1
The Stock Performance after an Open-Market Repurchase
announcement in the Financial Crisis
An Empirical Study from 2007 to 2009
Abstract
In this paper the increased popularity of open market stock repurchases by US firms is researched in the time set of the recent financial crisis from October 2007 till September 2009. The drawn data sample from Bloomberg consists of 289 stock repurchase announcements. The relevance of the undervaluation-, the optimal leverage-, the excess capital – and the dividend substitution hypothesis are tested by performing event studies based on the market model method and the Fama-French three-factor model. The results are regressed against nine variables based on firm characteristics in two models to explain the market reaction. For the event window (-1,+1) the results show a significant positive cumulative abnormal return of 1.859% for the market model and 1.776% for the Fama-French three-factor model. The results from the regression analysis provide evidence that the abnormal returns are consistent with the undervaluation hypothesis, based on the firms market capitalization and the past stock price return.
Author: W.T.A. Schaaf - 5603714 E-mail: willem.schaaf@student.uva.nl Date: June 6, 2014
Supervisor: dr. P.J.P.M. Versijp University of Amsterdam Finance Department
2 Table of Contents
I. Introduction 3
II. Literature 6
III. Data and methodology 13
A. Obtaining the data 13
B. Event studies 14
C. Univariate analysis 16
D. Multivariate regression analysis 18
IV. Results 22
A. Descriptive statistics 20
B. Event studies 27
C. Univariate analysis 30
D. Multivariate regression analysis 36
V. Conclusion 42
VI. Bibliography 45
3 I. Introduction
In the past decade open-market stock repurchases have been the commonly used method by firms to repurchase their shares. The popularity of stock repurchases by firms has been growing ever since new legislation by the SEC was entered into force in the eighties. Studies investigating the effect on stock prices after a repurchase announcement report significant positive abnormal returns. Though there can be many causes in explaining these abnormal returns, the most cited one is associated with the undervaluation hypothesis. The paper of Comment and Jarrell (1991) finds that “firms tend to announce open-market repurchase plans following a decline in their share price when their stock is more likely to be undervalued”.
In the recent financial crisis, starting in 2007, the financial markets experienced steep declines as stock prices fell across the board. Reaching an all-time high of 1576.09 points on 11 October 2007, the S&P 500 index lost over 57% of its value within the next 18 months. On the 6th of March 2009 the S&P 500 index set an intraday low at 666.79 points. In this 18 month period trillions of dollars in stock market value were wiped out. The recent crisis is different compared to the dot com bubble burst in 2000. Even though the NASDAQ experienced declines over 70% in the 2000-2002 period, the broader based S&P 500 index lost close to 37% of its value. In this period mainly growth firms experienced declines, of which many went bankrupt in the later years. In the recent financial crisis both value and growth firms1 declined. Stock prices declining across the board has not been examined extensively on this topic. This thesis offers meaningful insights in the firm’s stock performance after an open-market repurchases announcement in this unique period of time. Taking in account the current stock prices, in which the S&P 500 index reached a new all time high in March 2013, the research period was presumably a period of market wide undervaluation2 for stock prices.
The objectives of this thesis are to study the stock performance of firms after a
repurchase announcement in the recent financial crisis. This stock performance is researched
by testing if the abnormal returns are significantly different from zero. The first hypothesis in this thesis is as follows: Firms announcing an open market stock repurchase experience
1 Value stocks are best described by characteristics as a high dividend yields, low price to book ratios and low price-to-earnings ratios. The opposite is the case for growth stocks. These are best described as stocks which usually don’t pay dividends, have high price to book ratios and prefer to reinvest retained earnings in capital projects.
2 Graph II shows a portrayal of the fall and subsequent recovery of the financial markets in the period 2006 till 2012. Between the fall and the recovery it is likely that many firms were dealing with undervalued stock prices despite overall lower market valuations and the general risk adverse attitude by investors.
4
significant positive (cumulative) abnormal returns. This hypothesis is tested by performing
event studies based on the market model method and the Fama-French three-factor model. In the academic literature the undervaluation hypothesis3 is the main explanation for the positive abnormal returns which stocks experience after an open market stock repurchase announcement. It implies that firms repurchase their shares to signal private information that their stock is currently undervalued. This is in general most likely firm- or sector specific undervaluation. Firms experiencing declining stock prices across the board, due to general fear on the financial markets and a lack of investor confidence, could also trigger management to signal undervaluation by announcing a repurchase plan. As investors were selling their shares and prices dropped all over the board with high market volatility, it caused market wide undervaluation and therefore offers meaningful insights in the explanation value of the undervaluation hypothesis and its relevance in the period of research.
The next set of hypotheses, to research the relationship between the abnormal returns and wide variety of company data, are as follows: The stock return after an open-market stock
repurchase announcement is linked to the level of cash, market capitalization, dividend yield, price-to-book-ratio, Tobin’s Q ratio, leverage, free cash flow yield, debt to equity ratio and the past return of the stock price. A multivariate regression analysis is performed to research
the hypotheses. The used variables in the regression are supported by empirical evidence from the academic literature and can give an indication which theories are consistent with the empirical results, and which are not.
The firm’s level of cash and free cash flow yield are related to the excess capital hypothesis. This hypothesis identifies that the distribution of excess cash to shareholders in lieu of dividends is another explanation for a stock repurchase. Stock repurchases offers firms more flexible ways of distributing excess capital than dividends (Dittmar, 2000). By announcing a repurchase plan there isn’t an expectation of recurrence and the size of the repurchase can easily be adjusted. Since the start of the crisis at the end of 2007 the money markets started to freeze and economic uncertainty increased. Then it became especially hard for firms holding limited quantities of cash to repurchase shares or continue to pay out dividends. This was in particular the case for growth stocks.
The firms leverage and debt to equity ratio are linked to the optimal leverage ratio hypothesis. This is another possible motive to announce a stock repurchase. It implies that firms use stock repurchases as a way of increasing their debt ratios. In the period of research
3 The undervaluation hypothesis is related to the signaling hypothesis and is more exhaustively discussed in the chapter II.
5
many firms had difficulties to (re)finance their debts. Since the start of the crisis there has also been a tendency for firms to rather deleverage than leverage. For growth stocks, which are mostly dependent on external funding or equity issuance, the optimal leverage ratio hypothesis is a less logical explanation to repurchase their stocks. This hypothesis is more applicable for value stocks. Firms with large reserves of cash and no needs to refinance their debt could see the optimal leverage ratio as a possible rationale behind announcing a stock repurchase.
The market capitalization of a firm is a variable related to the undervaluation hypothesis as Dittmar (2000) suggest that the market capitalization of a firm can be an indication of the possible information asymmetry. Large firms are, compared with small firms, covered by more analyst and financial press so there is more detailed information available for investors. When a small firm announces a stock repurchase it is more likely to be perceived as a signal of undervaluation because of the possible information asymmetry.
The stock performance in the past is another variable related to the undervaluation hypothesis. As previous literature described that firms tend to announce a stock repurchase when they perceive that their stock price is undervalued. The steep declines in the period of research could lead to stock prices which were out of sync with the actual economic value of the firm. For management this could be a logical reason to announce a stock repurchase and communicate the undervaluation of the firm’s stock price to the market. With the addition of the past return there is more insight in the relation between the abnormal returns and the stock price performance before the announcement.
The data in the research sample is drawn from Bloomberg. The period of research starts at the 1st of October 2007 and ends at the 30th of September 2009 for US stocks which are part of the S&P 500 index, NASDAQ 100 index and the S&P MID 400 index. The initial sample consists of 417 stock repurchase announcements in the period of research. All incomplete observations, missing daily stock prices for example, are excluded from the sample. A final sample of 289 observations is used for the empirical research in this study.
The results in this paper find significant positive cumulative abnormal returns for the three day event window (-1,+1) based on the market model method and the Fame-French three-factor model. The results are significant at the 99% level. The results of the multivariate regression analysis indicate evidence for the undervaluation hypothesis. The variables related to the optimal leverage hypothesis report significant coefficients, though the relationship between the coefficients and the cumulative abnormal returns are not in correspondence with the literature. Therefore the evidence is limited for the optimal leverage hypothesis. The
6
models based on the market model method and the Fama-French three-factor model are all significant at the 5% level in explaining the cumulative abnormal returns for the three day event window.
In the next section of this paper is the literature review. In this part the existing literature in the field of stock repurchases and the relevance of this paper is discussed. In part three the methodology of the event studies and regression analysis are described. The selected variables which are part of the empirical studies are explained. Part four start with the descriptive statistics and discusses the results of the empirical studies. Part five is the conclusion of the paper. The appendix contains extra data.
7 II. Literature
The academic literature about the price impact after a stock repurchase announcement is rich. In the past decades the subject has been exhaustively researched and most studies reported significant positive abnormal returns based on both short as long run event windows.
The list of motives for firms to repurchase part of their own publicly traded shares is lengthy. These motives include the undervaluation-, excess cash flow-, management entrenchment-, dividend substitution- and optimal leverage ratio hypothesis. All the described motives are plausible, though the most prevalent proclamation in this area of research is the undervaluation hypothesis. Previous research by Vermaelen (1981), Dann (1981), Ikenberry, Lakonishok and Vermaelen (1995), Stephens and Weisbach (1998) and Chan, Ikenberry and Lee (2004) all find strong evidence in support of the undervaluation hypothesis.
For firms there are three common methods to repurchase their shares; tender-offers -, Dutch auctions - and open-market repurchases. Till 1981 all stock repurchases were executed as a fixed price tender-offer (Vermaelen, 1981). This method specified in advance the tender price, the amount of shares sought and the expiration date of the offer. Shareholders have to decide whether they want to participate in the tender offer of the firm. If the offer is oversubscribed, the firm may purchase a number of shares bounded by the stated number sought.
A Dutch-auction offer specifies a range of prices in which the tendering shareholder chooses the minimum acceptable selling price. Compared to a tender-offer there is not a single offer price though the number of shares sought is also publicly announced. The shareholders inform the offering firm the number of shares they are willing to sell at their minimum price within the specified price range of the firm. The firm now determines the lowest price that will realize the number of shares sought. This price is then paid to all shareholders who tendered shares at or below the endogenously determined price.
Open-market repurchases have been the most popular method for firms to repurchase their own shares in the past two decades. In 1995 approximately 90 percent of the dollar value of all announced repurchase programs was performed by the open-market repurchases method (Ikenberry et al, 1995). In the following years up to now the percentage of open-market repurchases in regards to the total dollar value of announced repurchases has increased further. For firms announcing an open-market share repurchase there is not an obligation to complete the program or even buy back any shares as stated in the announcement. Typically the buyback of shares is over a long period of time, which could be up to a couple of years.
8
Beforehand there is also no certainty that the dollar value or the number of shares announced is actually repurchased by the firm. Since there is no obligation to actually repurchase the shares, which in fact means that it effectively gives management the option to buy back shares when they perceive the share price as undervalued. This option has value, even if the stock is not undervalued at the time of the announcement. It is not without reason that firms, especially when stock prices are at their highs, announce these buy-back programs. In correspondence with research by Ikenberry et al (1995) and Stephens and Weisbach (1998) it is shown that a significant fraction of firms that announce an open-market repurchase do not buy back any shares at all. This clearly states an important difference between the method of open-market repurchases compared to tender-offers and Dutch-auction offers.
For all three methods described, previous research finds significant, positive abnormal returns for firms that announce a share repurchase. Comment and Jarrel (1991) examined a sample of 1197 firms between 1984 till 1988 that announced to repurchase part of their own outstanding shares. The method of repurchase did show differences between the abnormal returns. The smallest effects were found for open-market repurchases with an average positive abnormal return of 2%. The positive abnormal return for fixed Dutch-auction offers in the study was 8% and for tender-offers the returns totaled 11%, the largest effect of the three described methods.
Research by Vermaelen (1981) and Ikenberry et al (1995) report abnormal returns of approximately 3% over a two-day open-market repurchase announcement. They reported that the stock repurchases primarily served as information signaling and thereby provides new information to the market. There was also evidence that in the long run there was an under reaction of abnormal returns for value stocks. The contrary was the case for growth stocks, where undervaluation is less likely to be an important motive. For these stocks there was no positive drift in abnormal returns observed over longer time periods.
Similar results were found by Nohel and Tarhan (1998), who, in order to observe differences in growth opportunities following share repurchases, used the Tobin’s Q to determine firm characteristics. They partitioned their sample into firms with a high and a low Tobin’s Q, where a low value of the Tobin’s Q indicates underinvestment. This resulted in higher abnormal returns for firms with a relatively low Tobin’s Q and sometimes even negative abnormal returns for firms with a relatively high Tobin’s Q.
Previous research by Ikenberry et al (1995) reports evidence for the undervaluation hypothesis in the nowadays so popular open-market stock repurchase. The undervaluation hypothesis assumes that there is asymmetric information between the managers, the insiders
9
of the firm, and the investors who are the outsiders. When managers perceive that the publicly traded stock of the firm is undervalued, they can repurchase shares. The repurchase announcement by management implies that the stock is trading below its true value. By repurchasing shares they act in the interest of long-term investors. When a firm announces an open-market repurchase, this should therefore be a positive signal about the value of the firm and the current price where the shares are trading. This is especially the case for smaller firms which stock is less covered by financial analysts. The performance of insider managers also varied across firm size and book-to-market ratios. The market responded most favorably to repurchases that are made by small and value firms (Hua Zhang, 2002).
The undervaluation hypothesis is also of interest in this study. The substantial decline of stock prices on a broad market level could, for management at some point in time, be an indication of undervaluation of their firm’s stock price. By looking at the stock prices now, we find the S&P 500 index at its all-time high. Taking in account the period of research, we had apparently a huge buying opportunity. The question of interest is if management can actually time their share repurchase and signal the undervaluation of the share price. Undervaluation is a more relevant concept when there is an economic crisis than in an economic boom, which implies that the undervaluation hypothesis is of great relevance in this research.
The excess cash flow hypothesis is another plausible reason for firms to repurchase their shares. By not distributing the excess cash in the form of dividends, a share repurchase offers a more flexible way for firms to distribute their excess cash. Especially with open-market stock repurchases there is no expectation of recurrence and the amount of cash used can easily be adjusted as there is no obligation to actually repurchase the sought shares. Studies by Nohel and Tarhan (1998), Dittmar (2000) and Grullon and Michaely (2004) find support for the excess capital hypothesis.
The excess cash flow hypothesis might have limited relevance in this specific period of time. With worldwide economic panic and a deep recession in the US, firms broadly speaking didn’t have excess cash available in comparison to the current environment of 2013. The lack of investment opportunities was probably not the main issue at that time as firms were more focusing on the successful continuation and survival of their current business operations.
The management entrenchment hypothesis focusses on management who uses share repurchases as a takeover defense. Management can use share repurchases to increase the debt-to-equity ratio and reduce the number of outstanding shares, hereby making the firm less
10
attractive as a takeover target (Lee, Ejara and Gleason, 2010). A study by Ajeet Lamba and Ian Ramsay (2000) reported results that out of 62 share repurchases from US listed companies, those with the intention to prevent hostile takeovers, share price reaction was negative. The opposite outcome was the result when repurchases did not have a takeover defense intention.
The management entrenchment hypothesis is another example of a less relevant hypothesis for this study. In the recent crisis the M&A dollar value in the US dropped over 50%4 from the high in 2007 to the low in 2009. Preventing hostile takeovers wouldn’t be the primary focus for management leading a company through a severe recession. There is also an interesting remark to be made regarding these numbers of M&A deal value. When the stock markets was at its peak, like in 2000 and 2007, the M&A activity is at highest. When the stock markets hits it lowest point, as it did in 2002 and in 2009, the M&A activity is at its lowest level. To actually test the management entrenchment hypothesis, the rationale behind the repurchase announcement needs to be researched. As it is not part of this study, the management entrenchment hypothesis will not be researched.
Research by Grullon and Michaely (2002) found results that support the dividend substitution hypothesis. In the past decades for firms in the US share repurchases became an important form of payout. The cash, that otherwise would have been used to increase dividends, is used to finance share repurchases. Particularly young firms have a higher tendency to pay out cash through repurchases and it has become the preferred way to initiate cash payouts. Large established firms have generally not cut their dividends, though these types of firms do also show a higher tendency to pay out cash through repurchases. The findings in their research indicate that firms have gradually substituted repurchases for dividends. With the economy experiencing its booms and bust in the past 15 years, firms are now more reluctant to volatility in their dividend policy. If a company announces a dividend cut in times of economic recession, investors will most likely be disappointed with as consequence a falling share price. Especially in the United States there is a tendency to consistency regarding dividends. A list of companies named the dividend aristocrats, including well-known names like Procter & Gambler, 3M, Coca-Cola Co. and Johnson & Johnson, show firms that have increased their dividend on a yearly basis in the past 50 years. This clearly demonstrates the emphasis on a consistent dividend policy. In the current market environment, with very low returns on market instruments and excess cash, firms are more
4 Source MergerStat. In 2007 the total value of all M&A activity was 1232.5 billion. In 2009 this dropped to 564.3 billion, a decline of over 50%.
11
attracted to use part of their large assets to fund a share repurchase than raising their
dividends.
The dividend substitution hypothesis is of interest in the researched period. As firms were cutting their dividends, it might be an option to substitute part of the dividend cuts with stock repurchases to satisfy the shareholders. In practice it is difficult to find out if firms substituted their dividend payout to stock repurchases, though the reported dividend yield could be an indicator. Besides, firms with no history of dividends were presumably more inclined to announce a stock repurchase instead of a dividend since there are no obligations to actually proceed with the repurchase.
The optimal leverage ratio hypothesis focusses on stock repurchases which are used as a way for firms to increase their debt ratios. When firms have a below target leverage ratio, it could result in a tendency of larger amount of share repurchases compared to firms with an optimal capital structure. When the repurchases are funded by additional debt, it reduces the equity and increases the debt. If management does not have the intention to increase the debt to equity ratio as a result of preventing a hostile take, the ratio is attributed to the optimal leverage ratio hypothesis. Research by Dittmar (2000) finds support of this hypothesis.
Adding debt to repurchase shares wouldn’t be the primary focus keeping in mind that refinancing debt was already becoming very difficult. Since the outbreak of the crisis firms have been more active at delivering and as the money markets freezed, only cash rich firms would be able to actually repurchase shares. Since an open-market stock repurchase announcement doesn’t lead to any legal obligation to actually repurchase the shares, it will be interesting to research the relationship between the debt-to-equity ratio and firm’s leverage to the announced share repurchase.
In the literature most of the studies related to the subject of share repurchases lay stress on the described undervaluation-, excess cash flow-, management entrenchment-, dividend substitution- and optimal leverage ratio hypothesis. These studies concentrate on the short and long term returns of firms who announce a stock repurchase. In this study, with the particular time set between 2007 and 2009, the relevance of the motives for firms to repurchase their shares is mixed. In such an exceptional period for the financial markets the effect on stock prices after a repurchase announcement hasn’t been researched. Ikenberry et al (1995) made in their study a distinction between firms based on company characteristics like the book-to-market ratio and the book-to-market capitalization. Though none of the used datasets include a severe bear market like the one experienced in the 2007 till 2009 period. In the dataset of Ikenberry et al the share repurchases announcements after the crash of October 1987 are excluded to
12
avoid having this unusual period dominating the study. In their findings a large amount of firms5 announced either new or increased repurchase programs largely in response to their low post-crash price. The crash of 1987 is also a unique moment in time for stocks as it happened in a very short time period of only a couple of days. This event is not comparable with the period of research in this study.
In the past decades US firms have been distributing increasing amounts of cash to shareholders through share repurchases. With open-market share repurchase programs representing the majority of these, the importance of this subject has increased. While earlier studies within this topic include data on US stocks that range primarily from 1980 till 1995, more recent studies have mainly focused on the Japanese, Canadian and European markets. This research is therefore an addition to the partly outdated studies on US stocks in the past decades and is performed in a unique period of time.
5 The total sample consisted of 1239 open-market repurchase announcement between January 1980 and December 1990. In the fourth quarter of 1987, following the stock market crash, 777 firms announced a share repurchase.
13 III. Data and methodology
In the data and methodology section a brief description is provided about the data that is used for this thesis. The empirical part of this study is divided into two parts. In the first part the abnormal returns after a repurchase announcement are calculated. The second part attempts to provide explanations for the market reaction by a univariate analysis and multivariate regression model. In the results section both empirical frameworks are presented with results.
A. Obtaining the data
The data in the research sample is drawn from Bloomberg. The Bloomberg database contains data for all US stock repurchases in the past twenty years. The data is drawn for the selected period of research, 1st October 2007 till 30st September 2009, for the US stocks which are part of the S&P 500 index, the NASDAQ 100 index and the S&P MID 400 index. These three indices are widely respected by investors and consist of 1000 stocks in total.
The S&P 500 is based on the market capitalization of 500 large companies having common stock listed on the NYSE or the NASDAQ. The companies which are part of this index are selected by a committee. There is also a minimum market capitalization requirement of at least $ 4.0 billion. This differs from the S&P MID 400, where a stock must have a total market capitalization that ranges between $ 750 million to $ 3.3 billion. The NASDAQ 100 is an index of the 100 largest non-financial companies listed on the NASDAQ. There is no capitalization threshold, though there are minimal requirements of daily average volume. These indices represent the US stock market regarding the research in the subject of stock repurchase announcements. The distinction between large and midcap companies is made between the S&P 500 and the S&P MID 400. With the addition of the NASDAQ 100 the growth stocks, which are mostly technology companies, are also represented in the data sample.
The initial sample consists of 417 stock repurchase announcements in the period of research. Out of this initial sample 128 announcements were excluded. All tenders offers, Dutch auctions and odd lot/small shareholders repurchases are not included in the used sample and only the open market stock repurchases are used. Since the number of announcements different than the open market repurchases is limited6, it has no statistical
6 A total number of eight observations were excluded as these were not the type of an open market stock repurchase announcements.
14
importance in using these announcements. All announcements which have a different date as the effective date compared to the announcement date are also excluded from the sample. The difference between these could be due to management approval of the repurchase plan. The rationale behind the removal of the observations has to do with the uncertainty involved with the actual effectiveness of the repurchase plan. At the moment of the announcement the stock might already experience abnormal returns.
Observations of firms which announced another repurchase within two months of a prior announcement are deleted. Specifically because some firms announced a repurchase after they actually bought their shares back on the market. In this way these firms communicated with their shareholders about the progress of their repurchase plan. With such a large amount of announcements within a short period of time it becomes impossible to choose the right date for performing the event studies described in the next section.
The observations within the sample which had another company related announcement on the same day, the day before or the day after the repurchase announcement are excluded7. When a company announces their quarterly earnings which also include a stock repurchase announcement, it is incalculable to determine which part of the returns are attributable to the repurchase announcement and which part to the earnings announcement. Another example could be a profit warning where the company announces a stock repurchase to compensate the weaker than forecasted results for the upcoming quarter.
The last deleted observations are those of incomplete data observations. The obtained data within Bloomberg has some missing closing prices for several observations. This might be due to external circumstances like trading halts by the SEC or due to data errors by Bloomberg. The final data sample used in the event studies and the multivariate regression model consists out of 289 observations.
B. Event studies
With the obtained data from the sample three event studies are performed. The abnormal returns are estimated by the market model method and the Fama-French three-factor model (Fama and French, 1993).
7 The date of the most recent company announcement (earnings (update), profit warning and corporate announcement) is compared to the stock repurchase announcement date. Bloomberg provides the date for these company announcements and this is then compared with the actual date of the stock repurchase announcement.
15
The abnormal returns are computed over the following three event windows:
- From -1 day to 1 day - Day 0
- From -10 days to 10 days
Hereby is day 0 the announcement day. The following hypothesis is formulated to research if the abnormal returns are significant:
Hypothesis I: Firms announcing an open market stock repurchase experience significant
positive abnormal returns.
To calculate the abnormal returns for the sample by using the market model at first the firm’s normal return is calculated. This model both captures the market trend and the firm's specific risk and assumes a linear relationship between the return of any stock and the return of the market portfolio. The following formula is used to calculate the normal return:
Ri,t = αi + βi Rm,t + εi,t (1)
In the model t stands for the time index, i = 1,2, ...., N stand for stock, Ri,t and Rm,t are the
returns on stock i and the market portfolio, respectively, during period t. The return in the market portfolio is measured by the variation in the difference benchmarks such as the S&P 500 index, NASDAQ 100 index and the S&P MID 400 index. The εi,t is the error term for
stock i.
The second model used to calculate the abnormal returns for the sample is the Fama-French three-factor model. This model captures three common risk factors in the stock returns, which are the overall market factor, factors related to firm size (SMB) and those related to book-to-market equity (HML). The following formula is used to calculate the abnormal returns:
Ri,t = αi + βi Rm,t + βi SMB + βih HML + εi,t (2)
In the model the t stands for the time index, i = 1,2,...., N stand for stock, Ri,t and Rm,t are the
16
the slope in the time series regression where SMB is related to firm size andHML is related
to book-to-market equity. The εi,t is the error term for stock i.
The abnormal return (ARi,t) of stock i on day t is the actual return minus the normal return as
expected with the described models.
ARi,t = Ri,t - αi - βi Rm,t (3)
The estimation period for the models starts 250 days prior to the event window. The cumulative abnormal return of stock i is calculated by adding up the abnormal returns over the event window and is defined as follows for the event window (-1,+1):
CARi= ARi, t=-1 + ARi, t=0 + ARi, t=1 (4)
To test for both event studies if the (cumulative) abnormal returns are significantly positive, a t-test is performed. The two sided t-test has the following hypothesis:
H0: The (cumulative) abnormal returns are not significantly different from zero.
H1: The (cumulative) abnormal returns are significantly different form zero.
The rejection regions are based on a 5% significance level: -1.96 < t > 1.96.
C. Univariate analysis
With the univariate analysis the relationship between the abnormal returns based on the market model method and nine firm specific variables8 is researched. The variables which are expected to be influential on the returns are split in two segments at the median9, hereby creating two individual segments of the variable, one below the median (LOW) and one above the median (HIGH). For both the segments the mean abnormal returns are calculated to study
8 The variables are CASHi, lnMARKETCAPi, DIVIDENDi, PBRATIOi, TOBINSQi, LEVERAGEi, CASHFLOWi, DEBTi and RETURNi and will more exhaustively discussed in the next part, the multivariate regression model.
9 Table XV in the appendix shows the results for the variables CASH, lnMARKETCAP and RETURN split at the top and bottom 30%. The results largely correspond with the findings of the univariate analysis where the variables are split at the median. The results for both the top and bottom analyses reports larger values.
17
the variables relationship on the market reaction after the open market repurchase announcement. With the variables split on the median the results will indicate whether the used variables have a significant impact on the (cumulative) abnormal returns of firms which are part of the sample. A two-sided t-test is performed per variable for the one and three day event window to test whether the segment’s mean abnormal returns are significantly different from zero. The hypotheses are as follows:
H0: The (cumulative) abnormal returns are not significantly different from zero for the
individual subgroup’s variable.
H1: The (cumulative) abnormal returns are significantly different from zero for the individual
subgroup’s variable.
The rejection regions are based on a 5% significance level: -1.96 < t > 1.96.
As described in the literature review, earlier studies found evidence that market reaction for firms with a large market capitalization is less positive compared to firms with a small market capitalization because of possible information asymmetry. By splitting the firms in two subgroups at the median, this univariate analysis will be a first step into understanding what the relationship is between the selected variables and the abnormal returns. Creating a group with a below median market capitalization and a group with an above median market capitalization provides possible evidence for the relationship between the selected variables and the abnormal returns.
To research if the market response is different between the two segments, a t-test is performed. The results of this test indicate whether the returns between the two subgroups are different from zero. The two sided t-test has the following hypothesis:
H0: The (cumulative) abnormal returns are not significantly different from zero between the
segments.
H1: The (cumulative) abnormal returns are significantly different from zero between the
segments.
18 D. Multivariate regression analysis
The multivariate regression analysis is used to research the relationship between the abnormal returns and wide variety of firm data. A cross-sectional regression analysis is performed and the regression is of the following form:
(C)ARi = ai + β1 * CASHi + β2 * lnMARKETCAPi + β3 * DIVIDENDi + β4 * PBRATIOi + β5 *
TOBINSQi + β6 * LEVERAGEi+ β7 * CASHFLOWi + β8 * DEBTi + β9 * RETURNi + εi
(5)
The used firm variables in the stated regression are drawn from Bloomberg. The next part provides a description of the variables and the rationale behind incorporation of these variables in the regression.
(C)ARi
The (cumulative) abnormal return is the dependent variable of the regression and it measures the stock performance after the repurchase announcement. The used event windows are (0,0) and (-1,+1). The (cumulative) abnormal returns are calculated as previously described and are based on the market model method and the Fama-French three-factor model.
CASHi
The variable CASH stands for all cash and near cash items divided by the market capitalization of the stock. It includes short term investments with maturities of less than 90 days and cash in vaults and deposits in banks. To actually repurchase stocks a firm needs to be able to have enough financial resources. The variable CASH is mainly related to the excess capital hypothesis. The expectation is a positive relationship between the level of cash and the abnormal returns.
lnMARKETCAPi
The variable lnMARKETCAP stands for the total market value of the firm. This is calculated by multiplying the total number of outstanding shares in the last quarter prior the stock repurchase announcement with the stock price. It is expected that the relationship between the variable lnMARKETCAP and abnormal returns is negative and this is related to the
19
undervaluation hypothesis. Small firms are less covered by equity analysts and are therefore more likely to be undervalued.
DIVIDENDi
The variable DIVIDEND stand for the dividend yield of the stock. The dividend yield is calculated by the most recent dividend per share divided by the stock price at the day of the announcement. If a firm doesn’t pay out any dividends, the dividend yield is equal to zero. The dividend substitution hypothesis suggest that firms finance their share repurchases with funds that otherwise would have been used to increase dividends. Since economic uncertainty was one of the main themes in the recent crisis, this creates the expectation that there is a negative relation between the market reaction and the dividend yield.
PBRATIOi
The variable PBRATIO stand for the price-to-book-ratio of the stock and is calculated by dividing the stock price at the announcement date by the latest quarter’s book value per share. A low price-to-book-ratio could indicate undervaluation of the firm’s stock price. A negative relationship between the abnormal returns and the price-to-book-ratio is expected. The variable is related to the undervaluation hypothesis.
TOBINSQi
The variable TOBINSQ stands for the ratio of the market value of a firm to the replacement cost of the firm’s assets. This ratio is useful for the valuation of a company. It is based in the hypothesis that in the long run the market value of company should be roughly equal the cost of replacing the company’s assets. The variable is related to the undervaluation hypothesis and is expected to have a negative relationship with the market reaction after a stock repurchase announcement as a relatively low Tobin’s Q could indicate undervaluation.
LEVERAGEi
The variable LEVERAGE stands for the firm’s financial leverage ratio which is measures by the average total assets divided by the average total common equity. The variable is added to test the optimal leverage ratio hypothesis and it is expected that firms who have below target levels of leverage will experience larger positive abnormal returns. The relationship between the abnormal returns and the variable LEVERAGE is expected to be negative.
20 CASHFLOWi
The variable CASHFLOW stands for the free cash flow yield which is calculated by the cash flow from operating activities minus the total capital expenditures divided by the share price at the announcement date. The relationship between the free cash flow yield and the abnormal returns is expected to be positive. The free cash flow yield is related to the excess capital hypothesis. This hypothesis is another reason for firms to repurchase their shares. By not distributing the excess cash in the form of dividends, a share repurchase offers a more flexible way for firms to distribute their excess cash.
DEBTi
The variable DEBT stands for the debt to equity ratio and is calculated as the sum of short and long term borrowings divided by total shareholder’s equity. The relationship between the abnormal returns and the debt to equity ratio is expected to be negative. Firm’s having a low ratio are most likely in a better position to finance stock repurchases by issuing debt. The variable is related to the optimal leverage ratio hypothesis.
RETURNi
The variable RETURN stands for the 20 day CAR prior to the announcement date. The event window is from day -21 to day -2. This variable measures the firm’s performance prior to the repurchase announcement and is related to the undervaluation hypothesis. A negative relationship is expected between the market reaction and the stock price past return.
The stated set of hypotheses tests whether the variables CASHi, lnMARKETCAPi,
DIVIDENDi, PBRATIOi,TOBINSQi, LEVERAGEi, CASHFLOWi, DEBTi and RETURNi
have a significant impact on the cumulative abnormal returns calculated by the market model method and the Fama-French three-factor model.
Set of hypotheses: The stock return after an open-market stock repurchase announcement is
linked to the level of cash, market capitalization, dividend yield, price-to-book-ratio, Tobin’s Q ratio, leverage, free cash flow yield, debt to equity ratio and the past return of the stock price.
21
The motives to add the variables are explained by the dividend substitution hypothesis, undervaluation hypothesis, the excess capital hypothesis and the optimal leverage ratio hypothesis which are more exhaustively described in the literature section.
22 IV. Results
The results section starts with the descriptive statistics of the used data sample. In the next part the event studies to calculate the abnormal returns are performed with an elaboration of the obtained results. The results of the event studies are computed over three different event windows. The next part is the univariate analysis and the last part of this section is the multivariate regression analysis to research the relationship between the abnormal returns and a wide variety of firm data.
A. Descriptive statistics
Table I shows the distribution on a monthly basis of the 289 open market stock repurchases observations which are part of the sample. On a yearly basis there are 112 observations in 2007, which only includes three months. The most of the repurchases in this sample have taken place in 2008 with a total of 154 observations, while in 2009 only 23 observations are registered. This is an interesting distribution because it shows that when the markets were at their lowest point, in March 2009, almost no firm announced a stock repurchase. The found results are similar by taking into account the actual dollar amount of stock repurchases. Since
Table I
The number of stock repurchase announcements per month
Data Number of announcements (%)
2007 2008 2009 January 22 7.61% 2 0.69% February 34 11.76% 0 0.00% March 12 4.15% 6 2.08% April 10 3.46% 4 1.38% May 17 5.88% 0 0.00% June 10 3.46% 5 1.73% July 10 3.46% 1 0.35% August 9 3.11% 1 0.35% September 13 4.50% 4 1.38% October 36 12.45% 9 3.11% November 43 14.88% 6 2.08% December 33 11.42% 2 0.69% All years 112 154 23 All years % 38.75% 53.29% 7.96%
The table reports the sample distribution on a monthly basis in the period October 2007 to September 2009. On average the most stock repurchase announcements were in 2007. In 2009, when the financial markets were at its lowest point, the number of announcements is low compared to the total sample
23
the third quarter of 2007 the dollar amount firms used to repurchase their shares drops over 80% to a low in the second quarter of 200910.
Graph I shows the distribution of stock repurchase announcements in the period of research. It is clearly visible that over time the number of announcements declines. This corresponds with the developments on the financial markets, whereby the declining stock prices seem to lead to less repurchase announcements by firms in the period of research.
Graph I
The distribution of stock repurchase announcement in the period of research
Graph II shows the performance of the benchmarks over the period January 2006 to December 2011. The returns are based on a start value of 100 in January 2006. The high of the market was reached in early October 2007. In the following months the markets fell and reached the lowest level in March 2009. In the period when markets declined the performance seemed for all three indices to be equal at a bottom-level, though the market recovery of the NASDAQ 100 and the S&P MID 400 unfolded at a much faster pace than for the S&P 500
10 Table XII in the appendix reports the actual dollar amount of quarterly stock repurchase by S&P 500 firms in the period between 2007 and 2009.
0 .05 .1 .15 .2 F rac tio n
01oct2007 01apr2008 01oct2008 01apr2009 01oct2009
24
index. At the end of the research period in September 2009 none of the indices were on the same level as in October 2007. In the years to come both the NASDAQ 100 and the S&P MID 400 surpassed their previous highs in 2010, while the S&P 500 continued to underperform compared to these indices.
Graph II
The development of the benchmarks from 2006 to 2011
The distribution between the three exchanges shows that most announcements were made from firms which stock is part of the S&P 500 index. The explanation here for is obvious as the S&P 500 index makes up 50% of the sample, whereas the S&P MID 400 counts for 40% and the NASDAQ 100 contributes 10%. The actual announcements shown in table II - , the distribution of these indices - is compared to the monthly distribution seen in table I more evenly spread with respect to the number of announcements per year.
Table II shows the number of stock repurchase announcements per index on a yearly basis. The last three months of 2007 have a total of 112 repurchase announcements. Out of the total research period almost 40% of the announcements took place in this relatively short period. In 2008, when the financial crisis unleashed, the number of observations totaled 154.
60 80 1 00 1 20 1 40 Pri ce 2006m1 2008m1 2010m1 2012m1 Date SP_MID400 NASDAQ100 SP_500
25
On a 12 month basis this is less than before. This is an interesting observation, because in October and November 2008, the period of the Lehman Brothers bankruptcy, the markets declined considerably. The number of repurchase announcement dropped to its lowest level since this crucial event.
Table II
The number of stock repurchase announcements per index
Data Number of announcements
2007 2008 2009 Total
S&P 500 65 83 11 159
NASDAQ 100 4 18 3 25
MID 400 43 53 9 105
All years 112 154 23 289
The table reports the sample distribution per exchange on a yearly basis. With a total of 1000 stocks in the sample the S&P 500 represents half of the sample. This is in line with the number of announcements from S&P 500 stocks. The same is the case for the NASDAQ 100 and the S&P MID 400 which shows a mainly even distribution between the exchanges.
The literature in the field of stock repurchases describes that firms tend to repurchase their shares when stock prices have declined and are more likely to be undervalued. Even though the financial markets didn’t recover yet, the average number of announcements in the sample dropped notably to under-five a month in 2009. After the market set its low in the beginning of March, the number of repurchase announcements continued to stay well under the 24 month average of almost twelve repurchases a month. From these results one could argue that many firms were too soon with their stock repurchases. The moment the market started to recover, they didn’t recognize the possible undervaluation of their stock price.
Table III shows the variables of the firm characteristics in the sample. The descriptive statistics are reported for the total sample and per exchange. These variables are used in both the univariate - and the multivariate regression analysis. The mean market capitalization for the sample is $ 17.289 billion. The difference between the S&P 500 and the S&P MID 400 is clearly visible. With a mean of $ 26.157 billon the S&P 500 stocks have a market capitalization which is over ten times larger than the mean market capitalization of the S&P MID 400 stocks.
The dividend yield, which has a mean value of 1.42%, differs between the exchanges. The stocks which are part of the S&P 500 index have the highest dividend yield of 1.70%. The NASDAQ 100 stocks, the smallest portion of the sample, pay on average the lowest dividend yield of 0.91%. As there are some fundamental differences between the stocks which
26 Table III
Descriptive analysis variables of firm characteristics
Total N Mean SD Min. Max. Skewness Kurtosis
Cash and Near Cash Items 289 2085.1 8174.47 0.18 121233 11.91 164.38 Market Capitalization 289 17289 33119 570 231909 4.033 21.505 Dividend Yield 289 1.417 1.688 0.00 13.23 2.39 12.74 Price-To-Book Ratio 289 3.552 3.159 0.52 26.09 3.54 21.13 Tobin’s Q Ratio 289 2.134 1.176 0.88 7.43 1.65 5.97 Leverage Ratio 289 3.707 4.223 1.06 39.50 4.55 30.51 Free Cash Flow Yield 289 6.029 6.695 -21.55 40.57 0.90 10.69 Total Debt To Equity 289 83.695 226.423 0.00 2708.03 10.33 116.89
S&P 500 N Mean SD Min. Max. Skewness Kurtosis
Cash and Near Cash Items 159 3162 10774.7 21.086 121233 9.169 95.777 Market Capitalization 159 26157 38215.6 2038.71 229854. 2.990 12.706 Dividend Yield 159 1.698 1.867 0.000 13.232 2.472 12.888 Price-To-Book Ratio 159 3.801 3.495 0.571 26.090 3.795 22.091 Tobin’s Q Ratio 159 2.120 1.084 0.892 6.159 1.196 3.895 Leverage Ratio 159 4.255 5.054 1.121 39.500 4.178 24.375 Free Cash Flow Yield 159 6.441 7.439 -21.548 40.570 1.314 10.791 Total Debt To Equity 159 107.43 298.996 0.000 2708.02 7.908 67.368
NASDAQ 100 N Mean SD Min. Max. Skewness Kurtosis
Cash and Near Cash Items 25 2596.6 3950.5 50.65 17811 2.799 10.574 Market Capitalization 25 22531 45256.7 1508.92 231909 4.198 19.906 Dividend Yield 25 0.912 1.338 0.000 5.439 1.871 6.524 Price-To-Book Ratio 25 4.980 3.671 0.982 17.033 1.868 6.411 Tobin’s Q Ratio 25 2.941 1.641 1.190 7.391 1.164 3.397 Leverage Ratio 25 2.365 1.516 1.173 8.425 2.885 11.830 Free Cash Flow Yield 25 5.739 6.112 -16.461 16.121 -1.694 8.273 Total Debt To Equity 25 45.243 41.329 0.000 168.885 1.369 4.755 S&P MID 400 N Mean SD Min. Max. Skewness Kurtosis Cash and Near Cash Items 105 332.6 364.297 0.176 2596.61 3.043 16.945 Market Capitalization 105 2612.5 1365.49 570.898 7308.84 0.963 3.822 Dividend Yield 105 1.111 1.377 0.000 5.708 1.644 5.428 Price-To-Book Ratio 105 2.835 2.207 0.515 13.383 2.383 9.571 Tobin’s Q Ratio 105 1.964 1.111 0.879 7.425 2.211 8.980 Leverage Ratio 105 3.189 2.970 1.064 18.442 3.059 13.067 Free Cash Flow Yield 105 5.478 5.564 -13.478 23.088 -0.143 5.600 Total Debt To Equity 105 56.475 56.262 0.000 377.177 2.366 12.433 The table reports the company characteristics of the firms in the sample as total and per exchange. The mean, standard deviation, the minimum and maximum values are reported in the table. The observations of the variables are registered on the event day when the stock repurchase announcement is made. The dollar values are denoted in billions.
are part of these indices, the NASDAQ 100 stocks are best described as growth stocks compared to the value stocks which are part of the S&P 500 index.
The price to book ratio varied between 1.75 and 4 for the S&P 500 index between 2000 and 2010. The highest values were registered at the highs of the market in 2000 and the lowest in the recent financial crisis in 2008. In the used sample we find an average price to book ratio of 3.55, which is at the top of the range. Between the used indices the lowest price to book ratios are on average found for the S&P MID 400 observations while the highest are
27
these of the NASDAQ 100. The explanation makes sense as growth stocks deal with different valuation methods than smaller sized value stocks like the ones from the S&P MID 400.
Another interesting variable is the total debt to equity ratio. Between the three exchanges we have clear differences. The mean value for the S&P 500 is 107.43 and thereby the highest which indicate that the S&P 500 stocks have larger debt to equity ratio’s than the stocks from the other exchanges. The difference is most notably visible for the S&P MID 400 observations where the debt to equity ratio is 56.48.
B. Event studies
For the event windows (0,0), (-1,+1) and (-10,+10) the abnormal returns are calculated by the market model method and the Fama-French three-factor model. The total sample of 289 stock repurchase announcements is used to calculate the market reaction. The calculated abnormal returns per exchange for both the models are reported in the appendix11.
Table IV reports12 the summary statistics for the abnormal returns for the described periods calculated by the market model method.
Table IV
Abnormal returns market model method per event window
Event Window (0,0) (-1,+1) (-10,+10) N = 289 Mean 0.856% 1.829% -0.244% Standard deviation 0.046 0.057 0.133 T-statistic 3.172 5.455 -0.312 P-value 0.001 0.001 0.622 Lowest CAR -12.82% -18.67% -85.12% Highest CAR 31.29% 43.37% 41.69%
Number of positive observations 169 191 148
Number of negative observations 120 98 141
The table reports the (cumulative) abnormal returns in percentages surrounding the stock repurchase announcement from the total sample in the period October 2007 – September 2009. The abnormal returns are calculated by the market model method and the parameters of the model are estimated over 250-day period prior to the selected event window.
11 Table XIII and table XIV report the cumulative abnormal returns per exchange for the three day event window based on the market model and the Fama-French three-factor model.
12 In the appendix graph III provides a visual reproduction for the market reaction of the event windows (-10, +10).
28
For both the one and three day event windows a significant positive market reaction is reported. The average abnormal return is 0.856% for the event window (0,0) with 169 of the observations showing a positive market reaction and 120 observations showing a negative market reaction. The reported t-statistic for the one day event window is 3.172 and exceeds the 5% significance level of t > 1.96. For the three day event window the average cumulative abnormal return is 1.829%. The number of positive announcements increases to 191 which is almost equal to two-thirds of the total sample size. The t-statistic for the reported cumulative abnormal return is 5.455. The reported p-value of 0.001 rejects the null hypothesis that the cumulative abnormal returns are not different from zero. The event window (-10,+10) doesn’t report any significant cumulative abnormal returns with a p-value of 0.651, therefore the null hypothesis is not rejected. The average cumulative abnormal return of -0.244% implies that in this particular event window no positive significant abnormal returns are realized at the 5% significance level.
Next to the calculation of the market reaction by using the market model method, the Fama-French three-factor model is used for the same event windows. Table V reports the results for this method of research. The results show significant positive abnormal returns for the one day event window. The mean abnormal return of 0.806% is with a t-statistic of 3.015 significant at the 5% significance level. The null hypothesis that the abnormal returns are not different from zero is therefore rejected. Compared to the market model the abnormal return is slightly lower though there is an increase in the number of positive observations. Comparable results are shown for the three day event window. The average cumulative abnormal return is 1.776% and the null hypothesis is also rejected at the 5% significance level. The p-value of 0.001 is low and therefore the results are highly significant. Different results are found for the event window (-10,+10). The average market reaction is 0.303% and the t-statistic is not significant at the 5% level with a reported value of 0.379. The null hypothesis is not rejected and the returns for this event window are not different from zero.
While not significant, the difference in the average cumulative abnormal return between the market model method and the Fama-French three-factor model is most notably visible for the event window (-10,+10). The results from the market model method show a negative market reaction of -0.244%, while these of the Fama-French three-factor model show a positive market reaction of 0.303%. These results indicate that if the event window covers a larger period of time, the results from the models start to diverge13.
13 In the appendix graph V till VII will show this occurrence where difference between the cumulative abnormal returns start to diverge when the event window covers a larger period of time.
29 Table V
Abnormal returns Fama-French three-factor model per event window
Event Window (0,0) (-1,+1) (-10,+10) N = 289 Mean 0.806% 1.776% 0.303% Standard deviation 0.456 0.057 0.136 T-statistic 3.015 5.282 0.379 P-value 0.001 0.001 0.353 Lowest CAR -13.93% -20.50% -88.47% Highest CAR 27.13% 42.36% 35.76%
Number of positive observations 173 190 161
Number of negative observations 116 99 128
The table reports the (cumulative) abnormal returns in percentages surrounding the stock repurchase announcement from the total sample in the period October 2007 – September 2009. The abnormal returns are calculated by the Fama-French three-factor model and the parameters of the model are estimated over 250-day period prior to the used event window.
The results for both the market model as the Fama-French three-factor model indicate that
companies announcing an open market stock repurchase experience significant abnormal returns for the one- and three day event window. Both the periods and models find results that
the market reaction is significant at the 99% level and therefore the null hypothesis that the abnormal returns are equal to zero is rejected. For the event window (-10,+10) the results indicate that no significant positive cumulative abnormal returns have been achieved. The short period following the announcement date shows a negative cumulative abnormal return. After ten days these abnormal returns stabilize. There is no clear economic reason behind these negative abnormal returns in the days following the announcement date. Research by Vermaelen (1981) has similar findings for the period of time. Possible explanations could include short selling activity and uncertainty about the actual amount of shares the firm is going to repurchase. In the event window (+2,+21) we find corresponding results which show stabilizing cumulative abnormal returns after ten days. For the event window (+2,+252) the findings show a clear upward drift in the abnormal returns14 over the whole time period after twenty days. One could argue that after the first twenty days the firm starts to actually repurchase shares on the market which has a positive effect on the stock price. Since in this study the actual share repurchases are not taken in account, there remains uncertainty in this matter.
14 In the appendix graph VIII shows the event window (+2,+252) where after twenty-five days the cumulative abnormal returns start to become positive and this trend continues over the whole period of time.
30
The found results for the event windows (0,0) and (-1,+1) are in line with the outcomes of papers by Vermaelen (1981) and Ikenberry et al (1995). The average positive market reaction after an open-market stock repurchase announcement is between 0.806% and 0.856% for the one day event window depending on the chosen model. For the three day event window the results are almost double as the average market reaction ranges between 1.776% and 1.829% depending on the chosen model.
C. Univeriate Analysis
The univariate analyses provide more insight in the individual company characteristics in relevance to the reported market reaction. The selected variables CASH, lnMARKETCAP, DIVIDEND, PBRATIO, TOBINSQ, LEVERAGE, CASHFLOW, DEBT and RETURNare split at the median and the abnormal returns of the event windows (0,0) and (-1,+1) are based on the market model method. The variables are split between a low and high segment, indicating the observations are below or above the median value of the variables.
The results from the categorization into groups are based on the variables with the expectation that these are relevant in explaining the abnormal returns. The reported t-value for the difference between the groups in the table tests whether there is any statistical significant difference in the market reaction between the two groups. The null hypothesis indicates that there is no significant difference in the abnormal returns for the divided – low/high median - segments of the variable. The used rejection region is at the 5% significance level: -1.96 < t > 1.96. For each analysis the total of 289 observations is split in the two segments and the reported median abnormal return for the total sample is 0.85% at the event window (0,0). For the event window (-1,+1) the median cumulative abnormal return is 1.74% which also matches closely to the mean cumulative abnormal return of the three day event window at 1.83%.
The variables are ordered according to the relevant hypothesis earlier described in section III.D. These hypotheses are the undervaluation hypothesis, the excess capital hypothesis, the optimal leverage hypothesis and the dividend substitution hypothesis.
Table VI reports results with the categorization based on the firm’s cash ratio and the free cash flow yield. These variables are added to test the excess capital hypothesis. The results for the variable cash show that firms in the segment with a below median cash ratio have a slightly higher abnormal return than firms with an above median cash ratio for the
31 Table VI
Univariate t-test analysis for the variables cash and free cash flow yield
CASH
Event Window (0,0) (-1,+1)
Low High Low High
T-value for difference between groups 0.65 0.01
N 289 289 289 289
Number of observations 145 144 145 144
Median whole sample 0.85% 0.85% 1.74% 1.74%
Median 1.20% 0.55% 1.76% 1.68%
Mean 1.03% 0.69% 1.83% 1.82%
Standard deviation 0.0377 0.0529 0.0495 0.0638
T-statistic 3.29 1.54 4.46 3.43
P-value 0.001 0.124 0.001 0.001
FREE CASH FLOW YIELD
Event Window (0,0) (-1,+1)
Low High Low High
T-value for difference between groups 0.25 -0.85
N 289 289 289 289
Number of observations 145 144 145 144
Median whole sample 0.85% 0.85% 1.74% 1.74%
Median 1.08% 0.75% 1.75% 1.68%
Mean 0.92% 0.79% 1.54% 2.11%
Standard deviation 0.0446 0.0473 0.0526 0.0611
T-statistic 2.49 2.01 3.52 4.16
P-value 0.014 0.045 0.001 0.001
The table reports the results for the subgroups CASH and FREE CASH FLOW YIELD which are split at the median from the total sample of 289 observations. The (cumulative) abnormal returns are calculated by the market model method and regressed on the event windows (0,0) and (-1,+1).
event window (-1,+1). The mean returns of 1.83% and 1.82% indicate that there is no evidence that the results between the segments are significantly different from zero as the reported t-value of 0.01 is not significant on the 5% level. The individual t-tests do show that the variable has significant explanatory value on the abnormal returns at the 5% level with t-statistics of respectively 4.46 and 3.43 for the segments below and above the median. For the variable free cash flow yield the results are similar. For both event windows there is no evidence that the abnormal returns between the below and above median group are significantly different from zero. The result does show a larger positive abnormal return of 2.11% for the segment above the median compared to 1.54% for the segment below the median, but the difference is not significant. The reported the t-statistics for all the segments are significant at the 5% level for the one day and three day event window. The results therefore indicate evidence for the excess capital hypothesis.
32
Table VII reports the results based on the categorization of the firm’s leverage and the debt to equity ratio. The optimal leverage hypothesis is tested by these two variables. For the variable leverage the results show that the segment below the median experience higher abnormal returns in both event windows compared to the segment above the median. These results are in correspondence with the literature. All the reported individual group t-statistics are significant at the 5% level. Despite the difference, the reported t-statistics for the whole sample are less than the critical value of 1.96 so there is no significant evidence that the returns between the segments are different from zero. This is the case for both event windows. The results for the variable debt to equity ratio match the results of the variable leverage.
Table VII
Univariate t-test analysis for the variables leverage and debt to equity
LEVERAGE
Event Window (0,0) (-1,+1)
Low High Low High
T-value for difference between groups 0.46 0.96
N 289 289 289 289
Number of observations 145 144 145 144
Median whole sample 0.85% 0.85% 1.74% 1.74%
Median 1.04% 0.76% 2.41% 1.34% Mean 0.98% 0.73% 2.16% 1.51% Standard deviation 0.0457 0.0462 0.0525 0.0610 T-statistic 2.56 1.93 4.89 3.01 P-value 0.011 0.055 0.001 0.003 DEBT TO EQUITY Event Window (0,0) (-1,+1)
Low High Low High
T-value for difference between groups 0.59 0.69
N 289 289 289 289
Number of observations 145 144 145 144
Median whole sample 0.85% 0.85% 1.74% 1.74%
Median 1,00% 0.76% 2.15% 1.52%
Mean 1.02% 0.70% 2.07% 1.60%
Standard deviation 0.0527 0.0382 0.0663 0.0464
T-statistic 2.30 2.22 3.71 4.18
P-value 0.022 0.027 0.001 0.001
The table reports the results for the subgroups LEVERAGE and DEBT TO EQUITY which are split at the median from the total sample of 289 observations. The (cumulative) abnormal returns are calculated by the market model method and regressed on the event windows (0,0) and (-1,+1)
Firms in the segment below the median debt to equity ratio experience higher abnormal returns than firms in the segment above the median. These difference are not significantly different from zero based on the reported group t-values of respectively 0.59 and 0.69 for the
