This thesis studies the abnormal returns following share repurchase announcements during the period 2002-2011. The sample consists of 697 non-financial firms listed on the AEX in the Netherlands and the DAX in Germany. The event window covers the 3 working days surrounding the announcement.
The research results in a positive statistically significant average abnormal return of 0.14% for the announcement day. The cumulative average abnormal return during the event window adds up to an abnormal return of -0.09%. The cumulative abnormal return is not statistically significant. The results are considerably lower than previous results from similar studies performed for the United States.
Additionally the relation between the tax rate on capital in the Netherlands and the abnormal return is studied. This resulted in a negative statistical significant coefficient. Therefore it can be expected that abnormal returns decrease after a increase in tax rate.
Further the research confirms the free cash flow hypothesis, market undervaluation hypothesis, capital structure adjustment hypothesis, dividend substitution hypothesis and capital market allocation hypothesis.
keywords: event study, share repurchase, open market repurchases, abnormal returns, Netherlands, Germany, undervaluation hypothesis, substitution hypothesis, free cash flow hypothesis, tax.
1. Introduction
A share repurchase is a program in which a firm uses excess cash to buy back its own shares. By repurchasing shares the number of shares outstanding decreases. A decrease in the number of shares outstanding increases the earnings and dividends on a per share basis. [1]
Nowadays share repurchases are a popular instrument for firms to distribute money. Though share repurchases have not always been such a widely used instrument. Until 1983 share repurchases were relatively uncommon in the United States because of legal restrictions. Most firms were discouraged of repurchasing shares by the risk of violating the anti manipulative provisions of the Securities Exchange Act (SEA) of 1934. In 1982 the Securities and Exchange Commission (SEC) adopted rule 10b-18. This rule provides guidelines for firms that want to repurchase shares on the open market without violating the 1934 SEA. Rule 10b-18 dictates that a firm with plans to repurchase shares should announce its plans publicly. The firm is not allowed to use more than one broker per day. [2] The daily repurchasing volume is restricted and trading is not allowed during the last half hour before the closing of the market. After that, starting in the mid-1980's, share repurchases became increasingly popular in the United States. [3] 4
Until the late 1990s share repurchases used to be forbidden in much of continental Europe. European countries used to prohibit repurchases or made them unattractive though harsh tax laws (Jariwala, 2012). In France and Germany repurchases used to be prohibited until 1998 (Ginglinger and Hamon, 2005; Hackethal and Zdantchouk, 2003). In the Netherlands repurchasing firms used to pay 33â…“% withholding tax on the difference between the repurchase price and the average paid up capital (Lasfer, 1998). Because of this harsh tax treatment repurchases were uncommon in the Netherlands. In 2001 a new tax legislation was introduced in the Netherlands. The new legislation reduced the tax costs on repurchases. The new law prescribes that Dutch firms need an approval at the shareholders meeting to engage in a repurchase program. This approval is valid for 18 months. Firms are not allowed to repurchase more than 10% of their shares. Finally, repurchases should be financed using distributable profits (Kim et al., 2011).
Grullon and Michaely (2000) used Compustat data to study whether share repurchases are a substitute for paying dividends. They discovered that the expenditures of earnings to share repurchase programs by US companies grew from 4.8% in 1980 to 50.1% in 1998. In 1998, for the first time in history, US companies spend more money on repurchasing shares than on paying dividends.
Similar to the findings of Grullon and Michaelly (2000), Von Eije (2007) reports that the propensity to pay cash dividends by EU15 countries declined from 83.5% in 1989 to 52.8% in 2005. The propensity to repurchase shares shows the opposite pattern. While the propensity was 5% in 1989, in 2005 the propensity increased to 17.8%. Von Eije also reports that the total value of share repurchases by the EU15 [5] countries (at 2000 prices) grew from € 6,146 million in 1989 to € 58,841 million in 2005. Repurchases in the United Kingdom account for almost 50% of the cumulative value of share repurchases by EU15 countries of € 252,943 million during the period 1989-2005. The total value of share repurchases (at 2000 prices) by Dutch companies grew from € 114 million in 1989 to € 7,383 million in 2005.
The literature describes multiple possible explanations for the increased popularity of share repurchases. Grinblatt and Titman (1998) suggest that managers previously did not understood the relation between dividends and share repurchases. The shift from dividends to repurchases could indicate a better understanding of that relations and the tax advantage of repurchases over dividends. Second, the adoption of rule 10b-18, as mentioned before, provides guidelines to firms to repurchase shares without violating SEC rules. Grullon and Michaely (2000) found that the aggregate amount of cash that was spend on share repurchases tripled during the first year after the approval of rule 10b-18. Even after controlling for changes in tax, market conditions and executive stock options Grullon and Michaely found that the adoption of the rule remains economically and statistically significant. Third, according to Bens et al. (2003) firms use repurchases to compensate for employee stock options. During the late 1990's employee stock options (ESO's) became popular. ESO's were used as a compensation tool for employees. The advantage of ESO's is that there is no actual cash payment, therefore earnings are not affected. Bens et al. found that managers use share repurchases to compensate for the dilution caused by the granting of employee stock options. The granting of ESO's eventually leads to an increase in the number of shares outstanding which reduces the earnings on a per share basis. Managers use share repurchases to alter the increase in the number of shares outstanding.
With the increasing popularity of share repurchases over the last 30 years researchers have started to study the abnormal returns that follow on the announcement of a share repurchase. Most of these studies have focused on the United States. The research resulted in an average abnormal return for US companies of 2.8% over the 2 days surrounding the repurchase announcement. More recently similar studies were performed for European countries. These studies generated the following abnormal returns: 1.31% for the United Kingdom (Oswald and Young, 2002), 0.67% for France (Ginglinger and L'Her, 2006) and 5.87% and 5.97% for Germany (Seifert and Stehle, 2003; Hackethal and Zdantchouk, 2006).
Researchers have developed theories to explain why firms repurchase and which determinants drive these abnormal returns. In this thesis the following 5 models will be reviewed and tested: The agency costs of free cash flow hypothesis, the capital market allocation hypothesis, the capital structure adjustment hypothesis, the dividend substitution hypothesis, the market undervaluation hypothesis and the share repurchase option hypothesis.
In this thesis I will focus on the abnormal return following share repurchase announcements in the Netherlands. In this thesis the 7 models mentioned above will be reviewed and tested. The results will be compared with results from similar studies for the United States.
The study of share repurchases in the Netherlands is particularly interesting because of the difference in the taxation treatment of dividends and capital gains in the Netherlands compared to the United States. In the Netherlands an actual capital gain tax does not exist. Therefore from the taxation perspective investors will prefer repurchases over dividends. This preference could indicate higher abnormal returns following a repurchase announcement.
To test the hypotheses all share repurchases by non financial companies listed on the Dutch AEX and German DAX during the period 2002-2011 have been gathered. The announcement data is collected using Zephyr. The data for the explanatory variables has been gathered using Thomson Datastream.
The results show a positive average abnormal return on announcement day of 0.14%. The cumulative average abnormal return over the two days surrounding the announcement adds up to an abnormal return of 0.09%.
The remainder of this thesis is structured as follows. Section two describes the methods to repurchase, the developed theories an reviews previous research. Section three describes the collected data and its sources. Section four describes the method used to calculate the abnormal return and the regressions on the abnormal returns. Finally in section five the empirical results are discussed. Finally, section six presents the conclusions.
2. Literature review
The first paragraph reviews the changes in regulations that initiated the growing popularity of share repurchases. The second paragraph describes the methods to repurchase shares. In the third paragraph the hypotheses are discussed and reviewed. Finally the fourth paragraph, reviews the treatment of capital gains in the US, Germany and the Netherlands.
2.1 The popularity of share repurchases
Nowadays share repurchases are a popular instrument for firms to distribute money. Though share repurchases have not always been such a widely used instrument. Until 1983 share repurchases were relatively uncommon in the United States because of legal restrictions. Most firms were discouraged of repurchasing shares by the risk of violating the anti manipulative provisions of the Securities Exchange Act (SEA) of 1934. In 1982 the Securities and Exchange Commission (SEC) adopted rule 10b-18. This rule provides guidelines for firms that want to repurchase shares on the open market without violating the 1934 SEA. Rule 10b-18 dictates that a firm with plans to repurchase shares should announce its plans publicly. The firm is not allowed to use more than one broker per day. [6] The daily repurchasing volume is restricted and trading is not allowed during the last half hour before the closing of the market. After that, starting in the mid-1980's, share repurchases became increasingly popular in the United States. [7] 8
However share repurchases used to be forbidden in much of continental Europe until the late 1990's. European countries used to prohibit repurchases or made them unattractive though harsh tax laws (Jariwala, 2012). In France repurchases were permitted but difficult to implement. After the law was changed in 1998 open market repurchases became popular in France (Ginglinger and l'Her, 2002). In Germany repurchases used to be prohibited until 1998. (Hackethal and Zdantchouk, 2003).
In the Netherlands repurchasing firms used to pay 33â…“% withholding tax on the difference between the repurchase price and the average paid up capital (Lasfer, 1998). Because of this harsh tax treatment repurchases were uncommon in the Netherlands. In 2001 a new tax legislation was introduced in the Netherlands. The new legislation reduced the tax costs on repurchases. The new law prescribes that Dutch firms need an approval at the shareholders meeting to engage in a repurchase program. This approval is valid for 18 months. Firms are not allowed to repurchase more than 10% of their shares. The shareholders meeting determines the minimum and maximum price to be paid for the shares. Finally, repurchases should be financed using distributable profits (Kim et al., 2011). In 2006 the Dutch legislation is changed again. From that moment quoted companies are permitted to repurchase up to 50% of their shares outstanding. The validity of the shareholders approval is extended to 5 years.
2.2 Methods to repurchase shares
Paying out dividends is the most well-known way of distributing money to shareholders. Alternatively a firm can choose to repurchase shares. By repurchasing shares the number of shares outstanding decreases. Share repurchases can be financed by the use of internal funds or by attracting debt. Four types of share repurchases can be distinguished: Open market repurchases, Fixed-price tender offers, Dutch auctions and Private negotiated repurchase (Vermaelen, 2005). The four types are described below.
Open market repurchases
This method is the most popular method to repurchase shares. Grullon and Ikenberry (2000) found that around 95% of all announced share repurchases involve open market repurchases.
In the case of an open market repurchase a firm announces its intention to repurchase shares and starts buying them on the open market like any other investor. The firm is not obligated to repurchase the total amount of shares that they stated to repurchase, they could repurchase less. In the United States there is no time limit in which a firm should finish the repurchase program. [9] In the Netherlands a firm should first get approval to repurchase at the shareholders meeting. At the shareholder meeting the minimum and maximum prices are also determined. From the moment of the approval the firm has 5 years to repurchase otherwise the approval expires. [10]
Fixed-price tender offers
In the case of a fixed-price tender offer the firm offers a fixed price to its shareholders if they tender their shares. Shareholders have a specific period of time to respond to this offer. The offered price generally offers a premium over the current stock price. Often the realization of the repurchase depends on enough shareholders tendering their shares. If the target number of shares tendered is not met the firm may cancel its offer. When the offer is oversubscribed each shareholder receives cash for its offered shares in proportion to the threshold. The balance is returned in stock (Grullon and Ikenberry, 2000).
Fixed-price tender offers can also be used by other parties. For example when a third party wants to take over a firm or obtain a large share in the firm. In this case the third party wants to acquire the shares of the firm. When this party chooses to buy the shares on the open market it will drive up the price which is not favorable. To avoid overpayment the party could announce a tender offer to the shareholders of the target firm. When enough shareholders tender the acquirer will receive all the desired shares at once at the cost of the offered price.
Dutch auction
Dutch auctions are related to tender offers. When a firm uses a Dutch auction to repurchase shares it announces different prices at which it is prepared to buy shares. Shareholders are able to sign up how many shares they are willing to sell for the given prices. After the lapse of the period the firm buys the shares for the price that allows the firm to repurchase all the necessary shares. All the tendered shareholders receive the same price, also the shareholders who tendered for a lower price.
Privately negotiated repurchase
A firm could also decide to repurchase shares from a major shareholder. This is called a privately negotiated repurchase. In the case of a privately negotiated repurchase the firm negotiates directly with the seller. This can be the case when the major shareholder wants to sell its position in the firm but the market is not sufficiently liquid to sustain such a large sale without severely affecting the stock price. Therefore the major shareholder could approach the firm directly and negotiate the sale of its shares against a large discount. It is also possible that the firm approaches the shareholders instead of the other way around. This could be the case when the shareholder is threatening to take over the firm and replace the management. When this situation occurs the firm can buyout the shareholder, this often happens for a huge premium over the market price. This type of transaction is also known as greenmail (Peyer and Vermaelen, 2004).
2.3 Why do firms repurchase?
This paragraph will introduce the most important hypotheses that have been developed to explain the existence of share repurchases and the prevailing abnormal return. First the hypotheses will be introduced briefly after which previous research into the hypothesis will be reviewed.
Agency costs of free cash flows
The agency theory highlights the conflict of interest between principals and agents. In listed firms managers act as agents on behalf of the shareholders. Shareholders expect the managers to make decisions as such that their share of the firm increases in value. With the separation of ownership and control shareholders lose control over managers. Giving managers the ability to put self-interest above the interest of the shareholders.
When a firm has excess cash and no value creating projects available managers will be tempted to invest the cash in value destroying projects just to manage a larger firm. Thereby putting growth and size of the company above profitability and value. The costs that arise from the conflict between maximizing growth and maximizing value, the conflict between managers and shareholders, is known as agency costs of free cash flow.
The primary method that managers use to distribute excess cash is by paying out dividends. Lintner (1956) concluded on the basis of interviews with executives that firms use a dividend payout target. He also found that the market puts a premium on stocks that offer a stable or slightly growing dividend payout. Therefore managers avoid making changes in their dividend payout.
Therefore paying out the excess cash as dividend is not an option. An alternative for investing the excess cash into value destroying projects or paying it out in dividends could be to repurchase shares. Share repurchases are likely to be perceived as good news because it reduces the cash that managers can invest in value destroying projects. By repurchasing shares the number of shares outstanding decreases and at the same time reduces the agency costs of free cash flow.
In order to decrease agency costs Jensen (1986) suggests that firms that hold a large amount of free cash flow should raise dividend or repurchase shares. By increasing the distributions to the shareholder, the free cash flow in the firms decreases. By having lower levels of free cash flow there is less money in the firm that the manager can waste. Therefore the agency costs decrease. Li and McNally (1999) confirm this by reporting that Canadian firms with a greater amount or free cash flow are more likely to repurchase shares. Stephen and Weisbach (1998) find a positive relation between free cash flow and the number of repurchased shares in the US.
Jensen (1986) suggest that firms with higher levels of free cash flow will experience higher abnormal returns.
Grullon and Michaely (2004) find that repurchasing firms experience a decline in profitability and a decrease in investments in the years after the announcement. From this they preclude that earnings will recover in the long run. Their evidence suggests that the reduction is free cash flow is a source of the positive market reaction to the repurchase announcement. Further they conclude that firms that overinvest experience higher abnormal returns in case of a repurchase.
Li and McNally (1999) report that firm size and market-to-book ratio do not influence the abnormal return on announcement date for Canadian firms.
Kahle (2002) reports a positive coefficient for free cash flow.
Capital market allocation hypothesis
Once of the central functions of the financial markets is to allocate capital among investment opportunities (Grullon and Ikenberry, 1999). In the perfect world only value creation projects should be able to collect the necessary capital. Due to market imperfections the efficient allocation of capital cannot always be achieved. One of these imperfections is the information asymmetry between managers and investors.
This theory can be seen as a complement to the previous theory. Like the previous theory this theory considers the distribution of excess cash to the shareholders as positive. This hypothesis comes with the suggestion that when companies run of out value creating investments they should return capital to their shareholders. Afterwards the shareholders can decide to reallocate their capital into companies that do have positive investment opportunities. Therefore share repurchases are perceived as good news when a firms does not have value creating investment projects available.
Nohel and Tarhan (1998) conclude that repurchases are a part of a restructuring package instead of a pure financial transaction. They find a positive correlation between asset sales and post-repurchase performance. Grullon and Michaely (2004) also support this hypothesis. The authors find that firms decrease their capital expenditures after a repurchase announcement.
Capital Structure adjustment hypothesis
Firms can use repurchases to increase their leverage ratio. By repurchasing shares the value of equity decreases which increases the debt-to-equity ratio. Because interest payments are tax deductible
An alternative theory why corporations repurchase their shares is to adjust their capital structure. By repurchasing shares leverage ratios increase. Alternatively share repurchases could be used to compensate for dilution caused by employee stock option plans or dividend reinvestment plans. A study performed by Chan, Ikenberry and Lee (2000) shows that repurchase plans are often announced around the execution of executive stock options.
A firms leverage ratio is used as a proxy for the way a firm is financed. Firms with a low leverage ratio can use a share repurchase to increase their leverage ratio. If an optimal leverage ratio exists firms can use share repurchases to reach this optimal ratio (Bagwell and Shoven, 1988). By increasing debt the value of the tax shield of the firms increases. Therefore firms with a low leverage ratio are more likely to repurchase.
Firms with a high leverage ratio are more cash constrained because they have high interest expenditures. Managers in a cash constrained firm should be more motivated to take the right decisions. Therefore the agency costs of free cash flow are expected to be lower. Therefore firms with a higher leverage ratio are expected to generate higher abnormal returns (Jensen, 1986).
The value of a firms increases with its leverage ratio. This increase in value comes to a halt when bankruptcy costs kick in. Therefore, until a certain level of leverage firms are in the position to repurchase. Above that level, they face financial distress.
Because the agency costs of free cash flow decrease with a higher level of leverage and the value of the tax shield increase with leverage, I expect the coefficient for this variable to be positive.
Cash flow signaling
Managers are often thought of having more information about a company's true value than shareholders. This difference in information is referred to as the informational asymmetry between managers and shareholders. All publicly available information is reflected in a firms stock price. Of course not all information that managers have is made public. Managers could use credible signals to convey their optimism of future earnings to the market. One of these signals could be a share repurchase. By announcing a share repurchase managers limit their own flexibility. When managers know that earnings will be lower than expected they will not announce a share repurchase plan because there will not be enough money left for the necessary investments. When a manager knows that the earnings will be higher than expected he can announce a share repurchase plan. With the higher earnings the necessary investments can be covered. Therefore share repurchases are considered a credible signal.
Dividend substitution hypothesis
Firms can distribute money to their shareholders either by using dividends or by using a share repurchase. According to the dividend irrelevance theory of Miller and Modigliani (1961) dividends and repurchases are, except for tax and transaction costs, perfect substitutes. However managers are reluctant to cut dividends. Lintner (1956) concluded on the basis of interviews with executives that firms use a dividend payout target. He also found that the market puts a premium on stocks that offer a stable or slightly growing dividend payout. This indicates that managers see dividends more as a commitment. Share repurchases on the other hand are no commitment, they can even be cancelled or just not completed.
Grullon and Michaely (2000) show that firms have been gradually substituting dividends for repurchases. They discovered that the expenditures of earnings to share repurchase programs by US companies grew from 4.8% in 1980 to 50.1% in 1998. In 1998, for the first time in history, US companies spend more money on repurchasing shares than on paying dividends.
When an investor decides to sell his shares to the firm following a repurchase announcement he will incur capital gain taxes. [11] An investor that decides not to sell his shares increases his share of ownership in the firm. The latter investor does not directly receive a tax bill in countries where a capital gain tax is charged. [12] He has to pay capital gain tax at the moment that he sells his shares. Therefore, in these countries, share repurchases can be considered the better alternative compared to dividends for long term investors. This might explain the popularity of share repurchases.
However the tax treatment of the two methods differs. In the United States there are two sorts of dividends; qualified dividends and non-qualified dividends. Qualified dividends are dividends that the investor has received from a stock that he held for more than 60 days during the 121 days surrounding the ex-dividend date. When the investors income is subject to the 25% personal tax bracket of higher, the dividend is taxed at the 15% level. When the investors income is subject to a tax bracket lower than 25%, the dividend is taxed at the 0% level. Non-qualified dividends are dividends that do not meet the 60 day holding period criteria. Non-qualified dividends are taxed at using the investors personal tax bracket. Capital gains with a holding period less than one year are taxed using the investors income tax bracket. Capital gains with a holding period that exceeds one year are taxed at 15%. [13] Because of this rather complicated taxation system, the tax rate that investors have to pay on their dividends and capital gains will not always match. Therefore the two methods are imperfect substitutes.
Grullon and Michaely (2000) estimate next year's dividend payment using a firm's dividend policy in previously years. The result suggests that dividend forecasts and repurchase announcements are negatively related. This evidence supports the dividend substitution hypothesis. Bagwell and Shoven (1989) suggest that firms have learned to substitute dividend for repurchases to provide tax benefits for their shareholders.
Market undervaluation hypothesis
Alternatively to the cash flow signaling theory there is this signaling theory. Managers can use share repurchases as signal to the market to reveal their disagreement with the current price of the publicly available information. Given their position and the fact that they have more information than outsiders, managers are probably in the best position to estimate the value of a firm. Therefore repurchase announcements can be perceived as of signal of undervaluation.
Ikenberry et al (1995) rank the firms post-announcement abnormal returns on their market-to-book value. They find that firms ranked in the lowest market-to-book value group experience a four year abnormal return of 45.3%. Firms in the highest market-to-book value groups experience no or negative abnormal returns. Surprisingly they found that the initial reaction of the market on a repurchase announcement is similar across all market-to-book-value groups.
Stephens and Weisbach (1998) find a negative correlation between a firms decision to repurchase and the firm's stock return in the quarter prior to the announcement. Indicating that firm's who's stock underperformed in the last quarter are more likely to repurchase. This observation is confirmed by many studies (Vermaelen, 1981; Comment and Jarrell, 1991; McNally, 1999). A negative relation is found between the pre-announcement return and the abnormal return on announcement date (Stephens and Weisbach, 1998; Kahle, 2002). Indicating that firms that underperform in the period before the repurchase announcement generate higher abnormal returns.
Stephens and Weisbach (1998) also find that 57% of the firms repurchase more shares then they announce. The authors conclude that their results confirms the view that managers exploit the flexibility of repurchases. And that this flexibility explains the growing popularity if share repurchases.
The undervaluation hypothesis highlights the importance of information asymmetries for abnormal returns. Smaller firms are expected to have greater information asymmetries. Vermaelen (1981) states that smaller firms are expected to signal more information when they announce to repurchase. He suggests that smaller firms have higher levels of informational asymmetries because of (1) less attention from financial media, (2) lower institutional ownership, and (3) high fraction of insider holdings. Accordingly many studies confirm the negative relationship between firm size and abnormal return on announcement date (Chan et al., 2004; Grullon and Michaely, 2002; Stephen and Weisbach; 1998).
In the Netherlands the undervaluation theory is tested by Erken (2012) and Fierkens (2010). Erken (2012) study confirms the implications of the undervaluation hypothesis, however the results are not statistically significant. Fierkens (2010) finds no evidence that supports the undervaluation hypothesis in the Netherlands.
Share repurchase option hypothesis
Ikenberry and Vermaelen (1996) add another explanation to the list. They state that the existence of open market repurchase programs provides managers the resources to use insider knowledge of the firm to benefit the long term shareholders by having the possibility to reacquire shares at some point in the future. Managers can decide to do this when in their opinion the market price has deviated from the true value.
The earnings bump hypothesis
When managers announce a share repurchase program they often state that they decided to repurchase shares to increase the earnings on a per share basis. The earnings per share increase when the relative decrease of the number of shares outstanding exceeds the relative percentage change in earnings. There is however a side note. When a firm repurchases shares it is in fact shrinking its asset base. Shrinking a firms asset base only adds value when the firm cannot earn the cost of capital on its marginal investments. Therefore the real source of the increase in price following the repurchase lies in the reallocation of capital into more profitable investments. What basically refers to the capital market allocation hypothesis.
2.4 Tax on dividends and capital gains
In this paragraph the tax treatment of capital gains [14] and dividends in the Netherlands and Germany will be compared to the treatment in the United States.
The Netherlands: In the Netherlands there is no tax on capital gains. Savings and investments are however subject to a 1.2% capital yield tax [15] . All savings and investments above the threshold of € 21,785 is taxed using this 1.2% tax rate.
Dividends are taxed at a 15% flat tax rate. [16]
The United States: In the United States capital gains are taxed using an individual's income tax rate when the holding period is less than one year. An investors income tax rate can vary from 10% to 35%. When the holding period exceeds one year the capital gain is taxed at a rate of 15%. [17]
In the United States two types of dividends are distinguished; qualified and non-qualified dividends. Dividends are 'qualified dividends' when the stock is held for more than 60 days during the 121 days surrounding the ex-dividend date. Qualified dividends are taxed at a 15% rate when the investor is in the 25% tax bracket or higher. The dividends are not taxed when the investors income is subject to a tax bracket below 25%.
Non-qualified dividends are all dividends that do not meet the criteria above. Meaning that the holding period of the stock is less than 60 days during the 121 days surrounding the ex-dividend date. Non-qualified dividends are taxed using the rate of the investors personal tax bracket. [18]
Germany: In 2009 a flat tax rate on capital gains was introduced in Germany. The flat tax rate is 25% and will be incremented with a 5.5% solidarity surcharge. Therefore the total tax on capital gains in Germany is 26.375%. Before 2009 capital gains with a holding period longer than one year were not taxed while capital gains with a holding period shorter than a year were taxed at xx%.
Dividends are taxed at the 26.375%.
Difference
In the United States dividends from individuals with a higher income are taxed at a higher tax rate. While in the Netherlands and Germany a flat rate is used. Long term capital gains are taxed at 15% in the US, while short term capital gains are taxed using a progressive tax rate. In the Netherlands an actual capital gains tax does not exist. Instead the total amount of savings and investments is taxed at the 1.2% level. In Germany capital gains are taxed at 26.375%, no distinction is made between long and short term capital gains.
The favorable tax treatment of capital gains in the Netherlands as compared to Germany might lead to higher abnormal returns following share repurchase announcement in the Netherland as compared to Germany. The capital gains tax rates from the Netherland and Germany will be incorporated in the regressions that will be performed later in this thesis. By adding the taxation rate to the regression I am hoping to find a relation between the level of tax and the level of abnormal return.
3. Sample selection and data sources
This section describes the data types and its sources that have been used for this thesis. The first paragraph describes the steps that have been taken to gather the necessary data. The second paragraph discusses the sample selection.
3.1 Data collection
The analysis uses data from Bureau van Dijk - Zephyr and Thomson Datastream. The sample covers all share repurchase announcements made by non-financial firms listed on the Euronext Amsterdam Stock Exchange (AEX) and the Deutsche Aktienindex (DAX) during the period January 1 2002 until December 31 2011.
Step 1 Share repurchase announcements
From Bureau van Dijk - Zephyr all open market share repurchase announcements by companies listed on the DAX and AEX are gathered. Open market repurchases account for approximately 95% of the repurchases. Only the announcements during the period starting on January 1st 2002 until 31st December 2011 are included. This resulted in 727 announcements by companies listed on the AEX and 55 announcements by companies listed on the DAX. Making a total of 782 repurchase announcements. The ISIN number is added as identifying variable.
Step 2 Abnormal returns
In order to predict the normal behavior of the share price surrounding the event date, as will be discussed in the section methodology, historical share price data is needed. The ISIN number and announcement date as gathered in step 1 were used to collect the data. For all events historical stock prices are collected starting 121 business days before the event until 10 business days after the event. To be able to model the 'normal' stock price behavior for the days surrounding the event, for the same period the returns of the AEX and DAX are gathered. This data is gathered using the Request table function of Thomson Datastream.
Step 3 Collected data for the regression analysis
In this thesis the 5 main hypotheses developed to explain the existence of share repurchase are tested. To be able to test the hypotheses we need to collect data that proof the hypotheses. Therefore the following data has been gathered using Datastream: DWFC (Free cash flow from operations), DWCX (capital expenditures) for the current year and the next year, DWNP (net profit), DWTA (value of total assets), DWND (value of net debt), MTBV (market-to-book value), POUT (payout ratio).
3.2 Sample selection
Using Zephyr 782 open market share repurchase announcements have been gathered. 727 of the 782 events are announced by firms listed on the AEX. The remaining 55 are announcements by firms listed on the DAX. Like in previous studies into repurchases, financial institutions are excluded from the sample. Financial institutions are regulated by the government and could therefore have other reasons to repurchase shares. For example: banks are exposed to the Basel accords. The Basel accords restrict banks to hold a minimum total capital ratio. Therefore banks will be less inclined to repurchase shares.
After excluding the financial institutions 657 AEX announcements and 40 DAX announcements remained. Making the final sample to a total of 697 announcements.
Graph 1 gives an impression of the distribution of share repurchase announcements over the 40 quarters of the period 2002-2011. The absence of share repurchases before the year 2005 seems odd. Interestingly the 'wave' of share repurchases comes to a hold during the fall of 2008. This decrease is most definitely caused by the collapse of Lehman Brothers in September 2008 and the beginning of the financial crisis.
[Insert graph A1 here]
The descriptive statistics of the variables TargetFraction and Dealvalue as shown in table 2 shows that there is a wide variability in the size of the share repurchases. The target fraction varies from 0.006% until 16.32%. The value of the repurchases provides a similar view, the distribution starts at € 1.5 million and ends at € 7 billion. Therefore graph 2 presents an overview of the total value of announced share repurchases during each quarter of the period 2002-2011. The graph shows a decrease in the value of announced share repurchase programs after the collapse of Lehman in September 2008.
[Insert graph A2 here]
Looking at graph 1 and 2 simultaneously shows that the number of announced repurchases decreased considerably after 2008. But there are still a few high spikes. This could indicate fewer but larger share repurchase announcements after 2008.
4. Methodology
In order to estimate the abnormal return surrounding the announcement date of the share repurchase programs we have to use event study methodology. The event study methodology will be covered in the first paragraph of this section. The second paragraph describes the methodology used to determine whether the selected variables have an effect on the level of abnormal return.
4.1 Event studies
Event studies have been used for several decades to measure the impact of an event on the value of a firm. Examples of events are: share issues, stock splits, dividend payments and many more. Bowman (1983) describes four basic types of event studies; Information content, Market efficiency, Model evaluation and Metric explanation. The Information content method is used to measure the impact of an information notice like an annual earnings announcement on the value of the firm. The Market efficiency method measures whether the market reacts efficiently on for example the announcement of a stock split. The Model evaluation involves the development of multiple expectation models which are evaluated using the observed abnormal returns. Finally the Metric explanation analyzes the excess return metric by attempting to identify factors which are associated with the metric. The latter two methods are extensions of the first two.
I will use the 5 steps described by Bowman (1983) to conduct this event study. The five steps are: 1) Identify the event of interest. 2) Model the security price reaction. 3) Estimate the excess returns. 4) Organize and group the excess returns. I will describe the four steps in detail below. The fifth step, the analysis of the results, will be covered in section 5. De Jong and de Goeij (2011) is used as guideline for the notation of the formulas.
Step 1: Identify the event of interest
The first step in the event study methodology is to identify the event date. In this study the event date is the actual date that a firm announces its plan to repurchase shares. The sample consists of all non-financial firms listed on the Amsterdam Stock Exchange (AEX) or Deutscher Aktien indeX (DAX) that announced plans to repurchase shares during the period 01/01/2002 until 31/12/2012. Only open market repurchases are included in the sample.
Step 2: Model the security price reaction
The security price reaction is modeled using the security returns for a period in which no event occurred and its correlation with the market index. All securities are listed on the AEX or DAX, therefore the AEX and DAX are used as market index. The period in which no event occurred is referred to as the estimation period. The estimation period covers the period [-120,-31]. See figure 1 for an overview of the time line around the event.
[Insert figure A1 here]
The actual security price model is estimated using the Capital Asset Pricing Model (CAPM) (Sharpe, 1964). The CAPM expresses the return of a specific security as a function of the return of the market. The formula for the CAPM is as follows:
E(ri) = rf + βi x [E(rM)-rf] (1)
E(ri) refers to the expected return of the security
rf is a measure for the risk free rate
βi measures the correlation between the stock return and the return of the market
E(rM)-rf the excess return of the market over the risk free rate
Using an Ordinary Least Squares (OLS) regression the following formula is solved as optimally as possible. This is the exact same equation as (1) except for a slight change in notation. Also an epsilon is added, without an epsilon the two sides of the formula will not add up.
(2)
for t = -120, -119, -118, ... ,-32, -31 -30
With E(εit ) = 0 and Var(εit )=σ²(εit )
Rit resembles the actual security price reaction
αi is a unknown constant
βi measures the correlation between the security and the market
Rm,t equals the market return
εit resembles a random figure
An OLS regression is a regression that attempts to find the best fitting parameters for the regression. As its name indicates it searches for the coefficients that minimizes the sum of squared vertical distances between the dataset and it prediction. In English: it searches for the optimal αi and βi that minimizes the sum of squared differences between the returns predicted by the model and the actual returns.
By running an OLS regression on the stock returns in the estimation period of every share repurchase announcement against the market returns in the same period estimates for the unknown parameters are determined. For every repurchase event this results in one α and one β.
Using the estimated α and β the predicted return can be calculated. We want to estimate the predicted return for the period [-10, 10]. By multiplying the estimated β with the market return and adding the α the predicted return is calculated. See formula (3).
Predicted return: (3)
for t = -3, -2, -1, ..... , 1, 2, 3
Step 3: Estimate the excess returns
The excess return is calculated by subtracting the predicted return from the actual return. This results in the following abnormal return (AR):
(4)
for t = -3, -2, -1, ..... , 1, 2, 3
the estimate for α for event i
the estimate for β for event i
the return on the AEX index on date t
Step 4: Organize and Group the excess returns
The individual excess returns over the event window are grouped into the cumulative abnormal returns (CAR). The event window covers the period [-1, 1] as shown in figure 1.
(5)
Once the CAR's are calculated the cumulative average abnormal return (ACAR) can be calculated using the following formula:
(6)
The standard deviation of the CAAR:
sd ACAR = (7)
Testing the significance of the returns
Using a t test it can be estimated whether the resulting returns from formula (4) and (6) are statistically significant from zero.
The t test for the AR from formula (4) and the CAR from formula (6) are as follows:
test statistic = ARit/σ(ARi) (8)
test statistic = (9)
4.2 Regression analysis
This paragraph will introduce the regression model that will be used to test the hypotheses. Using an OLS regression, as mentioned before, we want to regress whether there is a relation between the level of the abnormal return and the level of the explanatory variables. OLS attempts to find the best fitting solution for all observations at once resulting in a coefficient and a measure for significance per explanatory variable. This results in the following two regressions for the AR and CAR respectively.
Below the regression model is presented. The cumulative abnormal returns (CAR) are calculated using formula (5).
for t = 0 (10)
for t = -1, 0, 1 (11)
The C in formula (10) and (11) refers to a constant.
4.3 Variables
This paragraph will present the variables that will be used in the regressions. First the variables used to test the hypotheses will be discussed. After which some additional variables will be presented.
Agency costs of free cash flow hypothesis
Free cash flow (fcf): The variable 'Free cash flow' is used to test the relationship between the abnormal return and the level of free cash flow in the firm. When a firm has excess cash and decides to distribute this excess cash to its shareholders, the firms sends a message to the market that instead of investing the money in negative NPV projects it is giving the money back to its shareholders. According to the hypothesis lower levels of free cash flow are associated with higher abnormal returns. Therefore I expect the sign for this variable to be negative.
Free cash flow is calculated as the net cash flow (DWFC) scaled down by the value of total assets (DWTA).
Capital expenditures (capex): The variable 'Capital expenditures' is used to test the relationship between the level of investments in the firm and the abnormal return on announcement date. When a firm has high capital expenditures this indicates high growth. Therefore firms with high levels of capital expenditures are less likely to distribute cash to their shareholders. Therefore I expect the sign for this coefficient to be negative.
Capital expenditures is calculated as capital expenditures (DWCX) scaled down by the value of total assets (DWTA).
Return on assets (roa): This variable is used to test the relationship between the profitability of the firm and the abnormal return. More profitable firms are expected to have more value creating investment opportunities. These firms are therefore likely to use excess cash to invest instead of to distribute. Therefore I expect the coefficient for this variable to be negative.
Return on assets is calculated as net profit (DWNP) scaled down by the value of total assets (DWTA).
Capital market allocation hypothesis
∆Capital expenditures (changescapex): The change in capital expenditures is used as a measure for the decrease in investment opportunities that a firm faces after a share repurchase. Instead of investing in negative NPV projects the firms distributes the money back to its shareholders. This will lead to a positive price effect. According to the hypothesis the capital expenditures of the firm decrease after a repurchase. Therefore I expect a negative coefficient for this variable.
The change in capital expenditures is calculated as the difference between the new and the old capital expenditures scaled down by the value of total assets (DWTA): (CAPEXt+1-CAPEXt)/DWTA.
Capital structure adjustments hypothesis
Leverage ratio (leverage): A firms leverage ratio is used as a proxy for the way a firm is financed. Firms with a low leverage ratio can use a share repurchase to increase their leverage ratio. If an optimal leverage ratio exists firms can use share repurchases to reach this optimal ratio (Bagwell and Shoven, 1988). By increasing debt the value of the tax shield of the firms increases. Therefore firms with a low leverage ratio are more likely to repurchase.
Firms with a high leverage ratio are more cash constrained because they have high interest expenditures. Managers in a cash constrained firm should be more motivated to take the right decisions. Therefore the agency costs of free cash flow are expected to be lower. Therefore firms with a higher leverage ratio are expected to generate higher abnormal returns (Jensen, 1986).
The value of a firms increases with its leverage ratio. This increase in value comes to a halt when bankruptcy costs kick in. Therefore, until a certain level of leverage firms are in the position to repurchase. Above that level, they face financial distress.
Because the agency costs of free cash flow decrease with a higher level of leverage and the value of the tax shield increase with leverage, I expect the coefficient for this variable to be positive.
The leverage ratio is calculated by scaling down the firms debt (DWND) by the value of total assets (DWTA).
Dividend substitution & taxation hypothesis
Dividend payout ratio (dividend): Paying out dividends is an alternative method for firms to distribute cash to their shareholders. In the Netherlands dividends are taxed at a higher tax rate than share repurchases. Therefore it could be expected that non-dividend paying firms generate higher abnormal returns. I expect the coefficient for this variable to be negative.
The dividend payout ratio is determined using datastreams payout ratio (POUT). POUT represents the proportion of the net income that has been paid out using dividends.
Market undervaluation hypothesis
Firm size (firmsize)): The logarithm op the value of total assets is used as a proxy for the size of a firm. The level of informational asymmetry is higher within small firms. Because of this asymmetry, small firms are more likely to use a share repurchase to signal information to the market. Big firms are better monitored by the market because more people are exposed to the performance of the firm. Therefore big firms are less likely to signal information to the market that investors do not already know. Hence I expect a negative coefficient for this variable.
Firm size is calculated using the logarithm of total assets (DWTA).
Market-to-book value (MTBV): Firms with a market-to-book value lower than one are called 'value firms'. The market value of these firms is lower than its book value. Firms with a high market-to-book value called 'growth firms' which are expected to have more profitable investment opportunities available (Hackethal et al., 2006). Therefore these firms are less likely to distribute money to their shareholders. Hence I expect a negative coefficient for this variable.
Market-to-book value is calculated using the market-to-book-value (MTBV)
Pre-announcement return (preCAR): The pre-announcement return represents the abnormal return during the period [-10,-1].
Other variables:
Capital gain tax (tax):
Target fraction (fraction):
4.4 Correlation between variables
To avoid the issue of multicollineairy in the regressions, the correlations between the variables have been checked. The correlation table is presented in table 4. For example: the table shows a high correlation between FCF and capex, leverage, MTBV. To avoid multicollineairy I will keep variables with a high correlation apart from another in the regressions.
[Insert table A4 here]
5. Empirical results
This section reports the empirical results following the performed regressions. The first paragraph reports the abnormal and cumulative abnormal returns calculated for the firm's stock returns during the event window. The second paragraph reports the result of the OLS regressions that have been performed in order to explain the calculated (cumulative) abnormal returns.
5.1 Abnormal and Cumulative Abnormal Return
Using the collected data as mentioned in section 4 and by following the method described in section 4 regressions were performed to approach the abnormal return of the firm's stock price surrounding the announcement of a share repurchase program. The abnormal returns are calculated for the days within the window [-10, +10]. The AAR's for both indexes and the total sample for the window [-10, +10] are presented in table X in the appendix. The table below presents the abnormal return for the window [-3, +3].
AEX
DAX
Total sample
Day
AAR
t-statistic
AAR
t-statistic
AAR
t-statistic
-3
0.001%
0.025
0.110%
0.485
0.007%
0.156
-2
-0.114%
-2.113**
0.061%
0.342
-0.104%
-2.003*
-1
-0.090%
-1.788*
-0.199%
-0.853
-0.096%
-1.954*
0
0.069%
1.053
0.656%
1.362
0.103%
1.524
1
0.008%
0.131
0.191%
0.589
0.018%
0.314
2
-0.070%
-1.280
0.043%
0.185
-0.063%
-1.193
3
-0.017%
-0.300
-0.076%
-0.426
-0.020%
-0.378
Table X
Table X shows a positive average abnormal return on announcement date for both the AEX and DAX firms that announce to repurchase shares. However, the AAR's are significantly lower than found in previous research, as presented table X in section 3. The AEX firms show an average abnormal return of 0.069% while the DAX firms show an average abnormal return of 0.656%. Both are statistically insignificant. Both indexes show a slightly positive AAR on the day after the announcement.
Table Y below presents the cumulative average abnormal returns calculated for different time windows. The cumulative average abnormal return over the event window, which covers the period [-1, +1], reaches -0.013% for the AEX firms and 0.648% for the DAX firms. Both CAAR's are statistically insignificant.
AEX
DAX
Total sample
Period
CAR
t-statistic
CAR3
t-statistic2
CAR2
t-statistic3
[-1, +1]
-0.013%
-0.139
0.648%
0.933
0.025%
0.247
[0, +1]
0.077%
0.871
0.847%
0.140
0.12084%
1.344
[-3, +3]
-0.212%
-1.452
0.786%
0.949
-0.155%
-1.062
[-10, -1]
-0.357%
-2.366***
-0.568%
-0.741
-0.273%
-1.835*
[-10, +10]
-0.625%
-2.789***
-0.779%
-0.626
-0.634%
-2.846***
Table Y
Table Y shows a statistically significant negative CAAR for the 10 days preceding the share repurchase announcement for the AEX firms as well as the total sample. The DAX firms show a similar, but insignificant, negative abnormal return. The negative abnormal returns preceding the repurchase program announcement indicates poor performance of the firm's stock, which supports the market undervaluation hypothesis. However, the market undervaluation hypothesis predicts a positive return following the repurchase announcement. This positive return should, at least partially, make up for the negative abnormal returns preceding the repurchase announcement. However the 10 post-announcement CAAR shows a negative CAAR for both indexes as well as the total sample.
5.2 Cross sectional regressions
In this section the results of the cross sectional regressions will be presented. The regression models, regression (10) and (11) in section 4.2, will be slightly adjusted to avoid multicollinearity issues. Therefore 5 regressions were needed to regress the initial model.
5.2.1 Regressions using AR[0]
First we will use the abnormal return on announcement date as dependent variable. The AR on announcement date shows a return of 0.103%. This return is almost statistically significant at the 10% level. The table below presents the results of the regressions.
[Insert table A7 here]
The regression output confirms the expected signs for the majority of the variables. In specific: ROA, dividendpayoutratio, size, MTBV, preCAR and fraction. However only the coefficients for size (partially) and fraction are statistically significant.
The agency costs of free cash flow hypothesis is tested using free cash flow (FCF), capital expenditures (capex) and return on assets (ROA). According to the theory, abnormal returns will be higher when the free cash flow is lower. A lower free cash flow implies that there is less money available that the manager can waste. Therefore the theory predicts a negative sign for FCF. However the regression output reports a positive coefficient for FCF. Similar the output reports a positive coefficient for capex, while the theory predicts a negative coefficient. The coefficient for ROA is according to the theory with a negative coefficient. This negative coefficient implies that less profitable firms will generate higher abnormal returns. These firms distribute excess cash to their shareholders which decreases the agency costs. However none of the coefficient is statistically insignificant
The capital market allocation hypothesis is tested using the percentage change of the capital expenditures using the current years capital expenditures as base (changecapex). According to the theory lower future capital expenditures reflects a lack of investment opportunities today. Instead of investing in bad projects the firms decide to distribute the excess funds to their shareholders using a share repurchase. The theory predicts that the distribution of excess funds today would lead to lower capital expenditures in the next year. Therefore a negative relation is expected between the abnormal return after the announcement and the increase in capital expenditures in the following year. However the regressions are inconclusive. Two of the regressions show a positive relation, while the other three show a negative relation. However all coefficients are statistically significant.
The leverage ratio of the firm (leverage) is used to test the capital structure adjustment hypothesis. According to the theory, firms with a low leverage ratio would use a share repurchase to increase their leverage ratio and in that way profit from the increased tax shield. Thus a negative relation is expected. The results however are inconclusive. The variable is included in 2 regressions, one reports a negative relation, the other regression a positive. Neither of the coefficient are statistically significant.
The firm's dividend payout ratio (dividendpayoutratio) is used to test the dividend substitution hypothesis. Firms that payout a high ratio of earnings as dividends are less likely to use a repurchase to distribute excess cash. Also firm that do not payout dividend have a higher information asymmetry. Therefore non-dividend paying firms are expected to generate higher abnormal returns. This is confirmed by the negative regression coefficients. The coefficients are not statistically significant.
The market undervaluation hypothesis is tested using the variables firm size (size), market-to-book value (MTBV) and pre-announcement abnormal return (preCAR). The theory predicts a negative coefficient for all three variables. Small size firms are expected to generate higher abnormal returns because of higher information asymmetry. The output of all five regressions confirm this expectation with a negative coefficient for firm size. Three of the five coefficients are statistically significant. Firms with a low market-to-book value ratio are perceived as value firms. Value firms are traded below their intrinsic value. Therefore these firms are expected to generate higher abnormal returns following a share repurchase announcement. The regression results confirm this expectation. However none of the coefficients is statistically significant. As shown in the previous section, repurchasing firms generate - on average - a negative cumulative abnormal return in the days before the repurchase announcement [-10, -2]. The undervaluation hypothesis assumes that firms that underperform in the period preceding the repurchas
