the UK seller can be rejected at a 95% confidence level. These results support the results found in the previous section on the AAR t-value results for the UK seller.

Table V: CAR Values of The Mean Return Model for the UK

UK Buyer UK Seller

Interval CAR t-value CAR CAR t-value CAR

-20,+20 0,39% 0,032658 0,33% 0,021811 -10,+10 0,80% 0,132101 0,73% 0,094030 -5,+5 2,00% 0,627450 0,89% 0,221260 -3,+3 1,14% 0,562795 2,83% 1,099431 -1,0 0,87% 1,507810 1,73% 2,346733* -1,+1 0,51% 0,582948 1,97% 1,783755 0,+1 0,94% 1,619358 1,86% 2,522902*

** and * indicate statistical significance at the 1 percent and 5 percent levels

United States

Appendix 7 gives an overview of the results of the t-test for the US buyer and US seller in the event period using The Mean Return Model. Table VI shows the returns for ‘day 0’ for the US buyer and US seller are 1,523746 and 5,806041 respectively. At the 95% confidence level the critical t-value is 2,009. For the US buyer the AAR t-value is not significant at the 95% confidence level. The null hypothesis can not be rejected so that there probably is no relationship present between the divestment activities of a acquiring firm and its value in the US. For the US seller the ‘day 0’ t-value is 5,806041, which is higher than the critical t-value of the 99% confidence level. Therefore the conclusion can be drawn that the null hypothesis can be rejected. So that there probably is a relationship present between the divestment activities of a divesting firm and its value in the US.

Table VI: AAR Results of ‘Day 0’ of The Mean Return Model for the US

US Buyer 0,76% 1,523746

US Seller 2,57% 5,806041**

(32)

______________________________________________________________________________ The CAR values and corresponding CAR t-values are presented in Table VII. For the US buyer there are no significant results present. For the US seller the timeframes -1,0; -1,+1; and 0,+1 are all positively significant at the 1 percent significance level. Here the CAR results support the outcomes of the AAR t-value of the US seller.

Table VII: CAR Values of The Mean Return Model for the US

US Buyer US Seller

-20,+20 0,21% -0,041594 5,14% 0,283620 -10,+10 -1,06% -0,101476 6,12% 0,658714 -5,+5 -1,10% -0,199742 5,43% 1,115863 -3,+3 -1,01% -0,288469 4,98% 1,610532 -1,0 0,63% 0,633273 3,06% 3,456289** -1,+1 0,12% 0,079057 4,31% 3,249686** 0,+1 0,25% 0,247186 3,82% 4,321261**

3.2.2. The OLS Market Model

In this section the results of the OLS Market Model are discussed for the UK and US buyer and UK and US seller. The descriptive statistics of the test are found in Appendix 5. Appendix 8 and 9 give an overview of the results of the t-test for the UK and US buyer and the UK and US seller in the event period using the OLS Market Model.

United Kingdom

(33)

Table VIII: AAR Results of ‘Day 0’ of OLS Market Model for the UK

UK Buyer 1,22% 4,488086**

UK Seller 1,64% 4,877872**

The CAR values and CAR t-values are given in Table IX. Only for the UK seller in the timeframe of 0,+1 there is a positive significant result on the 95% confidence level. This positive significant effect is in line with the findings for the UK seller on ‘day 0’.

Table IX: CAR Values of The OLS Market Model for the UK

UK Buyer UK Seller

-20,+20 0,47% 0,041737 0,24% 0,017251 -10,+10 1,01% 0,176042 0,85% 0,120694 -5,+5 1,76% 0,588168 0,71% 0,191565 -3,+3 0,88% 0,462615 2,67% 1,137191 -1,0 0,77% 1,405112 1,70% 2,531586 -1,+1 0,49% 0,602394 1,84% 1,827085 0,+1 0,95% 1,742521 1,78% 2,647978*

United States

For the US the OLS Market Model is used as well to estimate the AAR t-values. In Table X the results of the OLS Market Models tests are presented. The AAR t-value of the US Buyer lies within the 95% confidence level (critical t-value: 2,009) so one cannot reject the null hypothesis for the US buyer. The US seller has a AAR t-value of 6,401255. This value is higher than the critical t-value at the 99% confidence level (t* = 2,678). The null hypothesis in this case can be rejected at the 99% confidence level. Here one can assume that there probably is a significant positive relationship present between the divestment activities of a divesting firm and its value in the US.

Table X: AAR Results of ‘Day 0’ of OLS Market Model for the US

US Buyer 0,80% 1,735228

US Seller 2,51% 6,401255**

(34)

______________________________________________________________________________ seller the timeframes close to ‘day 0’ -1,0; -1,+1; and 0,+1 have a significant positive t-value at the 99% confidence level. This assumes, in line with the ‘day 0’ AAR t-value, that there probably is a positive relationship present between the sellers divesting activities and the value of the firm in the US.

Table XI: CAR Values of The OLS Market Model for the US

US Buyer US Seller

-20,+20 -0,98% -0,052292 5,24% 0,325363 -10,+10 -1,09% -0,112914 6,10% 0,740110 -5,+5 -1,05% -0,207322 5,34% 1,237775 -3,+3 -0,83% -0,259178 4,82% 1,755225 -1,0 0,58% 0,634938 2,77% 3,524862** -1,+1 0,09% 0,068584 3,89% 3,307631** 0,+1 0,31% 0,335552 3,64% 4,637213**

3.3. The Size Effect

Appendix 10 to 13 give an overview of the descriptive statistics of the AAR of the two models implemented. For these tests the following hypotheses are used:

H0 : There is no relationship present between the size of the divestment activities of a acquiring firm and its value in the UK and in the US.

H1 : There is a relationship present between the size of the divestment activities of a acquiring firm and its value in the UK and in the US.

H0 : There is no relationship present between the size of the divestment activities of a divesting firm and its value in the UK and in the US.

H1 : There is a relationship present between the size of the divestment activities of a divesting firm and its value in the UK and in the US.

(35)

Table XII: Critical Values df = 25

p 0,1 0,05 0,025 0,01 0,005 0,001 0,0005

t* 1,316 1,708 2,060 2,485 2,787 3,450 3,725

Confidence Level 80% 90% 95% 98% 99% 99,80% 99,90%

Source: Moore and McCabe (2006)

3.3.1. The Mean Return Model United Kingdom

Table XIII gives an overview of the ‘day 0’ AAR and the AAR t-values for all four subgroups. The “High Value Group” has AAR t-values of 4,686478 and 0,798567 for the UK buyer and UK seller respectively. The AAR t-value of the UK buyer is significant at the critical 1 percent significance level. It is positively significant at the 99% confidence level. Here one can reject the null hypothesis for the UK buyer and assume that the size of a divestment probably does have influence on the value of the firm. For the UK seller the AAR t-value is not significant at the 95% confidence level. Here one cannot reject the null hypothesis for the UK seller.

The “Low Value Group” has AAR t-values of 2,432450 and 5,826033 for the UK buyer and UK seller respectively. For the UK buyer the AAR t-value is positively significant at the 95% confidence level. The critical value of 2,060 is lower than the AAR t-value. So for the UK buyer one can reject the null hypothesis and assume that there probably is a relationship present between the size of the divestment activities and the value of the firm.

For the UK seller the AAR t-value is positively significant at the 99% confidence level. The null hypothesis for the UK seller can be rejected and therefore the size of the divestment activities is probably affects the value of the firm.

Table XIII: AAR Results of ‘Day 0’ of The Mean Return Model for the UK

UK Buyer “High Value Group” 1,37% 4,686478**

UK Buyer “Low Value Group” 1,24% 2,432450*

UK Seller “High Value Group” 0,36% 0,798567

UK Seller “Low Value Group” 2,86% 5,826033**

(36)

______________________________________________________________________________ confidence level. Two values have a CAR t-value that is higher than the critical value at the 99% confidence level and one has a CAR t-value higher than the critical values at the 95% confidence level.

Table XIV: CAR Values of The Mean Return Model for the UK

UK Buyer High UK Buyer Low UK Seller High UK Seller Low

Interval CAR t-value CAR CAR t-value CAR CAR t-value CAR CAR t-value CAR -20,+20 2,47% 0,20566591 -1,69% -0,080894 0,14% 0,00745931 0,52% 0,025741 -10,+10 1,89% 0,307096778 -0,28% -0,026053 -1,90% -0,199588033 3,35% 0,324690 -5,+5 2,52% 0,781718796 1,49% 0,264689 -0,84% -0,167803932 2,63% 0,485657 -3,+3 2,43% 1,186185059 -0,15% -0,040905 2,40% 0,756988708 3,26% 0,946149 -1,0 1,07% 1,823916413 0,68% 0,667408 0,21% 0,227613475 3,25% 3,299748** -1,+1 0,74% 0,837936099 0,28% 0,181842 -0,13% -0,094979039 4,06% 2,755285* 0,+1 1,04% 1,776226704 0,84% 0,821579 0,03% 0,029201678 3,68% 3,746196**

United States

In the US, first the results of the “High Value Group” are investigated. The “High Value Group” US buyer and US seller have AAR t-values on ‘day 0’ of 2,505980 and 7,708463 respectively (Table XV). Both AAR t-values are significant at the 95% confidence level. The AAR t-value for the US seller is even positively significant for the 99% confidence level. The US buyer has a AAR t-value that is significant at the 5 percent significance level and the US seller has a AAR t-value that is significant at the 1 percent significance level. Here one can reject the null hypothesis for both the US buyer as well as for the US seller.

The “Low Value Group” has AAR t-values of 0,471083 and 1,309794 for the US buyer and US seller respectively. Both values are not significant at the critical t-values of the 5 percent and 1 percent significance levels. In this case the null hypothesis for both the US buyer as well as the US seller cannot be rejected. This means that there probably is no relationship between the value of the firm and the size of the divestments for either the buyer or seller in the “Low Value Group” in the US.

Table XV: AAR Results of ‘Day 0’ of The Mean Return Model for the US

US Buyer “High Value Group” 1,11% 2,505980*

US Buyer “Low Value Group” 0,41% 0,471083

US Seller “High Value Group” 4,32% 7,708463**

US Seller “Low Value Group” 0,81% 1,309794

(37)

For the US the CAR and CAR t-values only have four significant outcomes in Table XVI. All of the outcomes lie around ‘day 0’ and all significant results come from the US seller. The “High Value Group” has three significant outcomes at the 99% confidence level. The timeframes for these outcomes are -1,0; -1,+1; and 0,+1. All three outcomes are significant at the 1 percent significance level. In the “Low Value Group” only one timeframe has a significant critical value. Timeframe 0,+1 has a CAR t-value that is positively significant at the 5 percent significance level.

Table XVI: CAR Values of The Mean Return Model for the US

3.3.2. The OLS Market Model United Kingdom

An overview of the AAR results with their corresponding AAR t-values is given in Table XVII. The “High Value Group” has a positive significant result for the UK buyer at the 99% confidence level. The null hypothesis for the UK buyer can be rejected. The UK seller does not have a significant t-value on the 1 or 5 percent significance level. Therefore the null hypothesis for the UK seller cannot be rejected.

In the “Low Value Group” both AAR t-values are significant. The UK buyer has a AAR t-value that is significant at the 5 percent significance level. The null hypothesis can be rejected on a 95% confidence level. The UK seller has a AAR t-value that is significant at the 1 percent significance level. The null hypothesis in this case can be rejected on a 99% confidence level.

Table XVII: AAR Results of ‘Day 0’ of The OLS Market Model for the UK

UK Buyer “High Value Group” 1,26% 4,941713**

UK Buyer “Low Value Group” 1,19% 2,382399*

UK Seller “High Value Group” 0,37% 0,916851

UK Seller “Low Value Group” 2,90% 6,021999**

US Buyer High US Buyer Low US Seller High US Seller Low

The Effect of Corporate Sell Offs on Shareholder Wealth