Rijks Universiteit Groningen
Faculty of Economics and Business
Master of Science: Business Administration, Specialization Finance Student: Frank Schepel
Student number: 1612905
Title: Value creation through mergers and acquisitions of high-tech firms Month and Year: August 2010
Supervisor: dr. ing. N. Brunia Words: 7589
Abstract
Shareholders of firms announcing a merger or acquisition of a high-tech company gained on average an abnormal return of 0.6% over the 3 days surrounding the announcement during period January 1, 1997 until December 31, 2009. However, there are regional differences in the average abnormal returns. In North America the abnormal return is almost 0% while this is for Europe and the rest of the world 1%. In a multivariate model different variables are used to try to explain the abnormal returns. Only size of the acquirer and to a lesser extend deals being cross border can explain abnormal returns. Across the three regions there are some differences in variables that can explain abnormal returns, for North America the cross border, share payment and size variables are of importance. While for Europe and the Rest of the World the period and size variable can explain part of the abnormal returns.
2
1. Introduction
In this research the value creation potential of a merger or acquisition of a high-tech firm is investigated. The main question in this research is, do mergers and acquisitions of high-tech companies create value for the shareholders of the acquiring company? Value creation is measured as the cumulative abnormal return over the three days surrounding the event. Most of the high-tech targets that are acquired are small private firms. According to Carpenter and Petersen (2002) small high-tech companies have difficulties acquiring external financing. An acquisition can make the needed capital available for the target. According to Kohers and Kohers (2000), the high growth nature of high-tech gives them the ability to create value for the acquirer. Ranft and Lord (2000) and Frick and Torres (2002) find that high-tech companies not always have the ability to develop products or special techniques internally because it can be too costly or takes too long, a merger or acquisitions appears to be a good alternative in that situation. According to Frick and Torres mergers and acquisitions can also be a good way for filling up a hole in the product line of a high-tech company. Not all acquirers are high-tech companies themselves, but due to the reliance on technology of almost every company, the acquisition of a high-tech company can create value by itself according to Kohers and Kohers (2000). If the mergers and acquisitions of high-tech firms indeed turn out to create value for the acquirer, than the result is of value for both professional and private investors as well as for managers of a company that want to acquire a high-tech company but at the same time want to maintain or increase shareholder value.
Some research has already been done on the announcement returns of acquirers of a high-tech firm.
Kohers and Kohers (2000) investigate announcement returns of acquirers that acquire a high-tech company over the period 1987 until 1996 and find positive abnormal returns upon announcement.
Benou and Madura (2005) also investigate announcement returns for acquirers of high-tech companies but they focus more on the influence of the use investment banks. Even though Kohers and Kohers use a sample with acquirers from multiple countries, they do not look at differences in abnormal returns or deal and company characteristics of acquirers from different countries. In this research differences in abnormal returns of acquirers from different regions will be investigated, as well as differences in deal and company characteristics.
In order to test the hypothesis a list of events is created with the help of the Zephyr database of Bureau van Dijk Publishing. The list consists of 3035 mergers and acquisitions and these are from the period January 1, 1997 until December 31, 2009. This sample consists of acquirers from a lot of different countries, however, the largest part of the deals are from the United States and Europe, together forming about 80 percent of the entire sample.
3 The main results of this study are that companies that announce a merger or acquisitions with a high- tech company experience on average a significant positive abnormal return of 0.6%. The returns are calculated as in MacKinlay (1997) and Brown & Warner (1985). Between the subsamples North America, Europe and the Rest of the World there are some significant differences in abnormal returns. For North America the abnormal returns are almost zero, while for Europe and the Rest of the World the average abnormal return is 1% over the event window. In a multivariate model is tested which variables can explain the abnormal returns. Similar to Moeller et al. (2005), the size of the acquiring company has a negative effect on the abnormal returns following the announcement, larger acquirers earn lower abnormal returns. Mergers and acquisitions that are cross border also earn significant lower abnormal returns. Also there are differences in the variables that influence the abnormal returns for different regions. The size variable is of negative influence in every region, the share payment and cross border variables are of negative influence in North America. The period variable has a positive coefficient in Europe while a negative coefficient in the Rest of the World subsample.
2. Literature review
In a simple model, Koller et al. (2005) state that value creation as a result of a merger or acquisition arises when the target is worth more to the acquirer than the price that the acquirer has to pay for the target. In a perfect market, value creation would never happen since everyone has the same information as the acquirer has and as such the target shareholders would require a premium that is equal to the extra value that the target is worth to the acquirer. However, price reactions do occur and there is a large amount of literature that tries to explain why price reactions exist. Roll (1986) states that acquiring firms may overpay for a target for because the manager suffers from hubris.
Roll (1986) argues that this overpayment can happen because the manager of the acquiring firm has calculated that the target is more worth to the acquirer than its market value is, even though this not the case. However the manager is confident that his calculation is correct and as such overpays for the acquisition. The argument that Roll (1986) gives for this is that the average manager only gets the opportunity to do a few acquisitions in his career and as such is not able to learn from his past mistakes. It is not likely that a manager of company that acquires a high-tech company suffers more from hubris than a manager whose company takes over a non high-tech company. Another reason why bad acquisitions are undertaken is constructed by Jensen (1986). He states that empire building is a reason why management engages in value decreasing mergers and acquisitions. But also for this argument it is unlikely that managers of firms that acquire high-tech firms are more prone to empire building than managers of firms that acquire non high-tech firms.
4 Mergers and acquisitions are a common practice, also mergers and acquisitions of high-tech companies and if they would be value decreasing by definition they would not be undertaken so often. Therefore it is also possible that mergers and acquisitions are capable of creating value.
According to Carpenter and Petersen (2002) small high-tech companies have difficulties acquiring external financing. Their reasoning is that the returns of R&D projects of a high-tech company are uncertain and skewed. Secondly there exist information asymmetries between the management of the firm and the suppliers of capital. Third, high-tech firms have often little collateral to offer. An acquisition can give the access to capital and enable the development or growth of products or services that otherwise would not be possible for the target company. According to Kohers and Kohers (2000), high-tech targets possess growth opportunities that make them suitable for creating value. Frick and Torres (2002) describe the market in which the high-tech companies operate as being fast paced and turbulent. Innovations replace each other rather quick and it is hard for companies to keep up with the pace of the market. As such acquisitions can be a good way to fill a hole in a company’s product line or to keep up with the market. However, not every company that acquires a high-tech company is a high-tech company. Kohers and Kohers (2000) argue that due to the increasing reliance of companies on technology, the acquisition of a high-tech company can create value by itself.
In order to check what type of deals and what type of acquirers create value, deal and company characteristics will be regressed against cumulative abnormal returns. A lot of research has been done on deal and company specific characteristics that influence abnormal returns following an announcement. Table 1 summarizes the deal and company characteristics that will be used
Table 1: Deal and company characteristics, their effect on abnormal returns and their expected effect on abnormal returns of companies that acquire high-tech firm
Deal characteristic Effect found Expected effect
Cash payment + 0
Cross border deal - -
Leverage + +
Market to book ratio - 0
Period + +
Public target - -
Same industry + +
Share Payment - 0
Size - -
5 throughout this research. Also it shows what effect other authors find that the variable has on abnormal returns. Next to that, the third column shows what the expected effect of a variable is on the announcement of a merger or acquisition of a high-tech company.
The method of payment is an often used variable to explain abnormal returns. Travlos (1987) finds that firms that pay an acquisition with shares experience lower abnormal returns than firms that pay an acquisition with cash. The reasons he gives for this effect is signaling of overpriced equity and co- insurance effects. Co-insurance effects arise when the two companies have not perfectly positively correlated cash flows. As a result the debt capacity of the new entity will increase and without capital restructuring, part of the gains of this extra debt capacity will accrue to the bond holders. Cash offers are expected to have mostly positive effects due to tax benefits. However, in a more recent study Chang et al. (1998) find no effect of the method of payment on abnormal returns. Benou and Madura (2005) find that the method of payment does influence the abnormal returns. According to them, acquisitions paid with cash experience higher abnormal returns. However, Kohers and Kohers (2000) find that the method of payment has no effect on the abnormal returns. According to them this is because the reaction is not that the choice to pay with stock automatically implies overvaluation but that the shareholders also see the growth potential of the high-tech target.
Bris and Cabolis (2008) investigate firm value and shareholder protection. They investigate whether the value of a firm changes when it is subject to a change of legal system. This occurs for instance when the firm is acquired by a company from another country. From that point on the acquired target is subjected to legal rules of the country where the acquirer is located. If this country has a legal system that is seen as having better investor protection, the premium paid for the acquirer is higher, and as a result the abnormal returns are lower. Also other factors like cultural differences and integrating the two companies are of influence on the abnormal returns and are all captured by the cross border variable.
Maloney et al. (1993) find that firms with higher leverage have higher average abnormal returns.
They state that this is due to the fact that leverage reduces the free cash flows, forcing management to make better acquisitions, a theory that is popularized by Jensen (1986). Dong et al. (2003) investigate overvaluation and the returns following an announcement of a merger or acquisition. The market to book ratio is used as a proxy for overvaluation. They state that companies with a high market to book ratio that announce a merger or acquisition signal to the market that their shares are overvalued and that they try to acquire less overvalued assets in order to lock in on value. However this is a bit ambiguous since a high market to book ratio can also mean high performance of the
6 acquirer. Moeller et al. (2005) find that firms that perform well over the year before the acquisition, have higher abnormal returns upon announcement.
The period in which the acquisition is announced can be very important. After the burst of the high- tech bubble in March 2000 (Greenwood and Nagel, 2009) the share price of high-tech companies dropped rather fast. While before the burst of the bubble the share price of high-tech companies was higher than ever before. Benou and Madura (2005) find that acquisitions that took place before the burst of the high-tech bubble received higher abnormal returns than acquisitions that took place after the high-tech bubble. According to Chang (1998) and Fuller et al. (2002) abnormal returns are lower when the target is a publicly listed company. Their argument for this is that private targets are not as liquid as public targets are. They cannot be bought as easily as public target can be bought.
This leads to a discount in the price the acquiring firm has to pay for a private target and as such the effect is positive. Both Kohers and Kohers (2000) and Benou and Ma0dura (2005) find this effect for the acquisition of high-tech targets to. According to Carpenter and Petersen (2002), many small high- tech firms have no or only limited access to outside financing. As soon as these companies go public through an IPO they experience a sharp rise in firm size. An acquisition can also give this access to capital and as such have a positive effect opposed to the acquisition of a public target.
Kennedy et al. (2002) find that acquisitions in which the target is active in the same industry as the acquirer have higher abnormal returns. Their explanations are that the acquisition will be friendlier than when the target is active in a different industry. Their expectation is that management of a company that is active in a related company has a better chance of a good future in the new company due to basic industry knowledge and as such will view the acquisition as more favorable.
According to them immediate perceived benefit is larger with related acquisitions than with diversifying acquisitions. They state that the higher benefit is caused by the fact that related product diversification is easier to achieve for the acquirer than unrelated product diversification. Unrelated diversification requires more time and resources to implement than related diversification.
Moeller et al. (2005) find that on average the abnormal return for the acquiring company is 1.1%.
However, upon announcement the shareholders of an acquiring company lose $25.2 million on average. This loss indicates a size effect at which the large companies lose value after the announcement and smaller companies gain value after the announcement of a merger or acquisition. Their explanation is that shareholders of the large firms reconsider their valuation. Also they state that the possibility exists that because the large firms where mostly serial acquirers, the acquisitions show their investors that the strategy of growth through acquisitions is not sustainable.
7 The expectation is that this is no different for firms that acquire high-tech firms and that smaller acquirers will have higher returns than larger acquirers.
3. Method
Following the arguments of Carpenter and Petersen (2002), Kohers and Kohers (2000) and Frick and Torres (2002) there is a solid basis to believe that the acquisition of a high-tech company will on average be value creating. As such the hypothesis is that the announcement of a merger or acquisition of a high-tech company will result in positive abnormal returns. To test the hypothesis daily prices are used, these prices are adjusted for dividends and stock splits to make sure that returns are not distorted by price reactions as a result of dividends and stock splits. Before calculating any returns to test the hypotheses, all the prices of the shares of the acquiring companies are converted to dollars. Prices are converted to a single currency to control for differences in monetary policy between different countries and monetary unions. Thorbecke (1997) states that stock returns react to monetary policy, he finds that stock returns react positively to expansionary monetary policy and negatively to contractionary monetary policy. Expansionary monetary policy leads to higher inflation and, all other things being equal, as a result to a devaluation of the currency opposed to foreign currency. By converting all prices to a single currency, price reactions as a result of differences in monetary policy are not measured. The time period of 1 day before the event until 1 day after the event is defined as the event window. Returns used for testing are calculated as daily logarithmic returns. In order to calculate whether the returns in the event windows are different from the returns that would be expected if the event would not have taken place, abnormal returns are calculated. For the calculation of the abnormal returns three methods are used, these methods are also used by MacKinlay (1997) and Brown and Warner (1980, 1985). The first method is the average abnormal returns method, the second market adjusted abnormal returns and the third market and risk adjusted abnormal returns. The abnormal returns 𝐴𝑅𝑖,𝑡 of stock 𝑖 at time 𝑡 are calculated with the average abnormal returns method through
𝐴𝑅𝑖,𝑡= 𝑅𝑖,𝑡− 𝑅�𝑖
where 𝑅𝑖,𝑡 denotes the return of stock 𝑖 at time 𝑡 and 𝑅�𝑖 the average return of stock 𝑖 determined over the period of 200 days prior to the event. The market adjusted abnormal returns are defined as follows
𝐴𝑅𝑖,𝑡= 𝑅𝑖,𝑡− 𝑅𝑚,𝑡
8 where 𝑅𝑚,𝑡 denotes the return on the MSCI world index at time 𝑡. The MSCI world index is chosen as the benchmark market to compare the returns of the acquirer with. In this research a worldwide sample is used and the MSCI world index consists of most of the markets that the acquirers in the sample are from. For instance the MSCI All Country World Index would add a large amount of indexes from countries from which there are no acquirers in the sample. Finally the market and risk adjusted abnormal returns are calculated. With this method the returns of the shares are regressed with the market by
𝐴𝑅𝑖,𝑡 = 𝑅𝑖,𝑡− 𝛼�𝑖− 𝛽̂𝑖𝑅𝑚,𝑡
where 𝛼�𝑖 and 𝛽̂𝑖 denote the regression parameters estimated over the 200 days before the event.
This method captures the magnitude of correlation between the share price and market swings.
After that, cumulative abnormal returns will be taken over the period -1 till +1. The significance of the average cumulative abnormal returns will then be tested. Following the paper of MacKinlay (1997) the significance will be tested on three different ways in order to check the robustness of the abnormal returns. First the significance of the average cumulative abnormal returns will be checked using the historical standard deviation of the returns over the 200 days before the event. Secondly the significance of the average cumulative abnormal returns will be tested using the cross-sectional standard deviation of the cumulative abnormal returns. The cross-sectional variance in used to mitigate the problem that the variance changes during the event opposed to the historical variance.
By applying this method, the average abnormal return will not be significant because the historical variance is lower than the variance around the event. Finally the standardized average cumulative abnormal return is calculated. According to MacKinlay (1997) the standardization of abnormal returns increases the power of the test. The abnormal returns are standardized as in Brown and Warner (1985). Because the abnormal returns do not follow a normal distribution the significance of the abnormal returns is also tested with the Wilcoxon signed rank test. According to Brown and Warner (1985) parametric test are well specified even when the daily returns do not follow a normal distribution and as a result of this the use of a non-parametric test is not required. However, the Wilcoxon signed rank test is added as a robustness check in this case
The sample used in the research contains acquirers of high-tech companies from around the world.
Because the acquirers come from different areas around the world, it is possible to check for differences in abnormal returns between different regions. With an ANOVA test the equality of means will be tested between Europe, North America and the region defined as Rest of the World to see whether the price reaction upon announcement is the same for all three regions. In addition to
9 check the results for robustness, an equality of median test will be conducted. The Kruskal-Wallis test will be used, this test is an extension on the Wilcoxon signed rank test and allows two or more samples of different sizes to be tested on a difference in their median. Again the non parametric test is added as a robustness check.
After that a multivariate model is estimated to see whether deal characteristics can explain the abnormal returns. The model that is estimated is as follows.
𝐶𝐴𝑅𝑖 = 𝛼0+ �(𝛽𝑖,𝑗∙ 𝑋𝑖,𝑗) + 𝜀𝑖
The 𝑋𝑖,𝑗 denotes the variables that are used in the equation, these are the variables that are described in Table 1 as the company and deal characteristics that influence abnormal returns. The significance of the coefficients of the variables will be tested with the Student T test.
A common statistical issue when estimating a cross-sectional regression is heteroscedasticity in the error terms. As a result the regression estimators will still be unbiased and consistent, however, they do not have the minimum variance. As a result inferences made about their significance can be wrong. To overcome this problem White’s correction to standard errors for heteroscedasticity will be applied to make sure that proper inferences can be drawn about the significance of the coefficients in the presence of heteroscedasticity. If there is no heteroscedasticity in the error terms, the variance of the coefficients will be the same as under a normal OLS estimation, otherwise they will differ (White, 1980).
4. Data
In this research a worldwide sample of mergers and acquisitions is used. The list of mergers and acquisitions is generated with the help of the Zephyr database of Bureau van Dijk Publishing. The deals are selected as such that the targets are active in a few pre-specified industries based on their US SIC code. The industries in which targets have to be active to be selected are listed in Table 2. In order for a deal to be used in the sample, the method of payment has to be known, the acquired stake has to be at least 51%, the acquirer has to be publicly listed and the deal has to be announced and completed between January 1, 1997 and December 31, 2009. Initially this resulted in a list of 7305 deals. However Zephyr also exported deals in which the acquirer is from one of the defined industries, but the target is from a non high-tech industry, as such these observations could not be used. Due to this flaw in the export the data had to be filtered in excel and as such it is not clear how many observations were dropped after each criterion. After filtering the list contained 5771
10 observations. Additional variables that are obtained from Zephyr are the countries the acquirer and the target originate from for creating the cross border variable. Zephyr also states whether the target is publicly listed or not and in which industry the acquirer is active to check whether the acquirer and target are from the same industry or not. After that the stock prices and the remaining deal and company characteristics listed in Table 1 have to be available in the DataStream database of Thomson Reuters. For retrieving the stock price and other variables the acquirers ISIN code is used, this results in the final list of 3035 observations.
Table 2: The target industry US SIC Code, the number of targets in each industry and the description of the industry
In Figure 1 the distribution of the deals over time is shown. From this figure it is clear that the number of deals was high at the top of the high-tech bubble in 2000 and quickly dropped after the bubble burst. In the period 2005-2007 the number of deals is high due to a growing economy and in the 2008 and 2009 again lower due to the recent economic crisis. In the period 1997-1999 there are relatively few observations because Bureau van Dijk Publishing started to record mergers and acquisitions actively from 2000 and updated their database over the 3 previous years, when making conclusions this fact will be taken into account.
US SIC Code Number of Deals Description
283 701 Drugs
366 628 Communications equipment
372 475 Aircraft and parts
376 13 Guided missiles, space vehicles
and parts
381 82
Search, detection, navigation guidance aeronautical and nautical systems and instruments.
384 489 Surgical, medical and dental
instruments and supplies
481 647 Telephone communications
11 Figure 1: distribution of mergers and acquisitions over time
Table 3 shows the geographical distribution of the acquirers of high-tech companies, this table shows that most of the acquirers are from North America and Europe. About 80% of the entire sample comes from these two regions. When differences between regions in the sample are reviewed, North America and Europe will be treated as separate regions. The other regions will be grouped to a single group named: rest of the world because they are too small to be treated as separate groups.
Table 3: Distribution of acquirers over geographical areas
Region Number of observations Percentage
Africa 24 0.8
Asia 332 11.0
Australia and New Zealand 98 3.2
Europe 1125 37.1
North America 1389 45.8
South America 67 2.2
Total 3035 100
The unit of measurement and the description of the variables that are used in the multivariate model are listed in Table 4. The method of payment variables, cash payment and share payment, are both dummies that are only assigned the value of 1 when the merger or acquisition involves either a 100%
0 50 100 150 200 250 300 350 400
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Frequency
Year
12 Table 4: Deal or company characteristics and their unit of measurement
Deal or company characteristic Unit of measurement Description
Cash payment Dummy variable The dummy has the value 1 if a
cash only payment is involved and 0 otherwise.
Cross border Dummy variable
The dummy has a value 1 if the deal is between two companies from different countries or 0 otherwise.
Leverage Ratio The ratio is calculated as the
net debt divided by the market value of the shares.
Market to book ratio Ratio
The ratio is calculated as the market value of the shares divided by the book value of the shares.
Public target Dummy variable
Dummy variable that has a value of 1 if the target company is publicly listed at the time of the acquisition or 0 if otherwise.
Period Dummy variable
Dummy variable that has the value of 1 if the deal took place during the internet bubble of 1997 until March 2000 and 0 otherwise.
Same industry Dummy variable
Dummy variable that has a value of 1 if the acquirer is active in the same industry as the target.
Share payment Dummy variable Dummy variable that has a
value of 1 if the deal involves a share only payment.
Size Dummy variable
The Dummy has a value of 1 if the acquirer has a market value at the time of announcement higher than 1 billion dollar.
13 cash or share payment. The cross border variable is a dummy that is assigned the value of 1 when the acquirer and target are from different countries and 0 otherwise. Leverage is calculated as the ratio of net debt divided by the market value of the shares. Net debt is defined as debt minus cash and short term investments, similar to the definition used by DataStream. The market to book ratio is calculated as the acquirers market value of the shares divided by the book value of the shares. For the public target variable is a dummy used that has the value of 1 when the target is publicly listed at any stock exchange or 0 otherwise. The period variable is a dummy for deals that took place during the high-tech bubble. The high-tech bubble is defined as the period 1 January 1997 until 1 March 2000 (Greenwood and Nagel, 2009). The amount of acquisitions in the sample is low for April 2000, with only 24 acquisitions. The average amount of monthly acquisitions over the period 1 January 2000 31 December 2009 is 47. This gives support that the internet bubble indeed burst in March 2000 as indicated by Greenwood and Nagel (2009). The same industry variable is a dummy variable that has a value of 1 when the acquirer and target are active in de same US SIC three-digit industry.
Acquisitions that take place within the same three-digit US SIC industry are seen as related diversifying acquisitions whereas acquisitions between two companies from two different industries are seen as unrelated diversifying acquisitions. The size variable is a dummy that has the value of 1 when the acquirer has a market value of 1 billion dollar or larger, indicating large acquirers. When the acquirers are ranked based on their market value and divided into deciles there is a jump in abnormal returns between the sixth and fifth decile. The tenth decile until the sixth decile have average abnormal returns of 1% or larger, while the fifth until the first decile have abnormal returns close to 0% or slightly below 0%. The average market value in the sixth decile is 850 million dollar, while the average market value in the fifth decile is 1.5 billion dollar. The abnormal returns of acquirers until 1 billion dollar are however, more in line with the acquirers of the sixth decile and because of that the cut-off point for the market value of 1 billion dollar is chosen.
Table 5 shows the occurrences of the company and deal characteristics that are taken into account in this research. From this table it can be seen that the number of deals that involved a share payment are rather low, in roughly one sixth of the deals the acquisition or merger is paid entirely with shares.
Two-third of the mergers and acquisitions are paid with cash and this is the case for the entire sample as well as for the different regions. When these numbers are compared to for instance Moeller et al. (2005), they find that about 24 percent of the deals are paid with only shares and 40 percent of the deals are paid with cash. About half of the acquirers are qualified as being large, they have a total market value of more than one billion dollar. About one sixth of the deals took place during the high-tech bubble, which is defined as the period 1 January 1997 until 1 March 2000. This
14 number is rather low since Zephyr contains not so much mergers and acquisitions over the period 1997 until 2000. About halve of the deals took place between companies active in the same industry.
Table 5: occurrences of deal characteristics
Entire sample America Europe Rest of the world
Cash payments 2009 931 729 349
Cross border 1830 1039 423 368
Public target 936 411 331 194
Period 492 193 285 14
Same industry 1316 650 459 207
Share payments 565 245 212 108
Size 1611 849 572 190
Number of deals 3035 1389 1125 521
In Table 6 the descriptive statistics of the abnormal returns are shown, the returns of the shares of the acquiring companies around their event dates appear to be positive. With all three methods a positive average cumulative abnormal return is found. However the mean in about twice as large as the median and as such the returns are positively skewed. Also, the returns are highly leptokurtic as is the often the case with high frequency financial data. As a result the Jarque-Bera test rejects normality of the data at a 1% significance level. A large part of the kurtosis is caused by a few outliers, if these are removed from the sample the kurtosis drops to around 7 for all three abnormal calculation methods. However, the data remains not normally distributed. Also the significance of the abnormal returns and the variables in the multivariate model are not changed. Therefore the choice is made to keep the outliers in the sample and not to remove observations based on their abnormal return.
Table 6: Descriptive statistics of the cumulative abnormal returns, calculated under the three methods
with a Descriptive
being significant at a 1% significance level.
Average adjusted Market adjusted Market and risk adjusted
Mean 0.56% 0.71% 0.60%
Median 0.25% 0.46% 0.30%
Standard deviation 0.072 0.070 0.070
Skewness 0.16 0.24 0.29
Kurtosis 13.5 14.4 14.7
Jarque-Bera statistic 13953a 16562a 17489a
15
5. Results
Table 7 shows the results of the significance tests of the abnormal returns. From this it can be concluded that with all three abnormal return calculation methods the abnormal returns are significant. Also with the adjustments in the calculation method of the T-value, cumulative abnormal returns remain significant. Normalized abnormal returns enhance the significance of the test value, as found by MacKinlay (1997) and Brown & Warner (1985). Using the cross-sectional standard deviation to calculate the T-values lowers the significance due to the higher standard deviation compared to when the standard deviation is calculated using the historical returns. However the abnormal returns are still significant at the 1% significance level using the cross-sectional standard deviation. Also under the non-parametric test, the Wilcoxon signed rank test, the abnormal returns are significant. From the three models the Market adjusted abnormal returns are the most significant, however this model is quite crude and in the event window there are a large number of returns that move contrary to what the market does, and as such are the abnormal returns larger then under the two other models. However, an ANOVA test does not detect any significant differences in the returns calculated with each of the calculation methods. It is therefore concluded that each of the models detects abnormal returns equally well.
Table 7: Significance tests of the average cumulative abnormal returns calculated with the three abnormal return models
with a being significant at the 1% significance level.
Average adjusted Market adjusted Market and risk adjusted Mean cumulative
abnormal return 0.56% 0.71% 0.60%
Historical T-test value 4.79a 6.02a 5.12
Historical standard deviation
a
0.12% 0.12% 0.13%
Cross sectional T-test
value 4.29a 5.54a 4.67
Cross sectional standard deviation
a
0.13% 0.13% 0.13%
Standardized T-test
value 5.17a 6.70a 5.38
Wilcoxon Signed rank test
a
5.00a 6.49a 5.06a
16 It might be that there are differences in the effects for different regions due to law, culture or something else. For this reason an ANOVA analysis is done between North America, Europe and the rest of the world. The different deal characteristics where rather equal across the regions and as such it is assumed that differences in deal characteristics do not lead to differences in abnormal returns between the regions. Table 8 shows the results of the ANOVA analysis of the means of North America, Europe and the Rest of the World group.
Table 8: ANOVA analysis of the difference between North America, Europe and the Rest of the World
With a significant at the 1% significance level.
Cumulative abnormal returns
Average adjusted Market adjusted Market and risk adjusted
Mean North America -0.01% 0.20% 0.11%
Mean Europe 1.02% 1.13% 1.01%
Mean Rest of the World 1.10% 1.18% 1.01%
ANOVA F-test value 8.19a 6.86a 6.28
Kruskal-Wallis
a
12.60a 10.71a 9.65a
From this table it is clear that there are some differences between the three sub samples. The
ANOVA test as well as the Kruskal-Wallis test rejects the hypothesis that the means or medians of the groups are the same. As can already be seen, the mean of North America is much lower than the means of Europe and the Rest of the World. Testing the subsamples pair wise with an ANOVA test determines that the differences between the groups are caused indeed by the low mean of North America, also the non-parametric Mann-Whitney U test shows that the differences are caused by the North America group. There is no significant difference between Europe and the Rest of the World group.
Table 9 shows the results of the multivariate model. When the whole sample is taken into account, only the size variable can explain part of the abnormal returns at the 1% significance level. The sign of the size variable is in line with the expectations, larger acquirers have lower abnormal returns than smaller acquirers as is the case in Moeller et al. (2005). Large acquirers have on average a 1.8% lower abnormal return than small acquirers. The cross border variable is only significant at the 10%
significance level and has a negative sign. This is in line with Bris and Cabolis (2008) as they find that in countries with strong investor protection returns are lower because larger premiums are paid. On regional level there exist some differences. In all the sub samples the size variable is negative and
17 significant at 1%. The period variable is positive and significant at a 10% significance level for the Europe sub sample and negative for the Rest of the World sub sample. This contradicts the expectation that mergers and acquisitions received higher abnormal returns during the high-tech bubble.
Table 9: Coefficients of the multivariate model
With the T-test values between brackets and a, b and c significant at the 1%, 5% and 10% significance levels.
Entire Sample North America Europe Rest of the World
Constant 0.020
(4.56a 0.023
) (3.11a 0.010
) (1.62) 0.028
(2.20b) Cash payment -0.001
(-0.40) -0.002
(-0.32) 0.004
(0.91) -0.013
(-1.18) Cross border -0.004
(-1.69c -0.007
) (-1.79c -0.002
) (-0.38) 0.002
(0.25)
Leverage -0.007
(-0.99) 0.007
(0.02) 0.002
(1.08) -0.001
(-0.71) Market to book
ratio 0.003
(-0.26) 0.005
(-0.70) 0.003
(0.41) 0.004
(-0.56)
Period 0.003
(0.70) 0.001
(0.14) 0.009
(1.82c -0.041
) (-1.87c
Public target
) -0.004
(-1.31) -0.006
(-1.50) 0.001
(0.32) -0.008
(-1.07) Same industry 0.001
(0.56) 0.001
(0.26) 0.001
(0.21) 0.006
(0.85) Share payment -0.003
(-0.72) -0.015
(-1.98b 0.007
) (0.92) 0.001
(0.10)
Size -0.018
(-6.80a -0.016
) (-4.03a -0.013
) (-2.72a -0.022
) (-2.96a
R-squared
)
0.017 0.025 0.013 0.038
Number of
observations 3035 1389 1125 521
It can be explained because the burst of the bubble appeared to have happened earlier in Asian markets that in the North American and European markets. The Nikkei and the Straits Times indexes started to decline sharply in the beginning of 2000 or even late 1999. Apparently for the North American market the high-tech bubble did not have any influence on abnormal returns. For the entire sample the method of payment is not of influence on the abnormal returns. For North America, the share payment variable is negative and significant. This is more in line with the results
18 of Travlos (1987) and Benou and Madura (2005). However it is contrary to what Kohers and Kohers (2000) and find. It is possible that this is a regional effect since Travlos (1987) and Benou and Madura (2005) use data from the US and Kohers and Kohers use data from multiple countries. Another result that stands out is the non significance of the public target dummy. In many articles on mergers and acquisitions this variable has a significant negative coefficient. Also in the articles on high-tech mergers and acquisitions of Kohers and Kohers (2000) and Benou and Madura (2005) the public target variable has a significant negative coefficient. It is possible that the public target variable was actually measuring something else, acquirer size for instance, however, leaving out the size variable does not change the significance of any of the other variables. Including insignificant variables to the multivariate model can lower the significance of the other variables, however none of the combinations with only a few of the variables used in the regression alters the significance of variables. Also regressing the variables separately with the abnormal returns shows that only the size variable is significant. As such it can be concluded that size and to lesser extend cross border partly can explain abnormal returns. But the explanatory power of the regressions is low, about 1-4% of the abnormal returns is explained by the regression.
6. Conclusion
In this study, the price effect on the shares of a company that announces a merger or acquisition of a high-tech company is investigated. In order to do this a worldwide sample of 3035 companies that announced mergers or acquisitions is used. On average these companies experience a significant cumulative abnormal return of 0.60%. This is line with Kohers and Kohers (2000), they find that companies that announce a merger or acquisition with a high-tech company have positive abnormal returns over the event window. However it is contrary to the results of Benou and Madura (2005), as the find that the abnormal returns over their event window are negative. From the coefficients of the variables that are estimated in the multivariate model only size and to lesser extend the cross border variable can explain abnormal returns for the entire sample. For the different sub samples the result is somewhat different. In the North American market the method of payment affects returns, whereas in the European and Asian market, deals that took place during the high-tech bubble earned different returns than deals that took place after the high-tech bubble. However the explanatory power of the regressions is only very low. A weakness of this research is that there is no control group used to check whether the results that are founds really differ from mergers and acquisitions in which the target is a non high-tech company. Also differences between regions are observed, however it is not determined what causes the differences. For future research it would be valuable to try to explain the difference between countries in abnormal returns and the different variables
19 influencing the abnormal returns. It could also be that not characteristics of deals such as the method of payment and size of the acquirer but processes during the deals can better explain abnormal returns.
7.References
Benou, G., & Madura, J. (2005). High-tech acquisitions, firm specific characteristics and the role of investment bank advisors. The Journal of High Technology Management Research , 16, 101-120.
Bris, A., & Cabolis, C. (2008). The value of investor protection: firm evidence from cross-border mergers. The Review of Financial Studies , 21, 605-648.
Brown, S. J., & Warner, J. B. (1980). Measuring security price performance. Journal of Financial Economics , 8, 205-258.
Brown, S. J., & Warner, J. B. (1985). Using daily stock returns: the case of event studies. Journal of Financial Economics , 14, 3-31.
Carpenter, R. E., & Petersen, B. C. (2002). Capital market imperfections, high-tech investment and new equity financing. The Economic Journal , 112, 54-72.
Chang, S. (1998). Takeovers of privately held targets, method of payment, and bidder returns. The Journal of Finance , 53, 773-784.
Dong, M., Hirshleifer, D., Richardson, S., & Teoh, S. H. (2006). Does investor misvaluation drive the takeover market. The Journal of Finance , 61, 725-762.
Frick, K. A., & Torres, A. (2002). Learning from high-tech deals. The McKinsey Quarterly Report , 1, 113-123.
Fuller, K., Netter, J. M., & Stegemoller, M. (2002). What do returns to acquiring firms tell us?
Evidence from firms that make many acquisitions. The Journal of Finance , 57, 1763-1794.
Greenwood, R., & Nagel, S. (2009). Inexperienced investors and bubbles. Journal of Financial Economics , 93, 239-258.
Jensen, M. C. (1986). Cost of free cash flow, corporate finance, and takeovers. The American Economic Review , 76, 323-329.
Kennedy, K. H., Payne, T. G., & Whitehead, C. J. (2002). Matching industries between target and acquirer in high-tech mergers and acquisitions. Technology Analysis & Strategic Management , 14, 149-162.
Kohers, N., & Kohers, T. (2000). The value creation potential of high-tech mergers. Financial Analyst Journal , 40-50.
Koller, T., Goedhart, M., & Wessels, D. (2005). Valuation: measuring and managing the value of companies. New Jersey: John Wiley & Sons.
20 MacKinlay, A. C. (1997). Event studies in Econmics and finance. Journal of Economic Literature , 35, 13-39.
Maloney, M. T., McCormick, R. E., & Mitchell, M. L. (1993). Managerial dicision making and capital structure. Journal of Business , 66, 189-217.
Moeller, S. B., Schlingemann, F. P., & Stulz, R. M. (2004). Firm size and the gains from acquisitions.
Journal of Financial Economics , 73, 201-228.
Moeller, S. B., Schlingemann, F. P., & Stulz, R. M. (2005). Wealth destruction on a massive scale? A study of acquiring-firm returns in the recent merger wave. The Journal of Finance , 60, 757-782.
Ranft, A. L., & Lord, M. D. (2000). Acquiring new knowledge: the role of retaining human capital in acquisitions of high-tech firms. The Journal of High Technology Management Research , 11, 295-319.
Roll, R. (1986). The hubris hypothesis of corporate takeovers. The Journal of Business , 59, 197-216.
Thorbecke, W. (1997). On stock returns and monetary policy. The Journal of Finance , 52, 635-654.
Travlos, N. G. (1987). Corporate takeover bids, methods of payment and bidding firms' stock returns.
The Journal of Finance , 42, 943-963.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica , 48, 817-838.
