The effect of share repurchases on stock returns in Europe from

Transcription

1 The effect of share repurchases on stock returns in Europe from Master Thesis Department of Finance Tilburg University Student: Marouane Ziani Administration number: Faculty: School of Economics and Management Study Program: Finance Supervisor: Dr. K.K. Nazliben Second reader: Dr. F. Castiglionesi Date of defense: Words:

2 Abstract This paper examines the effect of open-market share repurchase announcements on share prices in the period from , including the financial crisis period ( ). In addition, it will look into the stock returns abnormal performance of the FTSEurofirst300 firms in different periods: pre-crisis period, during the financial crisis and post-crisis period. This paper found an average abnormal return of 1.67% at the event date (t=0) over the period , which is positively significant at the 1% level. Including the financial crisis period ( ), we found a higher average abnormal return of 2.40% at the event date (t=0), which is positively significant at the 1% level. In addition, the cumulative abnormal returns are positively significant for different event windows: (-1,+1), (-3,+3) and (-5,+5). The results are similar to Vermaelen (1981), where significant positive abnormal returns are found during the crisis period Firms might offer a premium when repurchasing their own shares, so that investors could signal positive information and the market could come up with higher share prices around the share repurchase announcement date (Vermaelen, (1981)). 2

3 Table of contents Table of contents Introduction... 4 Literature review and Hypotheses development Trade-off between share repurchase and paying out dividends Methods of repurchasing shares Motives to repurchase shares Empirical Results Hypothesis development Data Methodology Results Conclusion References Appendix

4 Chapter 1: Introduction Introduction In recent years, open-market repurchase programs have become an important payout method for many European firms to distribute (excess) cash to shareholders. Share repurchase programs increased in popularity in Europe during the late 1990s (v.eije and Megginson, (2008)). Since the late 1990s, open-market share repurchases were allowed in Europe. Before share buybacks were prohibited by the government in many European countries, since repurchase announcements were not considered as firm commitments. According to Babenko (2012), managers could do share repurchase announcements to mislead the market, since the open-market repurchases are announced without any commitments. Therefore, managers are not obligated to complete the share repurchase program after the announcement. So, managers could use share repurchase announcements to manipulate the stock price, which could lead to an artificially increase of the firm s share price (Fried, (2001)). Furthermore, Babenko, Tserlukevich and Vedrashko (2012) found that the market responds more favorable to share repurchase announcements, when insiders purchase stocks in the pre-announcement period. This is important information for outside investors to measure the credibility of the undervaluation signal in the open market repurchase announcements. In this paper, we will look into managers their main motives to execute open-market share repurchases and it will analyze the share price impact of European firms after a repurchase announcement in the period , including the financial crisis in Europe ( ). According to Reuter s data 1, the totaling European share repurchase value reached a peak of $ billion in 2007, compared to the total deal value of $38.3 billion in Looking to the buybacks of US firms between 2009 and 2014, we notice a share repurchase value of over $2 trillion. In this study, we will look into the period , since share repurchases were allowed in Europe in the late 1990 s. In addition, we excluded the period , because many European firms listed on the FTSEurofirst300 index, had less data points in this period. The time period , includes the different time periods: pre-crisis ( ), crisis ( ), post-crisis ( )

5 Many studies have been performed in the past to examine the effect of share repurchase announcements on the movement of share prices. Vermaelen (1981), studied the effect of share repurchase announcement during the oil crisis in the period 1970 to 1978 and found positive abnormal returns. In this study we also measure the effect of share repurchases on share prices for the period 2008 to 2012 and we still expect positive returns, but lower abnormal returns as result of the lack of confidence and the cautious behavior of investors after the financial crisis, which started at the beginning of During the financial crisis, many financial institutions went bankrupt and banks needed to be rescued by the national government. The purpose of this paper, is to examine how share repurchase announcements affect share prices in the period from Moreover, this study looks into the motivations behind firm s decision to repurchase shares and measures the effect on firms share prices. The economic motivations for share repurchases are: First, to improve shareholder s value. In addition, share repurchases decrease the number of outstanding shares which increases shareholder s claim on firm s future profitability and thus leads to higher earnings per share which benefits current shareholders. Sequentially, this could lead to an increase of market optimism and will also lead to a higher stock market activity. Second, share repurchases could give a boost to firm s share prices, since firms believe that their shares are undervalued by the market, so managers could signal undervaluation by announcing share repurchases. Therefore, share repurchases could have a positive significant impact on the investor s portfolio (investor s confidence will also increase), because a share repurchase means that the firm has confidence in its future performance. So, when the firm believes that its shares are undervalued, buybacks represent the best use of cash at that time period. Third, share repurchases could be used for liquidity purposes, which could help to increase the stock s liquidity in the market and could stabilize firm s share price. According to Babenko (2009), share repurchases could also increase pay-performance incentives of employee compensation and could lead to higher employee effort and higher share prices. The reason for this, is that employees could exercise their stock options earlier and receive higher compensation, so this might increase employees work incentives and could lead to a lower level of unemployment. Including the legal aspects of share repurchases, financial institutions were prohibited to execute share repurchase programs before the late 1990 s. The EU law had the aim to limit market manipulation, the result was a relatively low volume of share repurchases in Europe (Siems & De Cesari, (2012)). In addition, the tax rate on dividend payments is higher than the tax rate on capital gains. Firms only pay taxes on their capital gains which is defined as the difference between the share 5

6 selling price and the firm s purchasing price and there are no taxes paid on their capital losses. So, share repurchases allow investors to lower their tax burden by optimizing the timing of the share buybacks. Share repurchase could be seen as a flexible payout to shareholders and is often chosen when firm s earnings are volatile and temporary (Siems and De Cesari, (2012)). Furthermore, the open-market repurchases in this paper, are related to the signaling undervaluation hypothesis, free cash flow hypothesis and the optimal capital structure as stated in the payout literature (e.g. Ikkenberry, Lakonishok and Vermaelen (1995), Grullon and Michaely (2004)). In addition, this paper will also examine the main drivers behind the abnormal returns. Dann and Vermaelen (1981), found that the signaling undervaluation hypothesis and the free cash flow hypothesis are the main explanation for the evidence of abnormal returns. This study will also include the optimal leverage theory. This theory states that firms pursue an optimal leverage ratio, unleveraged firms use share repurchase programs to increase their debt ratio to reach their optimal capital structure. Our sample consists of companies from the FTSEurofirst 300 and it will focus on share repurchase announcements between January 2005 and December 2015 to measure the effect on share prices in the pre-crisis, during the crisis and post-crisis period. We use the FTSEurofirst 300 data, because larger firms are more likely to announce their motives to repurchase their shares in the open-market than smaller firms. Since larger firms provide more public information about their financial performance compared to smaller firms. Recently, share repurchases become very popular in Europe, according to Reuters European firms have increased their cash returns to investors to stimulate their stock prices. In addition, Reuter s data found that many European firms have announced share repurchase value of $8 billion in 2015 and it was more likely that the volume of repurchases will increase: European firms had over the $1.5 trillion in cash on their balance sheet. In addition, we may expect that the undervaluation of shares and the distribution of excess cash to shareholders are still important arguments for firms to buy shares back. Empirical evidence found by Dann and Vermaelen (1981), is that the effect of abnormal returns is bigger for smaller firms than for larger firms. Smaller firms provide less private information about their performances, therefore we may expect that smaller firms will have more information asymmetry and hence higher abnormal returns after share repurchases than larger firms. Besides that, we will also include the differences in country characteristics when the firm announces a share repurchase. The motives for the allocation of share repurchases and dividends are country specific determined. According to Mitchell and Dharmawan (2007), share repurchases and dividends could be seen as complements in Germany and UK, while in France share repurchases are substitutes for dividends, this is shown in their study by a negative 6

7 coefficient of cash dividends and supports Grullon and Michaely (2002) and Skinner (2008) their findings. There are nowadays two important reasons for explaining the increase in the use of share repurchase programs by firms. First, is that firm s management might use open-market repurchases to signal positive forecast about its firm performance (Bhattacharya (1979), Miller and Rock (1985), and Vermaelen (1984)). In addition, these studies suggest that share repurchases could be seen as a costly signal about future cash flows when markets are inefficient. Stephens and Weisbach (1998) provide similar results, repurchasing firms buy their shares back when their shares are perceived as undervalued. Stephens and Weisbach found for repurchasing activities a negative correlation with prior returns. So, firm s repurchase decision could reveal significant information about future earnings to the market. Second, is that share repurchase programs could reduce the amount of free cash flow at management s level. According to Jensen (1986), firms might repurchase to reduce the potential over-investment by management. In addition, this refers to the cash-flow-signaling hypothesis, which implies a reduction in the growth opportunities and a decline in the return on assets. In that case, firms prefer to payout cash in the form of share buybacks. Furthermore, Grullon and Michaely (2004) could not find evidence for the signaling hypothesis that repurchasing firms will have better prospects relative to other similar market participants. In some cases, repurchasing firms have underperformed relative to their peer group. In addition, there is evidence found for the cash-flow-signaling hypothesis. Grullon and Michaely (2004) found that repurchasing firms decrease their current level of capital expenditures and R&D expenses, and have a significant decline in the level of cash reserves on the firm s balance sheet. At the same time, agency problems are more likely to arise when firms become more mature, because firm s growth opportunities will decline, hence lower investments are required. When firms have lower growth opportunities, current firm s assets will play a bigger role in determining firm value, which leads to a lower level of systematic risk (Berk, Green, and Naik (1999)). For example, the free cash flow hypothesis suggests that firms should pay out cash to its shareholders when agency problems lead to overinvestment by management. Firms should increase their payout when they have less investment opportunities. According to Easterbrook (1984) and Jensen (1986), share repurchase programs could help to reduce the agency costs related to possible overinvestment of firm s free cash flow, which is in line with the free cash flow hypothesis. In addition, a decline in systematic risk leads to a lower level of the cost of capital. Grullon and Michaely (2004) found evidence that repurchasing firms experience a significant reduction in systematic risk, relative to non-repurchasing firms. A decrease in the firm s investment 7

8 opportunities is not good news, however the market reacts positively to share repurchase announcements. When profitable investments become scarce then a decline in the amount of free cash at management s disposal leads to lower agency costs and systematic risk (Jensen, (1986)). Several studies (Boehme and Sorescu (2002) and Grullon, Michaely and Swaminathan (2002)) support this evidence of significant decline in risk and cost of capital for repurchasing firms relative to non-repurchasing firms. Furthermore, Bagwell and Shoven (1989) and Opler and Titman (1996) show that firms may use repurchase programs to increase their leverage ratio. According to Dittmar (2000) firms are more likely to counter hostile takeover attempts during periods of peak merger activity. During the mid-1980s, firms faced hostile takeover attempts, the use of management stock options increased significantly during this period. So, more firms may have a higher fraction of preferred repurchases to dividends. In order to test the evidence of abnormal returns after share repurchase announcements, an event study will be conducted in which the actual returns around the event date are compared with the estimated normal returns of the FTSEurofirst 300 firms during a non-event period. Next to the event study, an OLS regression will be performed to examine if the free cash flow theory and signaling theory still hold. In this paper, I will look into different event windows around the announcement date (t=0) to test the cumulative abnormal returns (CAR), which is the dependent variable and will be performed over different windows: three-day event window (-1, +1), seven-day event window (- 3,+3) and 11-day event window (-5,+5). In addition, the estimation period of this event study will be 250 days to 21 days before the event day. The regression model consists of an Ordinary Least Square (OLS) multivariate regression analysis and is performed to test several specific variables that are related to the share prices. The independent variables of the OLS regression model are: (ln) Total assets, Long-term debt ratio, Market value to book ratio, Return-on-assets, Return-on-equity, Cash flow to sales and Share repurchase ratio. In addition, one dummy variable is created: a Crisis dummy. The data section will provide a full description of each independent variable and of the dependent variable (CAR). The main questions that need to be answered by this study are as follow: 1. Does the European market respond differently to share repurchase announcements in the period ? In addition, how are repurchase announcements affect share prices? 2. How do firm performance affect share prices? 8

9 Further, this paper is organized as follows: chapter 2 starts with the literature review section. This section provides theoretical background information about share repurchases, to gain more understanding about share repurchase announcements and the performance of abnormal returns. This section will begin by comparing share repurchases with paying dividends and will examine why firms prefer share repurchases to dividend payouts. Furthermore, it includes different methods of share repurchase announcements and provides the main motives for managers to repurchase shares. Afterwards, there will be an overview of the empirical results from previous studies related to the effect of share repurchases on share prices. The last section includes the hypotheses development, which will be tested by the event study. Chapter 3 includes the data collection, specification of the independent variables, including their descriptive statistics and its correlation matrix. In chapter 4, an explanation of the used research methodology in this study will be given. Furthermore, chapter 5 will include the empirical results from the event study and the multivariate regression analysis. In addition, chapter 5 also includes the discussion of the empirical findings. The last chapter includes the main conclusions that can be drawn from the findings. In addition, this section will also provide the limitations of this study and provide recommendations for further research. 9

10 Chapter 2: Literature review and Hypotheses Development Literature review and Hypotheses development First, we start by explaining how share repurchases are defined in this study. Therefore an important question in this study is: What is a share repurchase? A share repurchase is an event where the firm is acquiring its own shares. The firm is distributing cash to its shareholder in return for a fraction of the outstanding equity. A share repurchase distributes cash to the shareholders that are willing to sell their shares (De Cesari, Espenlaub, Khurshed, and Simkovic, (2012)). After the share buy back, the number of outstanding shares will decrease in the market and as a result the price of the remaining shares will increase. Additionally, the remaining shareholders will lead to an increase in the size of shareholders claim on firm s future profits, which improves future financial ratios such as dividend per share (DPS) and the earnings per share (EPS). Recent years, share repurchase increased in popularity as payout method in the United States and Europe. There are many forms of share repurchases done in different timeframes. There is no specific timeperiod given and share repurchase announcements are not always fully exercised or completed. Grullon and Ikenberry (2000), found that more than 95% of the share repurchases are classified as open-market repurchases. In addition, this study will focus on the effects of share repurchase announcements on share prices. In their papers, Modigliani and Miller (1961) assume perfect capital market conditions, which means efficient markets, no information asymmetry between the shareholders and the firm, no bankruptcy costs and taxes. In the case of perfect markets, Miller and Modigliani s irrelevance proposition on payout policy may hold, firm value is then not affected. According to the Modigliani-Miller-Theorem, firm value is determined by the firm s cash flows generated from its assets. Share repurchases could be seen as substitutes for dividend payouts. However, the perfect market does not hold in reality, due to market imperfections and information asymmetries. So, the decision between share repurchase and dividend payout is still relevant for firm value. 10

11 2.1 Trade-off between share repurchase and paying out dividends Firms have basically two ways to distribute excess cash to shareholders: repurchasing shares and paying out dividends. Managers often prefer the announcement of share repurchase programs over paying out dividends, because it is financially more interesting for the firm. First, it provides more financial flexibility to the firm and depending on the repurchase method, firms could have less obligations when repurchasing shares compared to dividend payouts. According to Dittmar (2000), an open-market repurchase, is the most used repurchase method, which means that the firm has no obligations to actually repurchase the shares. In addition, there is no significant negative impact on firms share price when the share repurchase program is canceled. This is because firms do not announce share repurchases on regular basis, so shareholders might have lower expectations after a share repurchase announcement. However, when firms decide to pay out dividends, then firms have the obligation to pay out the amount of cash once the dividend announcement is made. Shareholders will expect that the firm will payout the similar dividend amount on regular basis. Furthermore, Grullon and Michaely (2002) state that when firms are fairly valued, then dividend payouts will be from an economic perspective, similar to repurchasing shares. However, there are still some important differences in tax treatment and investors income. According to Barclay and Smith (1988), share repurchases might be preferred over paying out dividends, because the taxation on capital gains is lower than on dividend income. Only the shareholders that made their choice to sell their shares during the share buyback will incur a capital gain on taxes. The remaining shareholders that keep their amount of shares will receive a pro-rata increase in their ownership, and will have higher EPS in the future, this makes their fraction of shares more valuable. The remaining shareholders don t have to pay any taxes, because they didn t have realized any profits yet. According to Thein (2013), there is a significant difference between paying out dividends and repurchasing shares, shareholders retain their rights as owner of the entity after dividend payout and the shareholders might expect that the firm will pay dividends in the future, till liquidation or until the shareholders sell their shares out. Furthermore, managers might prefer share repurchases over dividends, especially when they own a lot of stock options. The value of stock options is determined by its demand in the market. When employees exercise a lot of stock options, the value of the stock option will get diluted, because of the higher supply of the outstanding stocks. In that case, repurchasing shares could increase the value of the stock options. According to Dittmar (2000), managers who own a lot of stock options, are more interested in share repurchases than dividend payouts, because it helps them to increase their wealth and dividend payouts does not. In addition, firms are in favor of repurchasing shares, specifically when the shares are undervalued. 11

12 2.2 Methods of repurchasing shares In this section, we provide more details about the different methods of share repurchases. The method of share repurchase might have significant effects on share prices and firm value. There are three methods that are mostly used after share repurchase announcements: the Dutch auction share repurchase, the fixed-price tender offer and the open-market share repurchase (Stephens and Weisbach, (1998)). The Dutch auction and the fixed-price tender offer methods are used by the firm, when the purpose of the firm is to withdraw a large amount of shares in a short time period The Dutch auction share repurchase According to Grullon and Ikenberry (2000), this method of repurchase allows managers to solicit information (before repurchasing shares) from shareholders to form a final transaction price. This transaction price is revealed at the end of the repurchasing process and is not set by the traditional approach. The process works as follows: managers start to announce a range of prices at which they will accept offers that are proposed by shareholders. Afterwards, shareholders will participate by telling the firm how much shares they will offer at which price and announce the amount of shares that they are rendering. When the offer period ends, then managers start to collect the individual offers and rank them by price. In addition, the exact price at which the share repurchase program is completed is defined by summing up the number of shares that are offered, beginning with the lowest managers price range. The transaction price will be fixed at the level, where the cumulative amount of shares equals the size of the repurchasing share program. In this case, only the shareholders who tender at or below the given transaction price are included in the share repurchase program, these shareholders will all receive the same transaction price per share. The shareholders, who tendered at a price above the transaction price, will be excluded The fixed-price tender offer The fixed-price tender offer allows the firm to offer a single repurchase price to all shareholders for a given number of shares and is mostly used for a limited period of time (Grullon and Ikenberry, (2000)). The firm could extend this time period, when shareholders are not offering the requested amount of shares. In that case, the firm could buy additional number of shares to complete the repurchase program. Furthermore, firms have the right to change the amount of shares that they wanted to repurchase. However, the announced amount of shares is the limit to repurchase shares in practice. 12

13 2.2.3 The open-market repurchase The last method is the open-market repurchase, which is the most used method for repurchasing shares in Europe and US over the last decade. According to Grullon and Ikenberry (2000), almost 91% of the total deal value of share repurchase programs in the US in the period , which is related to open-market repurchases. According to Michel et al. (2010), an open-market repurchase is not always fully completed, because it is more time-consuming compared to the fixed-price tender offer and Dutch auction share repurchase methods. According to Oded (2005), firms can buy their shares back over a period that could vary from months to years. In addition, Oded (2005) found empirical evidence that share repurchase announcements could be seen as good news for the market, even as only a small proportion is actually repurchased by the firm. According to Barclay and Smith (1988), an open-market repurchase programs could provide incentives to managers to use private information to benefit at the expense of the shareholders. In addition, firms can use repurchase programs to affect the liquidity of the firm. 2.3 Motives to repurchase shares The Signaling/ Information Hypothesis The Miller-Modigliani theorem (1961) states that when markets are inefficient, firms can reveal information about their future cash flows through changes in firms pay out policy. Many studies support this argument, Bhattacharya (1979), Miller and Rock (1985) and Vermaelen (1981) found that firms payout decisions are explicit signals about future earnings. According to Grullon and Michaely (2004), the Signaling hypothesis has three underlying assumptions: 1. share repurchase announcements might be associated with positive price changes; 2. when firms start to repurchase shares, it could reveal good news about firms cash flows or profitability; 3. if the market expects positive changes in future profitability, share repurchase announcements will immediately follow. In addition, many studies (e.g. Ikenberry, Lakonishok, and Vermaelen (1995)) found a positive relationship between share repurchases and stock prices volatility. 13

14 2.3.2 The Free Cash-Flow Hypothesis According to the agency theory, it is more likely that firms with excess cash will spend it on valuedestroying investment projects, which could reduce firm value (Grullon and Michaely, (2004)). Many studies found (Grossman and Hart (1982), Easterbrook (1984), and Jensen (1986)) found similar results that when the amount of cash is minimized at management level, then it will become more difficult for managers to invest in negative NPV projects. In that case, it might be better to lower the amount of excess cash and to increase the level of payout to equity holders. Furthermore, firms have different payout methods that are important for the transformation from its growth phase to the mature phase. According to Grullon and Michaely (2004), when a firm is at its growth phase, we may expect many positive NPV projects, higher earnings growth, higher capital expenditures and lower levels of free cash flow. At some point in time, the firm s growth will decline (lower level of systematic risk), and in that case the firm s assets will play a significant role in determining firm value (Berk, Green and Naik (1999)). In addition, when the firm s risk profile declines, the return on investment and its profitability will also decrease. In this case, a decrease in the return on investment will lower the number of investments and could reduce firm s earnings growth rate. When the firm has lower investment opportunities, it is more likely that it will undertake less new investments. However, firm s free cash flows will increase after the decline in investments. So, the potential for managers to overinvest could increase when the firm is at the mature phase. Therefore, shareholders might prefer higher payout ratios, because share repurchases are related to lower investment opportunities and lower systematic risk. Simultaneously, share repurchase announcements by management could reveal information about the investment opportunities in the market, so managers could use share repurchases to reduce capital expenditures as reaction to the lower investment opportunities The Optimal Capital Hypothesis According to Dittmar (2000), firms could use share repurchases to distribute excess cash to shareholders and to decrease its level of equity. This will increase firms leverage ratio and could help firms to reach their theoretical optimum leverage ratio (Bagwell and Shoven, (1989)). In addition, we might expect that the capital structure plays an important role for determining whether the firm announces share repurchases or not. So, it is more likely that when firms have lower leverage ratios than their optimal leverage ratio, then it will repurchase shares. 14

15 2.3.4 The Take-over Deterrence Hypothesis According to Dittmar (2000), firms could perform share repurchases to prevent hostile take-overs. In addition, it is not always the case that the threat is real, however firms prefer to be prepared in case of hostile take-overs (Thein, (2013)). Many studies (Dittmar (2000); Thein (2013)) found that when share repurchases are correctly timed, share prices will go up for a certain period of time. So, share repurchases could be used as a good take-over defense strategy. In addition, Dittmar (2000) found that potential target firms are more likely to repurchase shares. 2.4 Empirical Results Since the 1980s, share repurchase were very popular in the United States. In Europe, share repurchases were for a long period of time forbidden and since the last decade it became very popular. Many studies have examined the effect of share repurchases on share prices. Most of the literature about the evidence of abnormal returns is mainly focused on U.S. event studies. Under perfect market conditions, we might expect that the market is perfectly informed and all information is incorporated in the share prices. However, in practice markets are not efficient in that case we could observe abnormal returns around the repurchase announcement dates. This section provides a summary of the empirical results of the abnormal returns from previous studies. First it will provide the results for the short run and second it will provide results for the long run. Looking to Table 1, we see an overview of empirical results from other papers. We could see that previous studies found positive cumulative abnormal returns (CAR) after share repurchase announcements (from ) in the U.S., which varies between 1.5% and 3.5% (event window: (-1,+1)) for the short-term period. 15

16 Table 1: Empirical results of abnormal returns Short-term period Study Sample period Number of obs. Country Event window (days) CAR Vermaelen (1981) U.S. (-1,+1) 3.0% Ikenberry et al. (1995) U.S. (-1,+1) 3.5% Grullon and Michaely (2002) U.S. (-1,+1) 2.7% Kahle (2002) U.S. (-1,+1) 1.5% Peyer and Vermaelen (2005) U.S. (-1,+1) 2.4% Ikenberry et al. (2000) Canada (-15,+15) 0.9% Rees (1996) U.K. (-2,+2) 0.3% Rau and Vermaelen (2002) U.K. (-2,+2) 1.1% Andres et al. (2013) Germany (-1,+1) 3.55% Zhang (2002) Japan (-1,+2) 6.0% Long-term period Study Sample period Number of obs. Country Event window (years) CAR Ikenberry et al. (1995) U.S % Ikenberry et al. (2000) Canada % In addition, we see positive CAR of 0.9% (event window: -15,+15) in Canada. Further, we see that repurchasing firms in the UK found evidence of positive abnormal returns varying from 0.3% to 1.1% (event window: -2,+2) for the short term period. Besides that, we notice positive abnormal returns of 3.55% (event window: -1,+1) in Germany for the short-term period. There is also evidence found of positive abnormal returns of 6.0% (event window: -1,+2) in Japan. 16

17 Furthermore, looking to the long-term period (2-3 years), we see higher cumulative abnormal returns compared to the short-term period. In the US, we notice positive CAR of 8.7% (event window: 3 years) and for Canada we found evidence of positive CAR of 14.9% (event window: 2 years) after the share repurchase announcement. Most studies about the evidence of abnormal returns are mainly focused on the short term period, assuming that the efficient market hypothesis holds. This theory states that share prices incorporate all relevant information in a short period of time after the repurchase announcement. According to Vermaelen (2005), there is evidence found of long-term abnormal returns, however interpreting long-term data could lead to many methodological and statistical issues. In addition, the long-term performance of abnormal returns could be explained by omitting variables. 2.5 Hypothesis development This section includes the formulation of the hypotheses based on the previous literature. It will examine whether the effect of share repurchase announcement on share prices during the financial crisis differs from the market reaction in the pre-crisis and post-crisis period. In this study, the crisis starts after the collapse of the investment bank Lehman Brothers on September The collapse of Lehman Brothers caused worldwide big financial shocks. According to Smith (2011), there was a decline in share repurchases in the U.S. during the financial crisis. The reason for this, is that strong firm performance and financial stability are important factors in periods of financial distress that could help firms to survive the crisis. In that case, it is more likely that only financially strong and larger firms will announce share repurchases during the crisis. The main research questions that will be answered by this study: 1. Does the European market respond differently to share repurchase announcements in the period from ? In addition, how do repurchase announcements affect share prices? 2. How do firm performance affect share prices? 17

18 A significant motive to announce share repurchase programs, is that managers might believe that their shares are undervalued by the market. In addition, share repurchase announcements could provide significant signals to the market. These signals might provide important information about the optimism of future firm performance to investors. Therefore, share prices could be positively affected after the share repurchase announcements, in that case we could observe abnormal returns. The signaling under valuation hypothesis might be related to the presence of information asymmetry. Higher market expectation of undervalued shares might lead to an increase in the level of information asymmetry between investors and managers. During the financial crisis, firms profitability and investments declined significantly due to the loss of consumers purchasing power (lower revenues) and lack of investors confidence, which resulted in lower share prices. According to Benmelech and Dlugosz (2010), credit ratings were not correctly determined at the beginning of the financial crisis and must therefore be downgraded. In addition, financial assets were mispriced, which increased the information asymmetry between insiders and outsiders in the firm. Moreover, investors might become uncertain about the determination of the firm s creditworthiness, this increases the level of information asymmetry. According to Peyer and Vermaelen (2009), markets could be in panic when bad news is announced. This could lead to a decrease in future earnings and credit rating downgrades. Therefore, it can be argued that the relevance of the signaling hypothesis is increased during the financial crisis, managers are more likely to repurchase shares when they are undervalued. The first hypothesis follows: Hypothesis 1: the signaling undervaluation hypothesis is important to explain the increase in abnormal returns in the period from Based on the previous literature, it could be assumed that the level of information asymmetry between the insiders and outsiders of the firm will be bigger for the smaller firms. The second hypothesis will be as followed: Hypothesis 2: smaller firms might have higher abnormal returns than large firms during share repurchase announcements, because of the increase in the level of information asymmetry. 18

19 Another important motive to repurchase shares is the free cash flow hypothesis. When firms reduce the amount of free cash flow at the management s level, then the agency costs will be lower and there might be an increase in the abnormal returns. However, firms might also have important motives to hold an excess amount of cash, which could lead to some benefits at the firm level. According to Opler, Pinkowitz, Stulz, and Williamson (1999), firms might have precautionary financial motives to hold excess amount of cash. In addition, firms might prefer to hold an excess amount of cash in the case of higher liquidity risk. When the firm s free cash flow is more exposed to risk, then the amount of excess cash that will be hold by the firm will increase to cover certain payments (Opler, Pinkowitz, Stulz, and Williamson, (1999)). When the firm s free cash flow is more volatile, especially during the financial crisis, the more amount of excess cash is needed by the firm to pay for certain expenses. This argument is valid for financial constrained firms, which are not able to use external financing to reduce liquidity risk. In that case we might expect that firms will demand for higher amount of excess cash when uncertainty is increased. In addition, the precautionary motive could lead to higher agency costs. Furthermore, firms might have lower earnings during the period of the financial crisis, therefore the amount of excess at management level might be significantly lower. The expectation is that the free cash flow hypothesis might have a weak explanatory power on the performance of abnormal returns. The third hypothesis follows: Hypothesis 3: The free cash flow hypothesis could be still important to explain the positive abnormal returns in the period from Finally, firms with higher leverage are more likely to have higher abnormal returns, because these firms have less free cash flow to waste by its management and will invest in projects with positive NPV to maximize firm value (Jensen, (1986)). The last hypothesis follows: Hypothesis 4: According to the Optimal Capital Hypothesis, high levered firms will have higher abnormal returns after share repurchase announcements in the period from

20 Table 2: an overview of the formulated hypotheses Hypothesis 1. The signaling undervaluation hypothesis is important to explain the increase in abnormal returns in the period from Motive Signaling undervaluation Effect on abnormal returns + 2. Smaller firms might have higher abnormal returns than large firms during share repurchase announcements, because of the increase in the level of information asymmetry. Signaling undervaluation + 3. The free cash flow hypothesis could be still important to explain the positive abnormal returns in the period from Free cash flow + 4. According to the Optimal Capital Hypothesis, high levered firms will have higher abnormal returns after share repurchase announcements in the period from Optimal capital structure + 20

21 Chapter 3: Data Data In this section, the details of the empirical tests will be given. First, it will explain how the data was collected and what criteria were used to select the sample. Second, the details regarding the regression analyses will be included and it will be explained how the regression analysis will be used to test the hypotheses. Finally, this chapter will provide descriptive data statistics used in the empirical tests and includes the correlation matrix of the independent variables. 3.1 Data collection In order to perform the empirical tests, a list of the FTSEurofirst 300 companies was required, obtained from the Thomson Reuters Database, which announced share repurchases in the period between 2005 and The FTSEurofirst 300 data is chosen in this study, to examine the effects of share repurchases of European firms after the buyback prohibition (before the late 90 s). The FTSEurofirst 300 index is a highly liquid market index that consists of the 300 largest companies ranked by its market capitalization and could be compared with the S&P 500 and consists of the 500 largest US firms. The data that is used in this paper, only includes large European firms, because these firms are more likely to adjust their pay out policy rather than to abolish it during the crisis period. In addition, larger European firms may have more diversified options of funding, for example: bonds, bank financing and (U.S.) private placement. However, smaller firms might have limited have limited resources and a limited reputation on the capital markets. Looking to Figure 3 (see Appendix), we could notice that the development of FTSEurofirst 300 index follows the same pattern as the S&P 500 index over the period ( ). For example, we see that both indices increase in the period At the beginning of 2008, we notice a significant decline in both indices at the start of the financial crisis. From , there is an upward trend in both indices. In addition, the FTSEurofirst 300 index outperforms the S&P 500 index in the period After this period, the S&P 500 started to increase in performance relative to the FTSEurofirst 300 index. Furthermore, Figure 4 (see Appendix) shows the development of the buyback yield index, the index performance is weighted by the product of the firm s market cap. weight and the buyback yield score. The index reflects the performance of the firms that return capital to its shareholders through share buybacks in Europe. 21

22 In addition, the MSCI Europe index (orange line) refers to the average market performance and we see that the repurchasing firms outperformed the average market performance over the period In addition, Figure 5 (See Appendix) is related to the US securities. This Figure shows the price movements in firms share prices over the period , which have bought back their shares in the last twelve months prior to the selection day. We notice a decline in share prices for the repurchasing firms at the beginning of From , we observe an upward trend in the yields for repurchasing firms. Comparing the figures for Europe and US, we notice that the European repurchasing firms have outperformed the US repurchasing firms over the period Furthermore, the list with share repurchase announcements was downloaded from the Zephyr Database and consists of companies of the developed Western European countries, which announced share repurchases between January 2005 and December In addition, the dataset includes Austria, Belgium, Denmark, France, Germany, Netherlands, Spain, Sweden, Switzerland, and UK. Next, the financial information related to the share prices and measures of firm performance as the Total assets, Market value to book, Long-term debt ratio, Return-on-assets, Return on equity and Cash flow to Sales are all obtained from DataStream. In addition, the share repurchase value is obtained from the Zephyr Database. In order to calculate the daily abnormal returns by the market model for the FTSEurofirst firms, this paper uses the FTSEurofirst 300 index as the benchmark market-index. In addition, the FTSEurofirst 300 index is obtained from the Thom Reuters Database. 3.2 Data clean up and sample selection In this paper, some firms from the FTSEurofirst 300 index were excluded in the case of missing information with regard to share prices and the specified independent variables, which are included in the sample. Starting with 179 firms from the list of the FTSEurofirst300 index, which have 5359 share repurchase announcements. The repurchase announcements include announced or completed share repurchase programs. Furthermore, for selecting the data, there are some criteria taken into account: All the financial information of the FTSEurofirst 300 active firms is obtained from the DataStream Database. Firms with no information about the share price or firm performance characteristics as defined in the methodology part are excluded from the sample. All the repurchasing firms in the sample are selected from the database Zephyr Database and the selected open-market share repurchase programs start from 2005 till

23 The selected amount of share repurchase included in the sample is at least 1%, because many firms have multiple events in a short time period and this could disturb the measuring the effect of abnormal performance in a specific period of time. In addition, the amount of repurchase differs a lot for each firm. Therefore, share repurchase announcements below 1% are removed, because it is more likely that these announcements will have less significant impact on the share price after the announcement. In addition, undefined amount of share repurchase is excluded from the sample, because this could make interpreting the results more difficult. Repurchase announcements that are rumored, withdrawn or waiting for shareholders approval are removed. This type of announcements might create uncertainty in the data. This could affect our results of the abnormal performance. The target firm in this sample will be defined as the parent company, so subsidiaries or trust funds are excluded from the sample. Including only the parent company will provide more consistency (similar type of firm) in the data. Subsequently, taken this set of assumptions into account: our sample consists of 125 firms including 316 share repurchase announcements. In addition, the dataset is fully controlled for missing values, so now we could start calculating the log-returns of the FTSE share prices and the FTSE index. The log-returns are calculated as follows: ri= ln(pt/pt-1), where Pt is defined as the closing price index on time t and Pt-1 is the closing price on time t-1. After the calculation of the log-returns, the concerning data will be used in Stata. To perform the regression analysis in Stata, it is important that the selected firms have (at least) an estimation window of -250 to -21 days. Furthermore, firms that have less data will be excluded. As result, the final sample will consist of 80 firms and 120 events. 3.3 Independent variables for performing a regression analysis In this section, a detailed description of the independent variables will be given. These specified variables will be included in the regression model. The independent variables could be seen as important determinants of firm performance. In addition, this study will examine the effect of the independent variables on the cumulative abnormal returns. It will look into how determinants of firm performance might affect cumulative abnormal returns. In order to specify the independent variables, we use definitions from DataStream and Zephyr Database. In addition, all the data for determining the different variables will be collected from DataStream. 23

24 Furthermore, this section includes the main drivers behind the abnormal performance after a share repurchase announcement. In addition, the significant repurchase motives and the corresponding variables will be given and the last section includes the descriptive statistics of the specified independent variables and the correlation matrix between the independent variables. Specification of the independent variables is as follows: 1. Total assets The variable Total assets defines the size of the firm and is determined by taking the natural logarithm of the total assets for each firm in the sample: Size of the firm = ln (Total assets) In addition, the firm size is related to the signaling undervaluation hypothesis. Undervaluation could be observed, when there is a big difference in the transparency of information between the managers (insiders) and investors (outsiders) of the firm. Therefore, firms might not be correctly valued. According to Ikkenberry and Vermaelen (1996), smaller firms are more likely to be undervalued, because larger firms provide more information to the public. So, smaller firms might have higher levels of information asymmetry than larger firms and are expected to have higher abnormal returns after share repurchase announcements. In that case, we might expect a negative relationship between firm size and the cumulative abnormal returns. 2. Market-to-book ratio (MVTB) The variable Market-to-book ratio could be defined as the market value of equity divided by the book value of equity. This variable is related to the signaling undervaluation hypothesis. The Market-to-book ratio could be defined as follows: Market-to-book ratio = The market value of firm s equity / the book value of firm s equity Which could be calculated as: (Book value per share + common shares outstanding) / (Market price per share + common shares outstanding) 24

25 According to Ikenberry, Lakonishok and Vermaelen (1995), firms with a lower MVTB ratio could observe higher abnormal returns after a share repurchase announcement, because a MVTB ratio below one indicates undervaluation. The Market-to-book ratio classifies two types of firms: growth firms and value firms. In addition, growth firms will have higher investment opportunities relative to value firms, therefore growth firms might have a higher market value of equity, this will lead to a higher Market-to-book ratio. Therefore, growth firms are less likely to distribute excess cash to their shareholders. However, value firms might have more incentives to payout their amount of excess cash to shareholders, because they might have less investment opportunities. So, we may expect a negative relationship between the Market-to-book ratio and the cumulative abnormal returns. 3. Long-term debt ratio (LTDebt) The variable Long-term debt ratio is defined as % of the Total capital: Long-term debt ratio = (Long-term debt/ Total capital) * 100 The Long-term debt ratio is related to the Optimal capital hypothesis and indicates how much of the firm assets is financed by debt. In addition, firms are willing to achieve an optimal leverage ratio, which could increase firm value. When a firm repurchases their shares back, the portion of equity in the firm will decrease and this increases the debt ratio. According to Vermaelen (1981), firms might also finance their share repurchase programs by issuing debt. In addition, firms will acquire a tax subsidy based on the deductibility of interest payments. This tax subsidy will benefit shareholders and this will lead to a higher share price. So, we may expect a positive relationship between the long-term debt ratio and the cumulative abnormal returns, in the case of positive market reaction and good investment strategy after a share repurchase announcement. 4. Return-on-assets (ROA) The variable Return-on-assets could be defined as the firm s profitability. The Return-on-assets is calculated as follows: Return-on assets = Net income/ Total assets 25

26 The share price impact could be more positive, when the firm announce a share repurchase after a profitable period. Therefore, managers could provide investors with relevant information about their allocation of assets. However, when firms have realized low profits in a specific period and these firm decide to invest in their production capacity, the result might be a negative ROA. Furthermore, profitable firms are less likely to distribute excess cash to their shareholders, because it will invest more in projects with a positive NPV. However, less profitable firms are more likely to distribute excess cash to their shareholders, because they might have less investment opportunities. So, we may expect a negative relationship between the Return-onassets and the cumulative abnormal returns. 5. Return-on-equity (ROE) The variable ROE could be defined as the net income divided by shareholder s equity. The ROE can be calculated as follows: ROE = Net income / Shareholder s equity The ROE is also related to the profitability of the firm and indicates that the share of profit is generated from the invested capital by the shareholders. This variable is related to the Free Cash Flow Hypothesis. In addition, a high ROE implies that the Net income relative to equity increases and that the firm s equity might be used in an efficient way, which will result in higher returns for shareholders. When firms decide to repurchase their shares by using their earnings, firm s Return-onequity will increase. In addition, the reinvestment of the earnings will only benefit shareholders when the ROE is high at that moment. Afterwards, the firm s outstanding equity will decrease after executing the share repurchase program. Therefore, it is more likely that firms with lower ROE will announce a share repurchase, because these firms might have lower reinvestment rates and are more expected to payout cash to their shareholders. So, we may expect a negative relationship between the ROE and the cumulative abnormal returns. 26

27 6. Operating Cash Flow-to-net sales ratio (Cash flow / Sales) The variable Operating Cash Flow to net sales defines the ability of the firm to generate cash from its sales. This variable is related to the Free Cash Flow Hypothesis. In addition, firms might repurchase shares to distribute excess cash to their shareholders and to limit the waste of money by firm s management and thus to lower the agency costs. Furthermore, the cash flow ratio will be for each firm calculated as follows: Cash flow- to-net sales ratio = (cash flow from operating activities / net sales)*100 According to Grullon and Michaely (2004), repurchasing firms are more likely to have higher levels of cash at their disposal, so we might expect a positive relationship between the operating cash flow-to- net sales ratio and the cumulative abnormal returns. 7. Repurchase ratio The repurchase ratio is defined as the size of the share repurchase program announced by the firm in period t and could be used as a control variable in the OLS regression model. It is expected that there is a positive relationship between the size of the repurchase program and the cumulative abnormal returns (CAR). In addition, the repurchase ratio is determined as a fraction of the total market capitalization of the firm. This variable can be calculated as follows: Repurchase ratio = Deal value of share repurchase program in a given period of time / Market value per company in a given period of time. 8. Crisis Dummy The variable Crisis dummy indicates the start of the financial crisis. This study uses the fall of Lehman Brothers as the beginning of the crisis, in 13 September In addition, the fall of Lehman Brothers caused worldwide big financial shocks in the financial markets. The Crisis dummy = 0 if the share repurchase announcement occurred before the financial crisis and 1 if the announcement was during the crisis period and 2 if the repurchase announcement was after the crisis period. 27

28 Table 3: The definitions of the independent variables that will be used to analyze the abnormal performance after share repurchase announcements in the period Variable Definition Hypothesis Effect The natural logarithm of the Signaling undervaluation Total Assets total assets Market-to- book ratio The market value of the common equity divided by the balance sheet common equity Signaling undervaluation Long-term debt ratio End-year's Long-term debt divided by the end-year's total capital Optimal Capital + Return on assets End-year's net income divided by the end-year's total assets Free cash flow Return on equity End-year's net income divided by the end-year's common equity Free cash flow Operating Cash flowto-sales ratio The end-year's operating cash flow divided by the end-year's firm's net sales Free cash flow + Repurchase ratio Deal value of share repurchase program in period t divided by the market value of the company in period t + Crisis Dummy The Crisis dummy is 1 if the share repurchase announcement occurred in the crisis period and 0 if the repurchase announcement was before the crisis period and 2 if the announcement was after the crisis period 28

29 3.4 Descriptive statistics Table 4 presents the descriptive statistics of the variables that will be used to perform the regression analysis. The calculation of the abnormal returns is based on 80 firms with 120 share repurchase announcements in the period In this study, Stata will be used in order to perform the regression analysis and to conduct the event study. Table 4: Descriptive statistics of the independent variables from 2005 to 2015 Eurofirst 300 firms Obs. Mean Median Std. Dev. Min Max Dependent variables CAR(-1, +1) % 0.78% 3.06% -7.49% 18.23% CAR(-3, +3) % 0.48% 4.81% % 31.82% CAR(-5, +5) % 0.47% 5.42% % 34.40% CAR(-20,+20) % 0.36% 10.93% % 38.74% Independent variables Total assets 225, MVTB 222, LTDebt 225, ROA 224, ROE 224, Cash Flow/ Sales 225, Share rep. ratio Table 4 includes the Mean, Median, Standard deviation and the minimum and the maximum value for the dependent and the independent variables. The number of observation for the cumulative abnormal returns refers to the total share repurchase announcements made by the firms listed on the FTSEurofirst 300 index that are included in the sample. In addition, the variable Total assets is the natural logarithm of the firm s total assets and indicates the size of the firm. Looking to Table 4, we notice that the mean of CAR decreases over the different event windows: (-1,+1), (-3,+3), (-5,+5) and (-20,+20). In addition, we notice a positive mean of 1.21% for CAR (-1,+1) and a negative mean of -1.07% for CAR (-20,+20). So, we observe higher positive cumulative abnormal returns surrounding the share repurchase announcement and this will decrease over time. Furthermore, it can be noticed that the number of observations is not the same for all the independent variables, because of some missing data points over the period

30 In addition, the variable share repurchase ratio has a lower number of observations, because some events do not include the deal value of the repurchase program. Finally, Table 4 provides a summary of the number of observations, Median, Std. Dev., Min. value and Max. value for all the variables included in the OLS regression model. Furthermore, Table 5 and 6 present an overview of the descriptive statistics for the repurchasing and non-repurchasing firms. Table 5: Overview of the descriptive statistics of the independent variables (Total assets, MVTB, LTDebt, ROA, ROE and Cash Flow/ Sales) for the non-repurchasing firms in the crisis period Crisis period Non repurchasing firms Obs. Mean Std. Dev. Min Max Independent variables ln Total assets 12,830 18,19 1,93 13,88 20,94 MVTB 12,830 2,74 4,36 0,23 29,77 LTDebt 12,830 43,58 1,65 10,79 81,94 ROA 12,830 3,67 4,93-8,13 24,35 ROE 12,830 12,17 20,97-47,24 98,67 Cash Flow/ Sales 12,830 13,27 7,08-2,71 29,06 Table 5 shows a summary of the descriptive statistics of the independent variables (Total assets, MVTB, LTDebt, ROA, ROE and Cash Flow/ Sales) related to the non-repurchasing firms in the crisis period. In addition, 10 firms that are listed on the FTSEurofirst 300 index, did not perform a share repurchase. Table 6: Overview of the descriptive statistics of the independent variables (Total assets, MVTB, LTDebt, ROA, ROE and Cash Flow/ Sales) for the repurchasing firms in the crisis period Crisis period Repurchasing firms Obs. Mean Std. Dev. Min Max Independent variables ln Total assets 102,640 17,00 1,64 13,95 21,51 MVTB 101,614 2,64 1,86 0,13 17,64 LTDebt 102,640 36,91 21,81 0,00 88,44 ROA 102,640 6,63 7,01-54,44 34,17 ROE 102,640 15,57 16,94-135,99 61,80 Cash Flow/ Sales 102,384 18,67 13,08-14,81 67,82 Table 6 shows a summary of the descriptive statistics of the independent variables (Total assets, MVTB, LTDebt, ROA, ROE and Cash Flow/ Sales) related to the repurchasing firms in the crisis period 30

31 As result, we could notice that the mean of the following variables: Total Assets and MVTB, was lower for the repurchasing firms compared to the non-repurchasing firms during the crisis period, which is in line with the undervaluation hypothesis. In addition, the mean of the variable Cash flow/sales was higher regarding to the repurchasing firms compared to the non-repurchasing firms, which is in line with the free cash flow hypothesis. Table 7: Correlation coefficients between the independent variables Firm size MVTB LTDebt ROA ROE Cash flow/sales Share rep. ratio Firm size 1 MVTB LTDebt ROA ROE Cash flow/sales Share rep ratio Table 7 summarizes the correlation coefficients between the independent variables used in the regression model. The variable Firm size refers to the firm s total assets Table 7 shows that some independent variables are highly correlated with each other. In addition, the variables ROA and ROE have a positive correlation of Further, the variables: MVTB, ROA and ROE are the most negatively correlated with the numbers: , and The correlation between the independent variables is relatively at a moderate level and the highest correlation value is

32 Chapter 4: Methodology Methodology This section will discuss the research methodology that will be used in this study. This study consists of three different parts this study will examine the effect of share prices after the announcement of share repurchases, therefore an event study analysis will be performed. According to Lasfer (2002) and Rau and Vermaelen (2002), to observe the effects of share repurchase announcements on share prices, it is important to study the share price behavior surrounding the announcement date. After determining the impact of share repurchase announcements on share prices, a multivariate regression analysis will be performed to examine the various drivers of abnormal returns behind the share price impact, by determining the effect of each independent variable. In order to perform an event study, different steps have to be taken. The following steps will be described by looking to the papers of Bowman (1983), De Jong and de Goeij (2011) about even studies methodology. 4.1 The firm s characteristics analysis In this section, we will look into specific firm characteristics that are important for analyzing, whether there are significant effects of abnormal performance after share repurchase announcements of the repurchasing firms for different periods: pre-crisis, during the crisis and post-crisis. The independent variables included in the regression model that will be analyzed are: Total assets, Book-to-market value, debt ratio, Operating Cash flow to sales ratio, Return on assets, Return on equity and Share repurchase value ratio. In addition, we include the Crisis dummy. Furthermore, Comment and Jarrell (1991) looked into a t-test on the difference on the mean, which will be used in this study to analyze the effect of each variable before and after the financial crisis. To perform a t-test, several steps have to be taken: To define t: t = X 1 X 2 σx 1 x 2 Further, we assume that: σx 1 x 2 = S²₁ n₁ +S²₂ n₂ To determine the number of degrees of freedom: ԁf = [(s²1 / n₁) + (s²₂ / n₂)] / [ S₁2 /n1 n₁ 1 + S₂2 /n₂ n₂ 1 ] 32

33 Furthermore, this analysis will be performed amongst all the repurchasing firms in the sample. 4.2 Research methodology: the event study In order to conduct an event study, this section will follow the event studies procedures (Bowman (1983), De Jong and de Goeij (2011)) to test for abnormal returns. There are different steps to follow for performing an event study analysis. In this study, we follow the three steps set by De Jong and de Goeij (2011). First, it is important to identify the event of interest and especially the timing of the share repurchase announcement. Therefore, it is important to give a clear definition of the event that is examined, so that the identified events are comparable. 2 Further, it is important to determine and include the specific time period of the event of interest over which the share prices are analyzed. In addition, to select the sample size, there are some important criteria that have to be satisfied for collecting the data. When a firm satisfies the criteria then it should be included. The criteria is discussed in paragraph 3.2 data section, which includes the requirements for including a firm and share repurchase announcement in the data sample. Second, we have to specify the benchmark model for determining the normal stock return behavior. In order to estimate the abnormal returns around the event date, it is important to estimate a benchmark model for the normal stock returns and to determine a good estimation window. According to MacKinlay (1997), the normal stock returns are defined as the predicted stock returns around the event date. The last step is to calculate and to perform an analysis for the abnormal returns around the share repurchase announcement. Furthermore, the abnormal performance will be statistically tested, to look whether the calculated abnormal returns are significantly different from zero. Afterwards, a multivariate regression analysis will be performed on the cumulative abnormal returns (CAR) in different event windows: (-1,+1), (-3,+3), (-5,+5) and (-20,+20). So, that we could examine the main drivers behind the abnormal returns after the share repurchase announcement Identification of the event and the estimation window

34 Starting with identifying the event of interest, which provides more detail about the type of share repurchase announcement that is used. This study will use event dates of completed and not completed share repurchase programs, only the rumored repurchase announcements were removed. Rumored repurchase announcements might lead to more uncertainty, which could lead to disturbing findings in the analysis of abnormal returns. However, Simkovic (2008) found evidence that share repurchase announcements do not always lead to actual share repurchases. Sometimes actual repurchases are not fully completed in the short term period after the repurchase announcement. In addition, investors could react late on news about share repurchase announcements, because of the presence of information asymmetry in the case of incomplete markets. Therefore, it is important to look to the abnormal returns that are close to the event date (t=0). According to Lie (2005), the positive impact on share prices after share repurchase announcements is often a temporary effect, so we take this into account for determining the event windows. Therefore, the event window will go from (- 1,+1) to (-20,+20). In order to choose the length of the estimation window, this study follows the guidelines of Brown and Warner (1985). The estimation window starts from 250 to 21 days prior (-250,-21) to the event date (t=0). According to Rasbrant (2011), a large estimation window might be useful to have a normal distribution approach of the t-test statistics. This study will examine further the calculation of the normal performance of the abnormal returns Measuring the normal performance of abnormal returns To find the normal performance of abnormal returns, first it is important to calculate the predicted normal stock returns. According to De Jong and de Goeij (2011), the choice of the benchmark model is significant for conducting an event study to determine the stock return behavior. In order to estimate the normal abnormal returns, the market model will be used. It is important to run a regression for each event of interest (repurchase announcement date). The estimation window starts from -21 to -250 (event date: t=0). As result from the regression, the alpha and beta will be determined. The calculated alpha and beta will be collected and will be used for predicting the normal returns during the following event windows: (-1,+1), (-3,+3), (- 5,+5) and (-20,+20). This study will define the abnormal returns as the residuals of the Market model: 34

35 Where is the return on period t for security i and is the chosen market portfolio. In this research, the FTSEurofirst300 index will be used as the market index. In addition, the error term is the zero mean disturbance term (MacKinlay, 1997). An alternative model to the market model is the CAPM model, where the excess returns are defined as: According to MacKinlay (1997), the abnormal returns under the market model will be jointly normally distributed. To perform a regression analysis of the market model, the dependent variable is defined as the natural log of the daily returns of the Eurofirst 300 firms, which is calculated by subtracting the daily share prices from each other. And the independent variable in this model, is defined as the natural log of the market return index by subtracting the daily market price index. The parameters of this model are:, and The abnormal returns will be defined as the residuals of the Market model: Where and are defined as the OLS estimates of the regression coefficients Model for the abnormal returns and cumulative abnormal returns According to De Jong and de Goeij (2011), the abnormal returns (AR) could be defined as the actual company s stock return (at the event date) subtracting the benchmark return or normal return (NR): Where is the actual return on the event date of company i at time period t. In addition, is the (benchmark) normal return of company i at time period t. In order to determine the normal stock returns, it is important to take into account the estimation period. The estimation period is prior to the event date and the time period t is the number of periods (days or months) from the event and is not the same as the calendar time (De Jong and de Goeij, (2011)). In addition, the average abnormal returns could also be calculated to improve the informativeness related to the share price movements around repurchase announcements (De Jong, (2007)). 35

36 The unweighted cross-sectional average of abnormal returns in period t is as follows: When the average abnormal returns deviate a lot from zero, it provides the evidence of abnormal performance. In that case the abnormal returns will be concentrated around the specific event and the average abnormal returns will reflect the effect of this specific event. All information that is not related to the event will be eliminated on average. In this study, we are also interested in the long-term performance of abnormal returns. This could be examined by using cumulative abnormal returns (CAR), which measures the performance of the cumulative abnormal returns in the period around the event date. In order to calculate cumulative abnormal returns, we have to sum up the abnormal returns within the event window (MacKinlay, (1997)). The share price effect will be tested prior and after the event date, in different event windows: (-1,+1), (-3,+3), (-5,+5) and (-20,+20). The cumulative abnormal returns in period t is as follows: CARt = In addition, the cumulative abnormal returns (CAR) are aggregated from the beginning of the event period t1 and ending at t2 (De Jong and de Goeij, (2011)). In that case, we obtain the cumulative average abnormal return (CAAR) in period t and is calculated as follows: Finally, statistical t-tests will be performed to test the significance of abnormal returns surrounding the repurchase announcement date. These tests might improve the drawing of conclusions about the effect of share price surrounding the event dates. 36

37 4.2.4 Testing the abnormal returns performance In order to test the significance of the abnormal returns surrounding the event date (t=0), it is important to calculate the abnormal performance in different event windows. In addition, a comparison between the period prior to the event and the period after the repurchase announcement is needed to observe the change in share prices over time. According to de Jong (2007), results from the statistical tests will provide more evidence whether the performance of the abnormal returns is significantly different from zero, hence the null hypothesis could be rejected. The null hypothesis is defined as follows: In the case of testing the null hypothesis, the t-statistics follows a student distribution with N-1 degrees of freedom. According to De Jong and de Goeij, (2011), this student t-test assumes that the average abnormal returns are independently and identically distributed. In essence, is an unknown parameter, therefore the estimator s could be used for calculating the of CAAR and could be constructed from the cross-sectional variance of the Abnormal returns in period t: Testing whether the cumulative average abnormal returns are significantly different from zero, the t-statistics could be calculated as follows: 4.3 Performing a regression analysis This study will perform a multivariate OLS regression analysis to examine the effect of share repurchase announcements. This is important to define the drivers behind the abnormal performance. In addition, the cumulative abnormal return is classified as the dependent variable (CAR) and the independent variables are specified in the Data Section 3.3. CARit= αi,t + β1ln (Total assets) + β2 MVTB+ β3ltdebt+ β4roa + β5roe+ β6fcf/sales+ β7rep. Ratio + β8crisisdumi (Di) + εi 37

38 In addition, we will define the Crisis dummy as Di, where Di=0,1,2 Performing a regression analysis over different periods from : Pre-crisis regression ( ): CARit= αi,t + β1ln (Total assets) + β2 MVTB+ β3ltdebt+ β4roa + β5roe+ β6fcf/sales+ β7rep. Ratio + β8 (0) + εi, where Di=0 During crisis ( ): CARit= αi,t + β1ln (Total assets) + β2 MVTB+ β3ltdebt+ β4roa + β5roe+ β6fcf/sales+ β7rep. Ratio + β8 (1) + εi, where Di=1 Post-crisis ( ): CARit= αi,t + β1ln (Total assets) + β2 MVTB+ β3ltdebt+ β4roa + β5roe+ β6fcf/sales+ β7rep. Ratio + β8 (2) + εi, where Di=2 38

39 Chapter 5: Results and Discussion Results This chapter will provide us with the empirical results from the conducting event study and includes the results from the regression analysis. The discussion of the results will be included at the end of this chapter. First, it discusses the analysis of the share price behavior for all the firms that are included in the sample prior, during and after the share repurchase announcement within the event window of 41 days. In addition, the results from the cumulative average abnormal returns (CAAR) will also be included. Furthermore, it provides an overview of the results from the OLS regression model, which includes the regarding independent and the dependent variables. Finally, these results will be used to test the hypotheses as stated in the literature review section Results from the event study In this section, we look into the share price impact as a result of the share repurchase announcement. In addition, the event study results will be presented in table 8 and 9 from It also includes the development of the average abnormal returns (see figure 1) and the development of the cumulative average abnormal returns (see figure 2). The results of AAR and CAAR during the crisis period ( ) will also be included in this section (see Table 10). Furthermore, the results from the performed event study will be used to answer the first hypothesis related to the signaling undervaluation hypothesis. Table 8 includes the results of the average abnormal returns (AAR) and the cumulative average abnormal returns (CAAR) for 20 days prior and 20 days after the share repurchase announcement (event day t=0). The result is an average abnormal return of 1.67% at the event day (at the repurchase announcement), which is positively significant at the 1% level with a t- statistic of Furthermore, the cumulative average abnormal returns is 1.96% in the event window (-1,+1), which is sufficient compared to the empirical results from other papers discussed in the literature review section 2.2. The cumulative abnormal returns for U.S. firms varies between 1.5% to 3.5%. The abnormal returns in the period could be higher, but many investors lost their confidence in the market during the crisis. In addition, the financial crisis started after the fall of Lehman brothers in

40 Table 8: The average abnormal returns and cumulative average abnormal returns in the event window (- 20,+20) over the period Day AAR CAAR t statistic Day AAR CAAR t statistic ** * * * * * *** ** * *** Table 8 shows the average abnormal returns (AAR) and the cumulative abnormal returns (CAAR) over the period varying from 20 days before until 20 days after the share repurchase announcement. Day 0 indicates the event day (t=0). In addition, the t-statistic for all the days are included (***, **, * significantly different from zero at the 1%, 5% and 10% level respectively) It can be noticed that the abnormal returns become more negatively significant in the last days (t=-10, t=-9, t=-8) prior to the event (repurchase announcement), which indicates undervaluation of the shares. At the event day (t=0), we still observe a positively significant AAR of 1.67% at the 1% level. After the event date, we notice negatively significant average abnormal returns in the trading days (t=3 and t=4, significant at the 10% and 5% level) and in the trading days (t=15 and t=17, significant at the 1% level). 40

41 Furthermore, Figure 1 and 2 show the development of the average abnormal returns (AAR) and the cumulative average abnormal returns (CAAR) over the event window of 41 days surrounding the share repurchase announcement. Looking to figure 1, we see explicitly a positive share price reaction surrounding the repurchase announcement, where the average abnormal return is increasing during the share repurchase announcement period. After the share repurchase announcements, we notice a sharp decline in the average abnormal returns. At least a couple days from the event day, the average abnormal returns start to become relatively constant. Figure 1 shows the average abnormal returns over the event window 20 days prior until 20 days after the share repurchase announcement (t=0). In addition, the x-axis represents the period of time in days and the y-axis represents the average abnormal returns scale. Looking to figure 2, we could see a clear upward trend of the cumulative average abnormal returns at the share repurchase announcement (t=0). Afterwards, the cumulative abnormal returns become relatively constant and then it starts to decrease after a couple days. 41

42 Figure 2 shows the cumulative average abnormal returns over the event window 20 days prior until 20 days after the share repurchase announcement (t=0). In addition, the x-axis represents the period of time in days and the y- axis represents the cumulative average abnormal returns scale. Table 9 shows the cumulative average abnormal returns (CAAR) for different event windows from The cumulative average abnormal returns in the event windows: (0), (-1,+1) and (-3,+3) are positively significant at the 1% level. Table 9: The cumulative average abnormal returns in different event windows from Event window (0) (-1,+1) (-3,+3) (-5,+5) (-20,+20) CAAR(%) 1.67*** 1.96*** 1.83*** 0.86** T-statistic P-value Table 9 shows the cumulative abnormal returns (CAAR) for all 80 firms and the total share repurchase announcements of 120 with the corresponding T-statistic and P-values over the different event windows from (0,0) to (-20,+20) (***, **, * significantly different from zero at the 1%, 5% and 10% level respectively) In addition, the CAAR in the event window (-5,+5) is positively significant at the 5% level. However, the CAAR in the event window (-20,+20) has a negative sign and is not statistically significant. So, CAAR is positively significant surrounding the share repurchase announcement (t=0). 42

43 Table 10: The average abnormal returns and cumulative average abnormal returns in the event window (-20,+20) over the period (crisis period) Day AAR CAAR t statistic Day AAR CAAR t statistic -20-0,0018-0,0018-0, ,0014-0,0558 0, ,0101-0,0119-2,6421 *** 19 0,0017-0,0572 0, ,0064-0,0183-1,6308 * 18 0,0007-0,0589 0, ,0009-0,0174 0, ,0022-0,0596-0, ,0022-0,0152 0, ,0054-0,0574 0, ,0076-0,0228-1, ,0065-0,0629-1, ,0039-0,0266-0, ,0059-0,0564-1, ,0015-0,0251 0, ,0038-0,0505-0, ,0039-0,0212 0, ,0052-0,0467 1,6651 * -11-0,0005-0,0217-0, ,0043-0,0520-1, ,0023-0,0194 0, ,0014-0,0476-0, ,0045-0,0239-1, ,0013-0,0463 0, ,0128-0,0367-3,3427 *** 8-0,0011-0,0475-0, ,0040-0,0327 0, ,0001-0,0464-0, ,0070-0,0397-1,7921 * 6-0,0010-0,0463-0, ,0010-0,0408-0, ,0065-0,0453-1, ,0051-0,0459-0, ,0094-0,0388-2,1496 ** -3-0,0017-0,0476-0, ,0081-0,0294-2,3146 ** -2 0,0010-0,0466 0, ,0006-0,0213 0, ,0004-0,0470-0, ,0011-0,0219 0, ,0240-0,0230 6,3987 *** Table 10 shows the average abnormal returns (AAR) and the cumulative abnormal returns (CAAR) over the period varying from 20 days before until 20 days after the share repurchase announcement. Day 0 indicates the event day (t=0). In addition, the t-statistic for all the days are included (***, **, * significantly different from zero at the 1%, 5% and 10% level respectively) Looking to Table 10, it presents the results of the average abnormal returns (AAR) and the cumulative average abnormal returns (CAAR) for 20 days prior and 20 days after the share repurchase announcement (event day t=0) during the crisis. The result is an average abnormal return of 2.40% at the event day (at the repurchase announcement), which is positively significant at the 1% level with a t-statistic of Furthermore, the cumulative average abnormal returns (CAAR) is 2.47% in the event window (-1,+1), which is still sufficient compared to the empirical results from other papers discussed in the literature review section 2.2., where the CAR for U.S. firms varies between 1.5% to 3.5%. However, we could still observe higher abnormal returns during the crisis, but many investors have lost their confidence in the market after the fall of Lehman Brothers. 43

44 In addition, it can be noticed that the average abnormal returns are statistically negatively significant at the 1% level in day (t=-8) prior to the event (repurchase announcement), which indicates undervaluation of the shares. Comparing the results between both periods ( and ), we observe that the trading days (t=-19, t=-18, t=-8) show for both periods statistically significant negative abnormal returns prior to the event day (t=0). At the event day, we observe for both periods positively significant abnormal returns at the 1% level. In addition, the trading days (t=3 and t=4) show both negatively significant abnormal returns after the repurchase announcement. In addition, the development of the AAR and CAAR during the crisis period follows the same pattern as over the period (see Appendix, Figure 6 and 7). Finally, the results from the event study show clear signs of undervaluation in the period before the event date (t=0), where AAR and CAAR are increasing surrounding the repurchase announcement. This is in line with hypothesis 1, according to Vermaelen (1981), firms might pay a premium to repurchase their own shares. 5.2 Results from the OLS regression analysis In this section, the results from the OLS regression will be provided in different event windows: (-1,+1), (-3,+3) and (-5,+5). In addition, this section consists of three regression analysis performed in STATA. The first regression analysis (see Table 11) includes the independent variables: Total Assets, MVTB, LTDebt, ROA, ROE and Cash flow/ Sales and CAR as dependent variable. The variable Repurchase ratio is excluded in all regressions, because including this variable in the regression analysis causes multicollinearity. The second regression model excludes the variable MVTB (see Table 12), which improves the significance of the constant term that is included in the regression model. 44

45 Table 11: The OLS regression results of CAR (-1,+1), (-3,+3) and (-5,+5) Variables CAR(-1.+1) CAR(-3.+3) CAR(-5.+5) Constant *** (0.0217) (0.0225) (0.0201) Total Assets *** (0.0012) (0.0013) (0.0011) MVTB *** *** *** (0.0011) (0.0011) (0.0010) LTDebt ** (0.0001) (0.0001) (0.0001) ROA *** (0.0006) (0.0006) (0.0005) ROE *** *** (0.0002) (0.0002) (0.0002) Cash flow/sales (0.0001) (0.0001) (0.0001) ** R Adj-R F-stat Significance F N ,298 Table 11 shows the OLS regression results from over different event windows: (-1,+1), (-3,+3) and (-5,+5). The numbers between brackets refer to the standard errors rounded off to four decimals (***, **, * significantly different from zero at the 1%, 5% and 10% level respectively) 45

46 The results from the first regression analysis show that the variable Total Assets is negatively significant at the 1% level in the event window (-5,+5). In addition, the variable MVTB is positively significant at the 1% level for all the event windows: (-1,+1), (-3,+3) and (-5,+5). The variable LTDebt is positively significant at the 5% level in the event window (-3,+3) and the variable ROA is negatively significant at the 1% level in the event window (-5,+5). Furthermore, the variable ROE is negatively significant at the 1% level in the event window (-1,+1) and (-3,+3). The variable Cash flow/sales is positively significant at the 5% level in the event window (-5,+5) and the constant term is positively significant at the 1% level in the event window (-5,+5). These results are mainly consistent with the formulated hypotheses. First, the variable Total assets shows a statistically significant negative relationship with the abnormal returns, which supports the signaling hypothesis. The second variable MVTB shows a statistically significant positive effect on the abnormal returns. According to Ikenberry, Lakonishok and Vermaelen (1995), the market-to-book ratio could be a proxy for overvaluation and should be negatively correlated with CAR. However, a positive coefficient of the MVTB suggests from management perspective a value-destroying policy to buy out corporate shareholders. The third variable LTDebt shows a statistically significant positive relationship with the abnormal returns, which supports the optimal capital hypothesis as described in section 2.5. The next variables ROE and ROA are statistically significant and have a negative relationship with the abnormal returns, which is related to the free cash flow hypothesis. And the last variable cash flow to sales shows a positive statistically significant effect on the abnormal returns, which is in line with the free cash flow hypothesis. Furthermore, the second regression analysis (which excludes the variable MVTB), is performed to improve the significance of the constant term. As result from this, the constant term becomes more positively significant over the different event windows. The results from the second regression analysis are shown in Table

47 Table 12: The OLS regression results of CAR (-1,+1), (-3,+3) and (-5,+5) Variables CAR(-1,+1) CAR(-3,+3) CAR(-5,+5) Constant * ** *** (0.0212) (0.0219) (0.0196) Total Assets * *** (0.0012) (0.0012) (0.0011) LTDebt *** (0.0001) (0.0001) (0.0001) ROA * * (0.0005) (0.0006) (0.0005) ROE *** *** (0.0002) (0.0002) (0.0002) Cash flow/sales (0.0001) (0.0001) (0.0001) R Adj-R F-stat Significance F N ,309 Table 12 shows the OLS regression results (excluding MVTB) from over different event windows: (-1,+1), (-3,+3) and (-5,+5). The numbers between brackets refer to the standard errors rounded off to four decimals (***, **, * significantly different from zero at the 1%, 5% and 10% level respectively) Looking to Table 12, we see that the constant term is positively significant at the 10% level in the event window (-1,+1), positively significant at the 5% level in the event window (-3,+3) and positively significant at the 1% level in the event window (-5,+5). Improving the significance of the constant term provides more reliable OLS regression results. 47

48 In addition, the variable Total Assets is negatively significant at the 10% level in the event window (-3,+3) and is also negatively significant at the 1% level in the event window (-5,+5). The variable LTDebt is positively significant at the 1% level in the event window (-3,+3) and the variable ROA is positively significant at the 1% level in the event window (-1,+1) and (-3,+3). In addition, the variable ROE is negatively significant at the 1% level in the event window (-1,+1) and (-3,+3). Based on the results presented in Table 11 and 12, the formulated hypotheses are confirmed excluding the MVTB variable. The Market value-to-book shows a positive relationship with the cumulative abnormal returns (CAR), which means that an increase of the MVTB leads to an increase of CAR. According to Liu and Swanson (2016), managers are more likely to increase the number of share repurchases when the market index increases. This suggests that firm s stock price increases with the market index than acquiring the corporate shares at low prices. Evidence found that managers repurchase shares when the stock returns are significantly decreased and are more likely to decline further which decreases the profitability in the market, so firms might overpay for acquiring its own equity. Successful stock price support means that the price paid in period t is below the stock price observed in the following time period (t+1). However, the empirical findings show support for overvalued equity (Liu and Swanson, (2016)). The third regression analysis includes the results from three OLS regressions for the different periods: pre-crisis, during the crisis and post-crisis, over the event windows: (-1,+1), (-3,+3) and (-5,+5). The results are presented in Table: 13, 14 and

49 Table 13: Results from the OLS regression over the event windows: (-1,+1), (-3,+3) and (-5,+5) in the precrisis period ( ) Pre-crisis period: Variables CAR(-1,+1) CAR(-3,+3) CAR(-5,+5) Constant *** *** *** (0.0312) (0.0267) (0.0240) Total Assets *** *** (0.0016) (0.0014) (0.0013) MVTB *** *** (0.0014) (0.0012) (0.0011) LTDebt (0.0001) (0.0001) (0.0001) ROA *** *** (0.0007) (0.0006) (0.0005) ROE *** *** * (0.0003) (0.0002) (0.0002) Cash flow/sales (0.0001) (0.0001) (0.0001) *** R Adj-R F-stat Significance F N Table 13 shows the results from the OLS regression in the pre-crisis period. The number between the brackets refers to the standard error. In addition, the coefficients and standard errors are rounded off to four decimals (***, **, * significantly different from zero at the 1%, 5% and 10% level respectively) 49

50 Looking to the results presented in Table 13, it can be noticed that the variables Total Assets, ROA and ROE are negatively statistically significant mainly over the event windows (-3,+3) and (- 5,+5) and is in line with the expectation in the pre-crisis period. Table 14: Results from the OLS regression over the event windows: (-1,+1), (-3,+3) and (-5,+5) in the crisis period ( ) Crisis period: Variables CAR(-1,+1) CAR(-3,+3) CAR(-5,+5) Constant *** Total Assets *** MVTB LTDebt * *** ROA * * ROE *** *** Cash flow/sales R Adj-R F-stat Significance F N Table 14 shows the results from the OLS regression in the crisis period. The number between the brackets refers to the standard error. In addition, the coefficients and standard errors are rounded off to four decimals (***, **, * significantly different from zero at the 1%, 5% and 10% level respectively) 50

51 Furthermore, Table 14 shows that the variables Total Assets (-1,+1) and ROE (-1,+1) and (- 3,+3) are negatively statistically significant during the crisis period and is in line with the expectation. In addition, the variable LTDebt (-1,+1) and (-3,+3) is positively statistically significant and is in line with the expectation in the crisis period. Table 15: Results from the OLS regression over the event windows: (-1,+1), (-3,+3) and (-5,+5) in the postcrisis period ( ) Post-crisis period: Variables CAR(-1,+1) CAR(-3,+3) CAR(-5,+5) Constant *** Total Assets MVTB *** * *** LTDebt *** * ** ROA *** *** ROE Cash flow/sales ** *** R Adj-R F-stat Significance F N (***, **, * significantly different from zero at the 1%, 5% and 10% level respectively) 51

52 In addition, Table 15 shows that the variables ROA is negatively statistically significant over the event windows (-1,+1) and (-5,+5). This is in line with the expectation during the post-crisis period. The variable Cash flow/sales is positively statistically significant and is in line with the expectation in the post-crisis period and supports the free cash flow hypothesis. Table 16 shows the regression results related to the average abnormal return (AAR), the cumulative abnormal return over 3-days (CAAR 3-Day) and the cumulative abnormal return over 7-days (CAAR 7-Day) regarding to the non-repurchasing firms over the period Table 16: The average abnormal return (AAR), the cumulative abnormal return over 3-days (CAAR 3-Day) and the cumulative abnormal return over 7-days (CAAR 7-Day) regarding to the non-repurchasing firms over the period Variable AAR CAAR 3-Days CAAR 7-Days Coef % -0.02% -0.05% (0.0001) (0.0004) (0.0008) T-statistic P-value R Adj-R F-stat Significance F... N 2, Table 16 shows the test statistics for the variables: AAR, CAAR 3-Day and CAAR 7-Day, over the period The number between the brackets refers to the standard errors and is rounded off to four decimals Based on the results from Table 16, it could be noticed that the AAR, CAAR 3-Day and CAAR 7-Day have no significant results over the period , which means that only repurchasing firms have experienced significant abnormal performance over the period This is in line with the expectation. 52

53 Discussion This study examined the effects of share repurchase announcements of the sample firms on the movement of share prices in the period , by using an event study methodology and performing an OLS regression analysis. The dependent variable in the regression model is the cumulative abnormal return (CAR) within the event windows: (-1,+1), (-3,+3) and (-5,+5). In addition, the regression model includes: Total Assets, MVTB, LTDebt, ROA, ROE, Cash flow/sales. The variable Share rep. ratio is excluded from the regression model, because this variable causes multicollinearity. The sample consists of firms listed on the FTSEurofirst 300 index. This section will include the answers on the main research questions that are formulated earlier in the introduction part. 1. Does the European market respond differently to share repurchase announcements in the period ? And how are repurchase announcements affect share prices? This study examined the share price impact of share repurchases announced by the FTSEurofirst 300 firms in the period (including the crisis period ). This paper found evidence of a statistically significant positive share price reaction (at the 1% level) after the share repurchase announcement at the event date (t=0) for the period This supports the empirical findings from the previous studies (e.g. Vermaelen (1981); Grullon and Michaely (2004)), where share repurchase announcements are used to signal positive news about the firm s future performance to the market and show positively significant abnormal returns. The empirical results for the European firms presented in this paper, are sufficient compared to the empirical results from other studies where the cumulative abnormal returns for U.S. firms varied between 1.5% and 3.5% in the event window (-1,+1). In addition, the signaling undervaluation hypothesis, the free cash flow and optimal capital hypothesis could be still seen as important motives to perform a share repurchase. 2. How do firm performance affect share prices? For performing the event study and the OLS regression, the sample consists of 120 share repurchase announcements done by 80 firms. As result, this study found an average abnormal return (AAR) on the share repurchase announcement day (t=0) of 1.67% positively statistically significant at the 1% level over the period Furthermore, the cumulative average abnormal return (CAAR) for the event window (-1,+1) is 1.96%, which is positively statistically significant at the 1% level. Including the results during the crisis period ( ), we observe a statistically significant positive AAR of 2.40% (at the 1% level) and CAAR of 2.47 over the event window (-1,+1). The results are sufficient compared to the evidence from other papers, where the cumulative abnormal returns (CAR) within the same event window (-1,+1) for US firms varied between 1.5% and 3.5%. However, the results compared to the CAR (-1,+1) for German firms are 53

54 relatively low, which was 3.55%. This could be explained as result of the financial crisis that started in Looking to Table 8, we could notice a statistically significant negative cumulative average abnormal return (CAAR) for some trading days (day -19, day -18, day -10, day -9 and day -8) in the period prior to the share repurchase announcement. The abnormal returns are negatively statistically significant at the 1% level (day -8) prior to the share repurchase announcement. After the share repurchase announcement (t=0), the average abnormal returns become not statistically significant in the post-announcement period (t=1 and t=2). This could indicate undervaluation, which means that managers are capable to time share repurchase announcements. In addition, the firm announces a share repurchase to give a signal to the market that its shares are undervalued. Furthermore, the results of AAR and CAAR during the crisis period (see Table 10), support the signals of undervaluation found in Table 8. The results from the event study are consistent with other papers, firms experience a positively statistically significant effect on the share price and firm value at the announcement day (t=0). From the firm and market perspective, there could be a fair valuation of the shares and is consisted with the results of Vermaelen (1981). In order to examine the share price reaction after a share repurchase announcement, this paper have performed an event study analysis and OLS regression analysis on the cumulative abnormal returns (CAR) for different event windows: (-1,+1), (-3,+3) and (-5,+5). The regression model includes CAR as dependent variable and Total Assets, MVTB, LTDebt, ROA, ROE and Cash flow/sales, as independent variables. In addition, the variable Total Assets refers to firm size in this study and the expectation is that smaller firms will have higher levels of information asymmetry than larger firms. So, it is more likely that smaller firms are undervalued. As result, the variable Total Assets shows a negative relationship with CAR for the different event windows: (-1,+1), (-3,+3) and (-5,+5), which means that an increase of the firm s size will lead to significant lower abnormal returns. This is in line with the undervaluation hypothesis. The following variable LTDebt is an indicator for the optimal capital hypothesis. It was expected that firms with higher leverage are more likely to have higher expected abnormal returns, because these firms will invest in positive NPV projects to maximize firm value (Jensen, (1986)). As result, the variable LTDebt shows a positively statistically significant relationship with CAR (-3,+3) at the 1% and 5% level in the period This is in line with the optimal capital hypothesis. 54

55 The variable MVTB is an indicator for testing the signaling undervaluation hypothesis. It was expected that firms with a low MVTB are more likely to repurchase shares to signal positive news to the market that their shares are undervalued. However, the results from the OLS regression analysis show that MVTB has a positively statistically significant relationship with CAR: (-1,+1), (- 3,+3) and (-5,+5). The reason for this could be that the firms that performed a share repurchase were mainly overvalued over the period Looking to Table 14, we observe a negatively relationship (not significant) with CAR (-5,+5) during the financial crisis ( ). This is in line with the undervaluation hypothesis. Looking to the variables ROE and ROA as indicators for the profitability of the firm, these variables show a negative relationship with CAR for different event windows: (-1,+1), (-3,+3) and (-5,+5). It was expected, that higher profitable firms are less likely to use excess cash for repurchasing shares, because profitable firms have more investment opportunities than less profitable firms. So, profitable firms will invest the excess cash rather than distributing it to its shareholders. The results are consistent with the hypothesis. The following variable Cash flow/sales shows a positive statistically significant relationship with CAR (-5,+5) at the 5% level in the period (see Table 11), which is related to the free cash flow hypothesis. This was expected, because it is more likely that firms with higher levels of excess cash will announce a share repurchase, to mitigate the waste of money by the firm s management and leads to lower agency costs. 55

56 Chapter 6: Conclusion Conclusion 6.1 Conclusion This study founds a statistically significant positive share price reaction after a share repurchase announcement over the period , which is in line with previous studies. In order to test the relationship between a share repurchase announcement and the share price reaction, we have included some control variables in the regression model: Total Assets, MVTB, LTDebt, ROA, ROE, Cash flow/sales and Share rep. ratio. However, the variable Share rep. ratio was not in line with the expectation, as it caused multicollinearity in the regression analysis. Therefore it was excluded from the regression model. Based on the performed OLS regression over the period with different event windows: (-1,+1), (-3,+3) and (-5,+5), the other variables: Total Assets, LTDebt, ROA, ROE and Cash flow/sales are statistically significant and in line with the expectation. In addition, looking to the OLS regression results from , the variable MVTB was not in line with the expectation. However, when we only look to the results during the crisis period ( ), we find a negatively (not significant) effect of MVTB on CAR, which is in line with the expectation. We can conclude that the results from the event study as the results from the OLS regression analysis are mainly consistent with the results provided in other studies. 6.2 Limitations and future research This section will include the limitations and some possible suggestions for future research. First, many firms listed on the FTSEurofirst 300 index performed multiple share repurchase announcements. Some events took in a short period after each other, where the share repurchase announcements with less days within the estimation window of (-250,-21) were filtered out. In that case, only the first event was included. In addition, the data sample includes only active firms, which is related to the survivorship bias. Furthermore, all the data about the independent variables and dependent variables included over the period , is collected from Thomson Reuters DataStream. Firms that do not provide data on a daily basis or insufficient data were excluded from the sample. As result, the sample size in this paper was relatively low, which consisted of 80 firms with 120 share repurchase announcements. For future research it could be good to expand the sample size and to focus also on the long term share price impact after a share repurchase announcement. Previous studies found higher levels of CAR (Ikenberry et al. (1995 and 2000)) in the long-term period. In addition, firms might have different motives to announce a share repurchase, analyzing these motives for the sample firms could lead to other new variables that are useful to include in the regression analysis. 56

57 References References Andres, C., Betzer, A., Doument, M., & Theisse, E. (2013), Open Market Share Repurchases in Germany: A Conditional Event Study Approach, Center for Financial Research, 13(2). Babenko, I., (2009), Share repurchases and pay-performance sensitivity of employee compensation contracts. J. Finance, 64, Babenko, I., Tserlukevich, Y., and Vedrashko, A., The Journal of Financial and Quantitative Analysis, Vol. 47, No. 5 (OCTOBER 2012), pp Bagwell, L.S., Shoven, J.B., (1989), Cash Distributions to Shareholders, Journal of Economic Perspectives, Vol.3, pp Barclay, M. J., & Smith Jr, C. W. (1988). Corporate payout policy: Cash dividends versus openmarket repurchases. Journal of Financial Economics, 22(1), Benmelech, E., & Dlugosz, J.,(2010), The Credit Rating Crisis, National Bureau Of Economic Research, 24, Berk, Jonathan B., Richard C., Green, and Vasant Naik, (1999), Optimal investment, growth options, and security returns, Journal of Finance, 54, Bhattacharya, S., (1979), Imperfect Information, Dividend Policy, and "The Bird in the Hand" Fallacy, The Bell Journal of Economics 10, pp Boehme, Rodney, and Sorin M. Sorescu, (2002), The long-run performance following dividend initiations and resumptions: Underreaction or product of chance? Journal of Finance, 57, Bowman, R. G. (1983). Understanding and conducting event studies. Journal of Business Finance & Accounting, 10(4), Brown, S. J., & Warner, J. B. (1985). Using daily stock returns: The case of event studies. Journal of Financial Economics, 14(1), Cesari D.A., Espenlaub S., Kurshed A.,and Simkovic M., (2012), The effects of ownership and stock liquidity on the timing of repurchase transactions, Journal of Corporate Finance, Vol. 18, Issue 5, pp Comment, R., & Jarrell, G. A. (1991). The relative signalling power of Dutch auction and fixedprice self tender offers and open market share repurchases. The Journal of Finance, 46(4),

58 Dann, L.Y. (1981), Common Stock Repurchases: An Analysis of Returns to Bondholders and Stockholders, Journal of Financial Economics, Vol. 10, pp De Jong, F., and de Goeij, P., 2011, Event Studies Methodology, Tilburg University Dittmar, A. K. (2000), Why Do Firms Repurchase Stock? The Journal of Business, 73(3), Easterbrook F.H., Two Agency-Cost Explanations of Dividends, American Economic Review, 1984, vol. 74, issue 4, Eije, V.H., and Megginson, W.L., (2008), Dividends and share repurchases in the European Union, Journal of Financial Economics, 89, Fried, J. M., (2001), "Open Market Repurchases: Signaling or Managerial Opportunism?" Theoretical Inquiries in Law, 2, Grossman, Sanford J., and Oliver D. Hart, (1982), Corporate financial structure and managerial incentives, in John J. McCall, ed.: The Economics of Information and Uncertainty (University of Chicago Press, Chicago, IL) Grullon, G., & Ikenberry, D. L. (2000), What do we know about stock repurchases? Journal of Applied Corporate Finance, 13(1), Grullon, G., & Michaely, R., and Bhaskaran S., (2002), Are dividend changes a sign of firm maturity? Journal of Business, 75, Grullon, G., & Michaely, R. (2002), Dividends, share repurchases, and the substitution hypothesis, The Journal of Finance, 57(4), Grullon, G., Michaely, R., (2004), The Information Content of Share Repurchase Programs, Journal of Finance, Vol. 59, pp Ikenberry D., Lakonishok J., and Vermaelen T., (1995), Market underreaction to open market share repurchases, Journal of Financial Economics, 39, Ikenberry, D. L., Lakonishok, J., & Vermaelen, T., (2000), Stock repurchases in Canada: Performance and strategic trading, The Journal of Finance, 55(5), Ikenberry, D. L., & Vermaelen, T. (1996). The option to repurchase stock. Financial Management, Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The American Economic Review, 76(2), Kahle, K. M. (2002). When a buyback isn ta buyback: open market repurchases and employee options. Journal of Financial Economics, 63(2),

59 Lasfer, M.A. (2002). The market valuation of share repurchases in Europe, Working paper, City University Business School. Lie, E. (2005). Operating performance following open market share repurchase announcements. Journal of Accounting and Economics, 39(3), Liu, H., Swanson, E.P. (2016). Is Price Support a Motive for Increasing Share Repurchases? Journal of Corporate Finance, Vol. 38: MacKinlay, A. Craig (1997), Event Studies in Economics and Finance,Journal of Economic Literature, 35(1), Mathias M Siems & Amedeo De Cesari (2012), The Law and Finance of Share Repurchases in Europe, Journal of Corporate Law Studies, 12:1, Merton H. Miller, and Franco Modigliani, (1961), Dividend Policy, Growth, and the Valuation of Shares, The Journal of Business 34, pp Michel, A., Oded, J., Shaked, I., (2010), Not all buybacks are created equal: the case of accelerated repurchase programs Financial Analysts Journal, 66 (6), Miller, Merton H., and Kevin Rock, (1985), Dividend policy under asymmetric information, Journal of Finance 40, Mitchell, J.D., Dharmawan, G.V., (2007). Incentives for on-market buy-backs: Evidence from a transparent buy-back regime. J. Corp. Finan. 13, Oded, J. (2005). Why do firms announce open-market repurchase programs? Review of Financial Studies, 18(1), Opler, T., Titman, S., (1996), The debt-equity choice: An analysis of issuing firms, Working paper, Columbus: Ohio State University. Peyer, U., & Vermaelen, T. (2005), The Many Facets of Privately Negotiated Share Repurchases, Journal of Financial Economics, 75, Peyer, U., & Vermaelen, T., (2009), The nature and persistence of buyback anomalies, Review of Financial Studies, 22(4), Rasbrant, J., (2011), The price impact of open market repurchases, Journal of Economic Literature, JEL classification: G14; G35 Rau P. R., and Vermaelen T., (2002), Regulation, Taxes, and Share Repurchases in the United Kingdom, The Journal of Business, 75, pp

60 Rees, W., (1996), The impact of open market equity repurchases on UK equity prices, The European Journal of Finance, 2(4), Simkovic, M.N., (2008), The Effect of Enhanced Disclosure on Open Market Stock Repurchases, Harvard Law School Skinner, D.J., (2008), The evolving relation between earnings, dividends, and stock repurchase, Journal of Financial Economics, vol. 87, issue 3, pages Smith, T., (2011), Share Buybacks Friend or Foe?, Fundsmith Stephens, C.P. and Weisbach, M.S. (1998), Actual Share Reacquisitions in Open-Market Repurchase Programs, The Journal of Finance, Vol. 53, No. 1, pp Thein, M. (2013). Share Buyback: A Blank Check for Management? ABAC Journal, 33(1), Vermaelen, T., (1981), Common Stock Repurchases and Market Signalling: An Empirical Study, Journal of Financial Economics, 9(2), Vermaelen, T., (1984), Repurchase tender offers, signaling and managerial incentives, Journal of Financial and Quantitative Analysis 19, Zhang, H., (2002), Share repurchases under the Commercial Law in Japan: market reaction and actual implementation, Pacific-Basin Finance Journal, 10(3),

61 Appendix Appendix This section includes the critical values that are used in this study (see Table 17). In addition, Table 18 provides an overview of the sample firms with their corresponding DataStream codes. The DataStream codes for the independent variables are also included (see Table 19). Furthermore, there is a summary of the descriptive statistics of the non-repurchasing firms over the period (see Table 20). In addition, there is also an overview of the nonrepurchasing firms included in the sample (see Table 21). Furthermore, Figure 3 shows the development of the FTSEurofirst 300 index relative to the S&P 500. And Figure 4 and 5 include the buyback yield index for the European securities and US securities. Finally, Figure 6 shows the development of the average abnormal returns within the event window (-20,+20) in the crisis period ( ). And Figure 7 shows the development of the cumulative average abnormal returns over the event window (-20,+20). Table 17: Confidence level and their corresponding Critical value Confidence level 1% 5% 10% Critical value This table presents the confidence levels and their corresponding critical values that are used in this paper (see de Jong and de Goeij, (2011)) Table 18: List of the European firms included in the sample Firms listed on the FTSEurofirst 300 index, included in the sample Name Code 1 ACS ACTIV.CONSTR.Y SERV. E:ACS(P) 2 ADECCO 'R' S:ADEN(P) 3 ADIDAS D:ADS(P) 4 AEGON H:AGN(P) 5 AIR LIQUIDE F:AIR(P) 6 ASML HOLDING H:ASML(P) 7 ATLANTIA I:ATL(P) 8 ATOS F:ATO(P) 9 BAE SYSTEMS BA.(P) 10 BBV.ARGENTARIA E:BBVA(P) 11 BNP PARIBAS F:BNP(P) 12 BP BP.(P) 13 BRITISH AMERICAN TOBACCO BATS(P) 14 CARREFOUR F:CRFR(P) 61

62 15 CENTRICA CNA(P) 16 COLOPLAST 'B' DK:COL(P) 17 COMPASS GROUP CPG(P) 18 CREDIT AGRICOLE F:CRDA(P) 19 DASSAULT SYSTEMES F:DSY(P) 20 DEUTSCHE BANK D:DBK(P) 21 DEUTSCHE BOERSE D:DB1(P) 22 DEUTSCHE TELEKOM D:DTE(P) 23 DSV 'B' DK:DSV(P) 24 EMS-CHEMIE 'N' S:EMS(P) 25 ENI I:ENI(P) 26 ERSTE GROUP BANK O:ERS(P) 27 ESSILOR INTL. F:EI(P) 28 EXPERIAN EXPN(P) 29 FERROVIAL E:FERC(P) 30 FRESENIUS MED.CARE D:FME(P) 31 GEBERIT 'R' S:GEBN(P) 32 GRIFOLS ORD CL A E:PROB(P) 33 HERMES INTL. F:RMS(P) 34 HEXAGON 'B' W:EKBF(P) 35 HOCHTIEF D:HOT(P) 36 IBERDROLA E:IBE(P) 37 ILIAD F:ILD(P) 38 INFINEON TECHNOLOGIES D:IFX(P) 39 ITV ITV(P) 40 DSM KONINKLIJKE H:DSM(P) 41 KPN KON H:KPN(P) 42 PHILIPS ELTN.KONINKLIJKE H:PHIL(P) 43 LEGRAND F:LRRS(P) 44 L'OREAL F:OR@F(P) 45 NATIONAL GRID NG.(P) 46 NOVARTIS 'R' S:NOVN(P) 47 NOVO NORDISK 'B' DK:NON(P) 48 ORANGE F:ORA(P) 49 PEUGEOT F:PGT(P) 50 PROSIEBENSAT 1 MEDIA D:PSM(P) 51 PROXIMUS B:PROX(P) 52 PUBLICIS GROUPE F:PUB(P) 53 RED ELECTRICA E:REE(P) 54 RELX REL(P) 55 REPSOL YPF E:REP(P) 56 RICHEMONT N S:CFR(P) 57 RWE D:RWE(P) 58 SAP D:SAP(P) 59 SCHINDLER 'R' S:SCHN(P) 60 SHIRE SHP(P) 61 SNAM I:SRG(P) 62

63 62 SODEXO F:SDX(P) 63 SOLVAY B:SOL(P) 64 SONOVA N S:SOON(P) 65 STANDARD LIFE SL.(P) 66 STMICROELECTRONICS (PAR) F:SGS(P) 67 THE SWATCH GROUP 'B' S:UHR(P) 68 SWISSCOM 'R' S:SCMN(P) 69 TELENOR N:TEL(P) 70 TELIA COMPANY W:TEL(P) 71 TESCO TSCO(P) 72 SAGE GROUP SGE(P) 73 UNIBAIL-RODAMCO H:UBL(P) 74 VALEO F:FR(P) 75 VINCI F:DG@F(P) 76 VIVENDI F:EX@F(P) 77 VODAFONE GROUP VOD(P) 78 WOLSELEY WOS(P) 79 WOLTERS KLUWER H:WSG(P) 80 ZURICH INSURANCE GROUP S:ZURN(P) Table 19: The independent variables with the corresponding DataStream codes Indep. variable Code Total assets MVTB Long-term debt ROA ROE (WC02999) (MTBV) (WC08216) (WC08326) (WC08301) Cash flow to Sales (WC08311) Table 19 presents the codes used from the Thomson Reuters Database for the independent variables 63

64 Table 20: An overview of the descriptive statistics of the independent variables related to the nonrepurchasing firms in the period Period Non repurchasing firms Obs. Mean Median Std. Dev. Min Max Independent variables ln Total assets 28,160 18,22 17,82 1,91 13,37 21,01 MVTB 28,160 2,82 1,88 3,44 0,23 29,77 LTDebt 28,160 43,05 38,98 16,42 10,10 82,32 ROA 28,160 4,28 2,11 5,24-8,13 24,35 ROE 28,160 15,49 14,14 18,24-47,24 98,67 Cash Flow/ Sales 28,160 14,76 12,89 8,79-2,71 58,62 64

65 Table 21: Overview of the non-repurchasing firms listed on the FTSEurofirst 300 index Company_id Non-repurchasing firms Code 1 ASSA ABLOY W:ASSB 2 ASSICURAZIONI GENERALI I:G 3 AVIVA AV. 4 BBV.ARGENTARIA E:BBVA 5 BAYER D:BAYN 6 DEUTSCHE POST D:DPW EDP ENERGIAS DE 7 PORTUGAL P:ECP 8 SAFRAN F:SGM 9 SOCIETE GENERALE F:SGE 10 UNITED INTERNET D:UTDI Table 21 shows an overview of the non-repurchasing firms (with the corresponding DataStream codes) listed on the FTSEurofirst 300 index in the period The sample of non-repurchasing firms excludes inactive firms and firms with less data points Figure 3: The development of the S&P 500 Index and the FTSEurofirst 300 index Figure 3 shows graphically the development of the S&P 500 index (green line) and the FTSEurofirst 300 index (orange line) over the period Source: Thomson Reuters. Retrieved from: 65