THE EFFECT OF SHARE REPURCHASES ON STOCK PRICE PERFORMANCE

Transcription

1 TILBURG UNIVERSITY THE EFFECT OF SHARE REPURCHASES ON STOCK PRICE PERFORMANCE Empirical evidences from The Netherlands and Belgium Master thesis Student name : Thanh Huyen Vu Student number : Administration number : Supervisor : Dr. Fabio Castiglionesi October 2017

2 Table of Contents 1. INTRODUCTION THEORETICAL FRAMEWORK AND HYPOTHESES SHARE REPURCHASE IN PAY-OUT POLICY METHODS OF CONDUCTING SHARE REPURCHASE Share Repurchase in Open Market Share Repurchase at Fixed Price Share Repurchase by Dutch Auction Negotiated Repurchase from Private Investors THE MOTIVES BEHIND SHARE REPURCHASE Signalling Hypothesis Free Cash Flow Hypothesis Capital Structure Adjustment Hypothesis Dividend Substitution Hypothesis Tax Efficiency Hypothesis Takeover Deterrence Hypothesis CURRENT STATE OF RECENT EMPIRICAL RESEARCH United State of America European Countries Asian Countries Other Regions and Countries HYPOTHESES TESTING RESEARCH METHODOLOGY EVENT STUDY METHODOLOGY OLS REGRESSION Dependent and Independent Variable Correlation Checking DATA SOURCES AND SAMPLE SELECTION DATA SOURCES SAMPLE SUMMARY RESULTS EVENT STUDY OLS REGRESSION CONCLUSION CONCLUSION LIMITATION RECOMMENDATION REFERENCES... 35

3 1. Introduction Since the start of share buyback as a legitimate financial program of firms, we have seen many criticisms against this activity. Some argue that buyback programs resulted from bad incentives of executives who aim to achieve higher EPS (Earning per Share) for higher compensation. Another opinion is that buyback damages the balance sheet and the ability of firms to confront challenges in the future due to using too much cash or issuing more debt. Share repurchase is also considered as the cost of innovation. It seems that investment opportunities cannot be taken because financial resources are already used for buyback. The consequence afterward is the unemployment. Facing various criticisms, however, buyback becomes more popular among companies all over the world. Although the third quarter of 2016 experienced a significant decrease in share repurchase value in comparison with the same period a year before, it still stayed at high level and it is still early to say about the end of Great Buyback Boom. The Business Insider said in 2016 that it seems like everyone are repurchasing shares, whereas the high-tech giant Apple was described as in the midst of a multibillion buyback program. What are the factors behind this popularity? As shown in the next sections of this research, there are a number of good motivations behind the decision of a share repurchase program, ranging from the undervaluation signalling to the takeover deterrence, which probably lead to positive reaction of the market. In reality, the trend of increase in share price after share repurchase programs was recorded in many countries and regions. Simultaneously, before announcements, the drops in stock price were also reported. While there are many evidences of negative stock price performance before and positive market reaction after share repurchase announcements over the world, there are limited number of researches conducted for Dutch and Belgian markets to answer the question of how the market react via share price around the moment of share repurchase announcement in The Netherlands and Belgium. Specifically, is stock price performance of the listed Dutch and Belgian companies positive or negative around the share repurchase announcements? This thesis aims to address this question. In addition, the other issue, investigated in this research, is the drivers of share repurchase. There are currently six hypotheses to explain motivations behind share repurchase announcements. They are Signaling Hypothesis, Free Cash Flow Hypothesis, Capital Structure Adjustment Hypothesis, Dividend Substitution Hypothesis, Tax Efficiency Hypothesis, and Takeover Deterrence Hypothesis. However, they did not show the equal strength in all cases. In fact, different datasets from different countries and regions showed evidences to support different hypothesis. This research tries to find out which factors determine the share repurchase program for Dutch and Belgium firms. Among many possible factors, the thesis focus on finding out whether the companies with different size of

4 total asset, different level of profitability, different pay-out policies experience different levels of stock price increase or not, which hypothesis is supported by Dutch and Belgian markets. To discover the answer for the questions stated above, data are taking from database of Bureau van Dijk Zephyr and Thomson Reuter Datastream. Particularly, for the first step, data of share repurchase announcements in the period of of the Dutch companies listed on Euronext Amsterdam and the Belgian companies listed on Euronext Brussels are obtained through Zephyr Database. Data for share repurchase announcement date and Datastream code are then imported to Thomson Datastream Database to get the daily stock price of firms to measure the market reaction to answer the first main question of the research. For the next step, to serve the empirical model of main factors leading to decision of share repurchase, Datastream will still be exploited to obtain the financial information of all companies. The financial information includes Total Asset (TA- representing size of firms), Free Cash Flow (FCF), Market-to-Book value (MTBV), Return on Asset (ROA), Leverage ratio, and Dividend Yield. The principal to build the desired dataset is that completed dataset includes firms that have all daily stock share price from 260 days before until 20 days after share repurchase announcements as well as all the necessary financial information stated above. In the end, the dataset contains totally 787 share repurchase announcements, dividing into 723 announcements for Dutch firms and 64 announcements for Belgian firms. The core method to find out the market reaction around event of share repurchase announcements is Event Study. The purpose of Event Study is calculating the abnormal return of events that is difference between the real return and normal return. The abnormal returns then are summed up based on the event window [-1,+1] that is mainly used in previous studies, as well as the pre-event period [-20,-2], [-5,-2], and post-event period [+2,+5], [+2,+20]. The Cumulative Average Abnormal Return (CAAR) for each country is calculated to show stock price performance on average. The result shows some differences between Dutch and Belgian market reactions. In Belgium, although the pre-event of [-20,-2] see a positive CAAR, the other pre-event [-5,-2] record the CAAR of -0.07%, reflecting the trend of stock price decrease as happening in other countries of previous studies. This is consistent with proposition that firms set time for share repurchase announcement to signal about undervaluation. For the event window [-1,+1] and post-event window [+2,+5], [+2,+20], CAARs are +0.78%, +0.09% and %, respectively. These figures meet the expectation about the increase in share price on and after a share repurchase program announced. However, they are not statistically significant probably dues to the small dataset size for Belgium. In contrast, almost all results of CAAR calculation for the Dutch subset are significantly different from zero at level 5% and 1%. The market in The Netherlands sees a decrease in stock price in the period before share repurchase announcement. For the period closed to the announcement, event window [-1,+1] presents a very strongly positive market reaction with CAAR of 4

5 +1.52% as expected. However, this trend does not continue after that. CAARs for post-event windows are negative. Surprisingly, the result of CAAR [+2,+5] is not significant. In another attempt, the subsets of The Netherlands and Belgium are pooled in one dataset to investigate the stock price performance in both countries as a whole. Once more time, the positive CAAR is found at announcement event window [-1,+1], significant at 5% level. The decreases in share price are recorded in both pre-event windows and post-event windows with statistically significant level of 1%, 5% and 10%. Furthermore, the share price performance of Dutch and Belgian firms around share repurchase announcement in three different periods- before, during, and after global financial crisis is tested. The differences in share price performance between three periods are small, except for event window [+2,+5]. In this window, while the negative coefficients are seen in the period of and , the post-crisis period shows an increase in stock price around share repurchase announcements. However, in this event window, no parameter shows statistical power. The statistical significance is observed for only one pre-event window in the pre-crisis and post-crisis periods. Whereas, the time of crisis experiences the decrease in share price for pre-event window [-20,-2] at 1% significant level and for post-event window at 10% significant level. This sub-period also shows an increase in share price in the window [-1,+1] at 5% significant level. Back to the second question of the research, to investigate drivers of share repurchase, the empirical model is introduced, in which the dependent variable is Cumulative Abnormal Return (CAR) of the event window [-1,+1] reflecting the stock price performance. The model also includes six independent variables. They are Firm-size, Market-to-Book value (MTBV), Free Cash Flow (FCF), Return on Asset (ROA), Leverage ratio and dummy variable Dividend. For the whole period from 2001 to 2015, as expected, the higher leverage companies see the higher raises in share price, while other factors like the size of firms, MTBV, level of free cash flow, level of profitability and whether dividend was paid or not, affect the share price in opposite dimension. However, these drivers do not show the equal effect on share repurchase. The determinants playing the most important role are Firmsize, and FCF with the coefficients are significant both in quantity and statistical meaning. This result strongly supports Signalling Hypothesis, and Free Cash Flow Hypothesis. For three subperiod, while the pre-crisis and post-crisis show the support for Free Cash Flow Hypothesis, the result for the crisis period of reflects the explanatory power of all five hypotheses, Signalling Hypothesis, Free Cash Flow Hypothesis, Capital Structure Adjustment Hypothesis, Tax Efficiency Hypothesis and the Dividend Substitution Hypothesis. These results, to a certain extent, are consistent with the findings in previous studies. In the research conducted by Lasfer with the data of European countries from 1985 to 1998, 11 share repurchase announcement in The Netherlands and 8 announcements in Belgium were collected and put together with the pool of announcements of other European countries. In this research, the author realised the CAAR for 5 days from the second day before to the second day after 5

6 announcement is +1.06%, significant at 10% level. The trend of increase in shares price continue over the +3 to +20 days after the announcement dates with CAAR equal 0.62%. However, it is not statistically significant. For the pre-event window of [-40,-3], stock market reaction is negative (CAAR= -0.17%) but like the result for post-event window, it is not significantly different from zero. More recently and in a larger scale, Manconi et al performed a research on a global sample from 32 countries including The Netherlands and Belgium, in the 11-year span from 1988 to Over the time span 3 days from one day before (day -1) to one day after share buyback announcement (day +1), the share price of firms in The Netherlands and in Belgium experienced a rise with CAAR of 1.54% and 1.46% respectively at 5% significant level. Expanding the event window to [-2,+2], the positive stock market reactions are even higher. The CAARs for Dutch and Belgian firms in this case are +2.02% and 2.07%. Also in this research, while conducting a range of OLS regression to assess determinants behind share buyback program, the author recognised that there was a strong negative relationship between prior return and CAR. In addition, the significantly negative coefficient of size of firms and CAR also reported, providing the evidence to strengthen the Signalling Hypothesis. In this study, the limitations could not be fully avoided. First, the dataset for Belgian market is small and incomplete from the original data source. There are only 79 share repurchase announcements made by Belgian firms in 15 years from 2001 to The missing data may lead to biased estimations and reduce the statistical significance for Belgium. The Belgian market reaction can be more reliable with better data collection. In addition, the result is also affected by the situation of multi-announcement, which means one company made more than one buyback announcement. The other limitation of this research is that the testing motivations of share repurchase are not done for all hypotheses. Only five hypotheses are tested while there is no variable in OLS regression model to test about Takeover Deterrence Hypothesis. To present the research in detail, the rest of the thesis is organized in 5 sections. Section 2 proposes some related theoretical frameworks and the recent empirical researches. For the next step, Section 3 Research Methodology, first introduces the method of Event Study. The second part of Section 3 provides the descriptions of regression model used in the research with full explanation about the dependent and independent variables. Section 4 Data Sources and Sample Selection, discusses about the issues of which data are needed, how and where to collect the such data, in which criteria data will be filled, step by step leading to the desired dataset. Furthermore, Section 5 Results, is dedicated to answer the questions of the research. The regression coefficients are given and analysed to conclude about the drivers of share repurchase program. Section 6 presents conclusions. Also, in this part, the limitations of the study as well as the recommendations for future research are discussed. In the end, a list of references is given. 6

7 2. Theoretical Framework and Hypotheses This section will discuss the main theoretical framework of share repurchase. It ranges from the share repurchase definition, the method of share repurchases, the motives behind share repurchase, as well as the trend of share repurchase development. The main part of this section is the proposals and findings of previous academic studies in a comparative context. Based on that, the hypotheses testing in the thesis are proposed at the end of this section Share Repurchase in Pay-out Policy Share repurchase is an activity of companies to re-acquire their own shares. Shareholders, who decide to sell shares back to companies, will be paid. Thus, share repurchase is considered as a method for companies to distribute cash to the selective shareholders while the other method of pay-out policy, dividend, is returned to all shareholders. Back in time, share repurchase is quite new in the history of pay-out policy. While dividend was used hundreds of years to return cash to shareholders, share repurchase have just appeared in the US as a legitimate practice from 1980s. Share buyback then became a strong trend among American companies. Although the amount of dividend continues to increase, the development of share buyback is considerably stronger, especially from 2000s. With the total amount of about $450 billion (data from Prof. Aswath Damodaran - Stern School of Business at New York University), in 2004, the value of stock repurchases start to exceed dividend. The year of 2007 is recorded as the peak of share repurchase activity with the value of $1150 billion while the amount of $820 billion was spent to return shareholders as dividends. The global financial crisis sunk the total value of share buyback to $650 billion in 2008 and to $420 billion in However, it recovered very fast in 2010 and reached $720 billion in From this time, although there was fluctuation year to year, the value of buyback was still at the high level. Share repurchase was approved in European later than in the US. Typically, in some countries like France and Germany, share repurchases were illegal until the legal reform in Together with the difference in corporate culture, which showed that European shareholders seem to prefer dividend to other kinds of pay-out, stock buyback in Europe did not experience a boom as happened in the US, according to Rau and Vermaelen (2001). However, inspire of conservative opinions over pay-out policy, share repurchases became increasingly dominant in European firms distribution program. According to Lasfer (2005), in 1997, the repurchase announcements made by European firms amounted to $47.2 billion, three time as much as value in 1996, which was equal to $14.2 billion. And in 2007, it reached to $157.8 billion. The share repurchase activity declined significantly after 2007 due to the crisis. The total value of buyback is only $38.3 billion in 2014, but it is expected to go up.

8 2.2. Methods of Conducting Share Repurchase It is commonly acknowledged that there are four different methods for companies to buy back their stocks. They are share repurchase in the open market, share repurchase at a fixed price, share repurchase by Dutch auction, and share repurchase by direct negotiation. Each one has its characteristic and thus, suitable for different purposes of offered companies Share Repurchase in Open Market Under this method, companies can execute the share repurchase program by buying back shares at the market price on the open market. This is the most favorite method of share repurchase. The study conducted by Grullon and Michaely (2004) in US showed that in the 17-year period from 1984 to 2000, there are 90% of the total share repurchase announcements undertaken in the form of open market repurchase. This popularity probably dues to the fact that open market share repurchases offer the cheapest and the most flexible way to buy back shares. This means companies can buy back share at lowest price and the buyback can take place in longer time in comparison with other methods of share repurchase Share Repurchase at Fixed Price Undertaking the share repurchases in this form means that companies offer to repurchase a certain number of their shares (also known as target number) at a fixed price. Normally, the offer price is higher than market price, which is considered as a way to stimulate shareholder to sell their shares. This method is often used when companies want to get back a large number of shares in the very short period in the attempt to avoid the risk of acquisition. All the tendered number of share will be repurchased if this number is smaller than the target number of company. In the other case, if this number is larger than the target number, the number of share repurchased is between target number and tendered number of share and specifically defined to ensure the fair deal for all participants. Normally, companies can buy back from different shareholders on the pro-rata basis Share Repurchase by Dutch Auction The mechanism of this method is quite similar to the method of fixed price tender offer. The point of difference is that under Dutch audition, companies will offer a range of price instead of a fixed price. Based on the offered prices, shareholders will quote the price at which they are willing to sell their shares. The company then will qualify these bids starting from the minimum price and moving up until getting the target number of repurchased shares. The price at which company finishes the target number of repurchased shares will be paid for all participants, including shareholders with lower price bids. With this mechanism, companies generally can repurchase shares at lower price under Dutch auction than under fixed price tender offer, but still at higher price than under open market. 8

9 Negotiated Repurchase from Private Investors The company will buy back its share from certain large shareholders at a price, which is negotiated between the company and the such shareholders. This method is often used when companies want to repurchase share to protect itself again potential takeover attempt, and when firms want to get share back from investors who undervalue company itself. The price in these cases will usually be higher than the current market price. In addition, this method is also employed if companies buy back shares from employees. In this situation, the price of share repurchased will be probably equal the market price. In the case that an investor wants to exit and sell shares back to company, the price may be lower than market price The Motives behind Share Repurchase Signalling Hypothesis Signaling hypothesis is considered as the predominant motivation behind the decision of firms share repurchase programs. Signaling hypothesis suggests that the share repurchase program is a way to transfer the signal of stock undervaluation. Specifically, managers are considered to have better information about company than outside investors. When managers believe that the stock price of the company lower than its true value, they can signal this belief by announcing a share repurchase program. The positive market reaction then will push the share price up to correct value. Many study showed that both share repurchase program and dividend plan can help mangers signal their firms value. However, the market reaction is different between them. Normally, the share price performance around a share repurchase program was significantly higher than that in dividend plan. According to Ofer and Thakor (1987), the reason for this difference is that share buyback programs convey more information content than dividend programs Free Cash Flow Hypothesis This hypothesis is also known as Agency Cost Hypothesis, which states that free cash flow should be distributed to shareholder through share repurchase if there is no profitable investment opportunity. In the other words, share repurchase is considered as a solution for managers to minimize wastefulness from manager due to rich cash position. While shareholders pursue the aim of maximize the firm value, firms managers want to maximize their benefit. Agency cost arises when managers work for their benefit, which is not in-line with benefit of shareholders. For example, if firms have excess cash while investment opportunities are limited, managers are more likely to invest in poor projects, which can destroy firm value, to pursue personal purposes. Therefore, if firms announce a share repurchase program, the market interpret that managers have less chances to waste cash. Thus, firms share experience the increase in price. 9

10 Capital Structure Adjustment Hypothesis Capital Structure Adjustment Hypothesis offers the idea that the target capital structure can be obtained by retiring share. When companies buy back their own shares, they reduce their equity and increase the leverage ratio. Especially, when companies use debt to finance repurchase program, leverage ratio even significantly go up. Therefore, share repurchase can be used by managers as a tool to achieve the target capital structure. However, according to Grullon and Ikenberry (2000), capital structure adjustment does not show as strong motivation for open market repurchase although for tender offer, it works very clear Dividend Substitution Hypothesis The hypothesis of Dividend Substitution discusses that share repurchase is a substitute of dividend in payout policy. Manager can choose repurchase shares to distribute cash to investors. However, in comparison with dividend, share repurchase presents more flexible in returning cash. While cash is returned to all shareholders in dividend paying program, only self-selected shareholders receive cash in share buyback. According to Damodaran (2014), the market will expect the companies continue to keep increasing dividend if such companies did the same thing in the past. This situation does not happen with share repurchase Tax Efficiency Hypothesis Tax efficiency hypothesis proposes that share repurchase is not just a substitute for dividend in payout policy of firms because shareholders can benefit more from share buyback than from dividend program. The tax rate of the capital gain tax, which applies for shareholder selling their shares back to companies, is normally lower than the tax rate applied for dividend. This makes the net cash distributed to shareholders higher. Thus, the increase of share price after share repurchase announcement is probably because market realizes the benefit of buyback program for shareholders in comparison with dividend Takeover Deterrence Hypothesis This hypothesis is first discussed by Bagwell (1991). In the research, he showed that share repurchase could offer a repurchase price higher than market price and thus make the available price for investor to sell stock higher. As a result, the cost of acquisition for potential acquirer increases. Therefore, firms with possibilities of being taken over are more likely to announce share repurchase program as protection mechanism against acquisition Current state of recent empirical research Around these hypotheses, an enormous number of empirical studies have been drawn to investigate the stock market behaviors and the drivers of such behaviors. 10

11 United State of America To examine the price behavior of the firm who repurchase their own shares, in 1981, Vermaelen conducted the model of abnormal returns around the announcement date with the sample of 131 tender offers made by 111 firms for the years , plus 243 open market repurchase made by 198 firms between 1970 and While he acknowledged the increase in stock price of the firms, which repurchase their share, he found that firm size plays an important role to signal true value of the firms. Small firms which are dominated by insiders can strongly convince the market over the undervaluation of firms via share repurchase since insiders will carry a significant burden in the case they buy the shares back at the price higher than intrinsic value. From the data on NYSE and ASE from 1980 to 1990, D Ikenberry, J Lakonishok, T Vermaelen (1995) provided the evidences of negative abnormal returns of -3.07% for the pre-announcement of share repurchase program. However, through 2 days after the public announcement, abnormal return was reported of 3.54%. They also found the effect of share repurchase in long term. Their study observed average abnormal four-year buy-and-hold return of 12.1% after announcement. In addition, the extra important result from this study was about the motivation of share repurchase. According to the researchers, undervaluation is an important reason of share repurchases for companies with high book-to-market ratio, while companies with low ratio, other reasons may play more important role. With longer horizon of study (20 years), but quarterly data instead of annual data used before, in 2005, Erik Lie showed in his study the mean abnormal stock returns of 3.0% for the event window [-1,+1]. At the same time, he found the evidences of improvement in operating performance of the firms that announced an open market share repurchase program. However, these results were only true for firms that had actual repurchase in the same quarter, but not for the rest. A research based on share repurchase announcements from 1980 to 1997, was conducted by Grullon, Gustavo, and Roni Michaely in Their evidence implied the positive market reaction to share repurchase announcements since these programs were expected to lead reduction in the agency costs of free cash flows and in the firm s cost of capital. This result is perfectly consistent with the Agency Cost Hypothesis. More recently, H. Almeida, V. Fos, M. Kronlund while acknowledged an increase in abnormal return, explored that EPS-motivated repurchases lead to reductions in employment, investment, and in cash holdings. It is regarded as a progress in the study of share buyback because they proved the causal effect of share repurchase and related factors whereas previous research only looked at the relationship between buyback program and factors like employment and investment European Countries In the year of 2002, Rau, P. Raghavendra and Theo Vermaelen conducted a research over effects of share repurchase in both short and long term with the data from the United Kingdom (UK). They 11

12 observed statistically significant CAR of 1.14% in the 11-day window surrounding the announcement. This figure is considerably smaller than that in US. Meanwhile, the result for long-term was a negative abnormal return of 7% in the year after the share repurchase announcement. The evidence of positive market behavior was also reported in Germany according to the study conducted by Seifert, Udo, and Richard Stehle in They obtained CAR of 4.793% on the announcement day for the dataset of 188 share repurchase announcements. This result is considerably higher than that of US but smaller than the outcome of other study for Germany conducted by Hackethal, Andreas, and Zdantchouk in Exploiting the sample of 785 buybacks plans from 1998 to 2003, they recorded the CAR of 5.97%. In this study, they also tested four hypotheses of primary motivation behind share repurchase and found some evidences to strongly support Signaling Hypothesis in Germany companies. In 2015, D. Andriosopoulos and M. Lasfer employed a sample of 970 buyback announcements in the UK, France, and Germany and they found stock prices decrease in the pre-event period (CAR=- 0.34%), but increase for the post event (CAR=0.31%). However, in contrast to US, these figures did not have statistical significance. Regarding to factors that influence firms market behavior, this paper concluded that taxation, shareholder protection, and the European Union s Market Abuse Directive did not affect significantly but firm internal conditions and reforms played more dominant role. In a larger scale, a research was conducted by Lasfer (2000) for 14 countries including Belgium, Denmark, Finland, France, Germany, Greece, Luxembourg, The Netherlands, Ireland, Italy, Spain, Sweden, Switzerland, and the UK. With the dominance of the number of UK share repurchase announcements (465 announcements out 642 total announcements), the dataset was divided in two subsets: the sample for UK and the sample for other European countries. For all observations from total dataset, the pre-event window [-40,-3] saw the insignificant positive cumulative average abnormal return of 0.12%. This result was not consistent with the finding from Ikenberry, J Lakonishok, T Vermaelen (1995), which showed that the share price decreased before buyback announcements. However, the subset of firms from other European countries except the UK experienced a fall of -0.17% in share price. In the announcement window [-2,+2], there was an increase in stock price for both subsets, as well as total dataset at 1% and 10% significant level. The post event window [+3,+20] also reported the positive share price performance but much smaller Asian Countries A decrease in share price in pre-repurchase period and an increase in post-repurchase period were also seen in some Asian countries. In Japan, Hatakeda, Takashi, and N.Isagawa (2004) employed 452 share buyback announcements made from November 1995 to November 1998 by firms listed in The First Section of the Tokyo Stock Exchange. The cumulative average abnormal return for pre-event from day -20 to day -1 was -2.94%, significant at 1% level while the abnormal return on the announcement day (day 0) was +0.91% and on the following day was +1.24%. 12

13 In the study conducted in Taiwan by Chang, Shao-Chi, Jung-Ho Lai, and Chen-Hsiang Yu (2005), it was observed the cumulative abnormal return CAR [ 1,+5] of 2.7%. At the same time, they also saw that firms with buybacks program showed a superior long-term performance in comparison with those reissued shares. Based on observations in Hong Kong, Firth, Michael, and Canna SF Yeung (2005) provided more evidence to support Signaling Hypothesis and once more time confirmed that the share repurchase program aimed to signal and correct undervaluation. Investigating the sample of 30 share buyback announcements from 2001 to 2005 in Thailand, Vithessonthi found the significant CAR of 5.37% for the 3-day period [-1,+1] and 6.83% for the event window [-2,+2]. The other finding of this research is the result of abnormal return for clean sample, which is 11 buyback programs announced not coincidently with other financial announcement of firms. Stocks in this sample experienced higher increase in price than in full sample. Specifically, CAR is 6.14% and 9.03% for event window [-1,+1] and [-2,+2], respectively Other Regions and Countries In other research employed Australian data, Brown and Christine (2007) measured the effect level of 1.2% on the announcement date of share repurchases. Using the discount-to-market price, the authors suggested the importance of tax benefit in the increase of share trading on the day of announcement. Taking the sample of 1060 Canadian share repurchase program from 1989 until 1997, Ikenberry and Vermaelen (2000) calculated the AR of less than 1% in the announcement month and abnormal performance of three-year holding period of 7%. They also showed that a better long-run performance was experienced by firms with repurchase program driven more by undervaluation than other factors. Analyzing 377 share repurchase announcements in Brazil during the crisis period of , Micheloud found that the stock prices did not react to buyback in the whole period but they did for the year of He also saw the evidences of negative relation between market price reaction and the size of companies, the evidences of positive relation between stock price performance and the number of shares announced for repurchase and the quantity of shares repurchased actually Hypotheses testing From previous findings based on datasets over the world, it is expected to see the positive market reaction around share buyback program. Therefore, first and foremost, the hypothesis testing is in the following. H1: The share price performance of firms in The Netherlands and Belgium is positive around the share repurchase announcement. 13

14 The share price before buyback announcement is also tested to check the time setting to signal about firm s value. H2: The share price performance of firms in The Netherlands and Belgium is negative before share repurchase announcements. The research period includes the global financial crisis Stocks experienced a deterioration in these years. According to Signalling Hypothesis, managers set the time, when the share price is low, to announce a share repurchase program to signal about undervaluation. Thus, this thesis will test the higher abnormal return in the crisis period. H3: There is a higher abnormal return around share repurchase announcements during the crisis period than in pre-crisis and post- crisis periods. 3. Research Methodology The thesis makes use of the method of Event Study. In addition, the method of OLS Regression (Ordinary Least Squares Regression) will be employed to investigate drivers behind share repurchase program. This section will explain about Event Study and OLS regression. 3.1 Event Study Methodology An event study is an important tool in finance. The main content of event study is to capture the effects of an economic event on the firm s value. The very first and primary published research was conducted by James Dolley (1933). From the sample of 95 stock splits, he recorded the price increase for 57 cases and price decrease for just 26 cases. Researches over event study were developed after that. The event study methodology as we employ today and as I will use in this research, was built by Fama, Fisher, Jensen, and Roll (1969) and was summarized by De Jong, De Goeij (2011). In particular, I will use the event study to investigate how the event of share repurchase announcement influent stock price of firms. This requires specifying the model for abnormal stock return behavior. This means that the abnormal returns (AR), which are differences between the actual return and normal return (The return in the case of no event happening), must be calculated. The positive AR will show the positive affect of event of share repurchase announcement. In contrast, negative affect of event will be represented by negative AR. I will clarify step by step on how to build the abnormal return model of event study methodology conducted in this research for the rest of this section. The part of result analysis will appear later in Section 5 of event study result. Step 1: Identify the timing for model 14

15 The announcement event window [-1,+1], which is mainly used in previous studies, will be chosen. To capture the market reaction before and after share repurchase announcement, the preevent period of [-5,-2], [-20,-2], and post-event period of [+2,+5], [+2,+20] will be considered. As the rule to choose the time for an estimation window, a window should not too long from event date to ensure the capture the trend of market, but also not too short to avoid the effect of event. In this research, the estimate window spans from day -260 to day -21 before event date of share repurchase announcement. Step 2: Calculate the coefficients of market return and return of each stock Using the estimate window, a range of OLS regressions are taken to achieve the coefficients of market return and return of each stock via market model. R "# = α " + β " R )# + ε "# For t = -260, -259, -258,..., -23, -22, -21 E (ε "# ) = 0, Var ε "# = σ 2 Where: R "# : The actual return of stock i at day t This return is calculated by the following formula: R "# = :;86< 5 =; >=? ;@ :;86< 5 =; >=? ;@A :;86< 5 =; >=? ;@A α " : β " : R )# : is an unknown constant The coefficient between market return and return of stock i at day t The market return at day t This return is calculated by the following formula: R ) = C=4<7; D4567 =; >=? ;@C=4<7; D4567 =; >=? ;@A C=4<7; D4567 =; >=? ;@A In this research, market price for the Dutch firms is the index of Euronext Amsterdam and for the Belgian firms is the index of Euronext Brussels. ε "# : The random figure Step 3: Calculate the normal return (NR "# ) The coefficients calculated in step 2 are used to achieve the expected returns of each firm for each announcement based on the following formula. NR "# = α " + β " R )# 15

16 For t = -20, -19, -18,..., 18, 19, 20 Step 4: Calculate the abnormal returns (AR "# ) Abnormal returns will represent the effects of share repurchase event on firms share price each day around announcement date. As introduced above, the abnormal returns are differences between the actual returns and the expected returns. AR "# = R "# NR "# For t = -20, -19, -18,..., 18, 19, 20 Step 5: Calculate the cumulative abnormal returns (CAR " ) and the cumulative average abnormal returns (CAAR) Cumulative abnormal returns are total effect of share repurchase event after concerning periods. Thus, they are calculated by summing up all abnormal returns during the event window periods. CAR " = AR ",#A + + AR ",#J = AR "# # J #K# A CAR of each announcement will be taken for different periods with different event windows [-20,-2], [-5,-2], [+2,+5], [+2,+20]. To capture the effect of share repurchase on subsets of Dutch firms and Belgium firms, the cumulative average abnormal returns (CAAR) will be calculated. Q CAAR = L M 5RA NOP 5 Step 6: Testing abnormal performance. The t-test is executed to test whether the calculated abnormal returns and cumulative abnormal return different from zero at a certain significance level. Thus, for abnormal return, the null hypothesis is the expected abnormal return of zero. H T : E AR "# = 0 The t-test is TS L = N OOP ; X ; N(0,1) Where: s # = A M (AR Q@A "KL "# AAR # ) 2 zero. For the cumulated abnormal return, the null hypothesis is the cumulative abnormal return of 16

17 H T : E CAR " = 0 The t-test is TS 2 = N NOOP X N(0,1) Where: s # = A M (CAR Q@A "KL " CAAR) 2 TS L and TS 2 then is compared to a certain critical value to conclude whether null hypothesis is rejected or not OLS Regression While regression is the method to help explain the relationship between one given variable (plays the role of dependent variable) and one or more other variables (independent variables), OLS regression provides the best linear unbiased estimators. Thus, OLS regression was used widely in economic and other research fields. In this research, OLS regression was firstly used to get the coefficients of market return and return of each stock when exploiting event study methodology. For the later part of thesis, OLS regression is the main method to find the motivations of share repurchase program by looking at the relation between various factors and cumulative abnormal return. The model and all variances will be explained in following sub section. The empirical model is CAR " = α + β L Firmsize " + β 2 MTBV " + β d FCF " + β e ROA " + β g Leverage " + β k Dividend " Dependent and Independent Variable Dependent Variable CAR " Cumulative Abnormal Return The dependent variable - CAR " in this model is cumulative abnormal return for the announcement event window [-1,+1], which reflected the total stock price performance in short time before and after share repurchase announcement. The cumulative abnormal returns calculated above from study event methodology are then regressed against a range of independent variables explained as following to find relationship between the price performance and relevant factors. Independent Variable Firmsize The size of firms is important factor related to market reaction around the share repurchase announcement since the firm size has the close relation with information asymmetry, while the positive stock price performance reflects the positive reaction of market after the manager released extra information about firm s health (according to signaling hypothesis). To be specific, large companies have lower level of information asymmetry since they have more exposure in front of public. The market thus has more ability to monitor them. In contrast, small firms and many of them 17

18 managed by family members, have higher level of information asymmetry. As a result, managers at small companies are more likely to announce share repurchase program to send signals of undervaluation to the market. Also, the signals of small companies are recognized more strongly than those of large ones. This above consideration allows expecting a negative relationship between Firmsize and CAR. To capture the change of abnormal return of the firm dues to the change of 1% in total asset of such firm, the independent variable Firmsize is defined as the natural logarithm of total assets. Independent variable MTBV- Market to Book Value MTBV - Market to Book Value is calculated by market value of a firm divided by its book value. According to Hackethal, Andreas, and Zdantchouk (2006), MTBV is determinant of buyback plan since companies with higher MTBV is expected to create more value in the future. Thus, companies with lower MTBV are more likely to be undervalued and they will prefer to employ a share repurchase program to signal about the true value. The negative relation between MTBV and CAR may be found in this regression. Independent variable FCF Free Cash Flow Free cash flow is the most leading measure to help investigate the existence of Free Cash Flow Hypothesis in share repurchase activity. As mentioned above, because of the benefit conflict, managers probably decide to invest the excess cash in negative NPV projects that is harmful for the value of companies. This so-called agency cost will be lower for companies with lower free cash flow and will be higher for rich cash companies. The market will react more positive with lower level free cash flow to show the belief that less money will be wasted by managers. Thus, the coefficient representing the relation between free cash flow and abnormal returns should be negative. In this research, FCF is defined as a fraction of free cash flow and the value of total assets. Independent variable ROA - Return on Total Assets ROA is defined as net profit divided by total asset. Investigation the relation between ROA and CAR allows finding out the effect of level of profitability on share price performance. This relation is negative in the explanation of Free Cash Flow Hypothesis. Firms with higher ROA are more profitable. They are expected to have more opportunities to invest in effective projects. Thus, they tend to employ cash flow for business project instead of distribution to shareholders via share repurchase. Firms with lower ROA, whereas, should pay out more to avoid agency cost due to the lack of valuable investment projects. The abnormal performance is better for these firms as a recognition of market about money wasteful limitation. Independent variable Leverage Leverage in this case is the ratio of total debts and total assets. Leverage is important factor dominant the stock price performance around share repurchase announcement since it represents 18

19 capital structure of firms. Share buyback will reduce the equity of firms, hence, make leverage ratio higher. In the case that firms decide to finance buyback program by debt, the leverage ratio become even much higher. Although higher leverage ratio makes firms riskier due to the increase of bankruptcy cost, it allows companies taking the advantage of tax shield. The optimal capital structure achieved by retiring equity and increasing the proportion of debt will help firms to get optimal cost of capital, thus maximum firms value. Firms with higher leverage ratio will be interpreted to be closer to optimal capital structure; thus, their abnormal performance is better than peers with the lower leverage ratio. Therefore, it is more likely that the research will observe the positive relationship between CAR and variable Leverage. Independent variable Dividend Dividend is dummy variable, which is set as 1 if firm paid dividend and 0 if it did not do that. As explained in Tax Efficiency Hypothesis, because of the lower tax rate of capital tax gain, share repurchase programs are preferable to dividend. Therefore, companies, which paid no dividend, can experience higher abnormal return than companies that had already dividend payment program. The coefficient of CAR and variable dividend then can be negative. This relation also proves the appropriation of market of share repurchase as a replacement for dividend. In summary, the description of all independent variables is showed in the following table 1. Table 1: Summary of Independent Variables Variables Proxy Define Expected Sign lntotalasset Firm size = ln (Total Assets) negative MTBV Market to Book Value = Market Value / Book Value negative FCF Free Cash Flow = Free Cash Flow/ Total Assets negative ROA Return on Asset = Net Profit/ Total Assets negative leverage Leverage ratio = Total Debts/ Total Assets positive dumdiv Dummy Dividend = 1 if firm pays dividend and 0 if it does not do that negative Correlation Checking The regression result will not reflect the importance of each factor to the dependent variable in OLS regression model if there is a phenomenon of multicollinearity, which happens when some predictor variables related highly to each other and together affect to outcome variable. Therefore, to avoid the multicollinearity, before regressions are performed, the correlation between six independent variables are checked to define which variables are correlated to others. The variables with high 19

20 correlation coefficients will not appear together in an OLS regression. Instead, the separate regressions will be performed. 4. Data Sources and Sample Selection The sources of data and the way to collect such data will be explained in this section. Also, for the second part of this section, the summary of dataset will be presented Data Sources To investigate the effect of share repurchase on stock price performance in The Netherlands and Belgium, I will use the data of share repurchase of Dutch companies listed in Euronext Amsterdam and of the Belgian companies listed on Euronext Brussels. The finance information of companies for regression model is also needed to collect to investigate the determinants of share repurchase. The process of getting data will be explained step by step as the following. First, all the data of share repurchase announcements of Dutch and Belgian companies are gathered from database of Bureau van Dijk - Zephyr. The horizon for research is 15 years from 2001 to The criteria of firm sector and Datastream code (ISIN code) are looked up to collect other data from Datastream later. Totally, 996 announcements are reported from Dutch companies and 142 from Belgian firms in this period. While a dominant part of firms with share repurchase programs are listed either in Euronext Amsterdam or in Euronext Brussels, there is a small part of firms not listed or listed on other stock exchanges. These companies are removed from the final sample. Since the financial companies are under the restrictions of the rules of capital, only nonfinancial firms are included in final dataset. Together with dropping the firms without ISIN code out of the sample, in total, Zephyr reported 838 share buyback announcements including 759 announcements by firms listed on the Euronext Amsterdam and 79 announcements by firms listed on the Euronext Brussels. In fact, these numbers are not logical. Considering the size of Euronext Brussels, it is expected that there are more actual announcements than in the dataset. Second, the ISIN code and announcement date gathered above are imported into Thomson Datastream Database to collect the data of stock price for event study. As discussed in Section 3 Research Methodology, to achieve the coefficients of market return and return of each stock, the return of stock from 260 days before announcement date to 21 days before announcement date are needed. Therefore, all historical stock prices in this period are collected. The market indices, which are AEX index for Dutch companies and BEL20 index for Belgian companies, are also collected for the same time. In order to calculate the normal stock return for event windows [-20,-2], [-5,-2], [-1,+1], [+2,+5], [+2,+20], the historical stock price of firms announced share repurchase and the market 20