Do share repurchases matter in declining markets? An analysis of the S&P 500 firms Wenhao Chang

(1)

1

Do share repurchases matter in declining markets?

An analysis of the S&P 500 firms

Wenhao Chang1 Date 21 June, 2013

ABSTRACT

I study the influence of share repurchases on the monthly stock returns of S&P 500 listed firms over the period 2000-2012. The Fama and French (1993) three factor model is estimated with an additional repurchase factor to measure the sensitivity of this factor in declining markets. I find that a portfolio of top repurchasing firms outperforms portfolios of zero and low repurchasing firms. Share repurchases do matter to shareholders, though more in declining markets. A portfolio of the top repurchasing stocks outperform up to 0.65% more per month in declining markets than in advancing ones.

RESEARCH IN THE EFFECTS OF PAYOUT policy on stock return is abundant (Aharony and Swary, 1980; Masulis, 1980; Asquith and Mullins, 1983; Ikenberry, Lakonishok and Vermaelen, 1995). There are two main kinds of payout policies: dividends and share repurchases. The vast majority of the literature focusses on the effect of dividends, leaving the influence of share repurchases on stock returns underexposed. However, since the last decade share repurchases play an increasingly larger role in the payout policy of companies. This applies for both the U.S. and Europe (Grullon and Michealy, 2004; Skinner, 2008; von Eije and Megginson, 2008). According to Fama and French (2001) the proportion of listed U.S. firms paying dividends has decreased from 66.5% to 20.8% from 1978 to 1999. Nowadays share repurchases account for almost the same worth as cash dividends and some say it is even the dominant form of payout (Skinner, 2008). In figure I the total dividends and share repurchases over the

(2)

2

period 2000-2012 of the S&P 500 listed companies are presented together with the S&P 500 index. The amount of share repurchases in 2006 and 2007 confirms the statement of Skinner (2008). Even though share repurchases have been a dominant form of pay out policy in some years, the research of share repurchases still leaps behind that of cash dividends. Another reason why share repurchases draws attention is that managers often view repurchases as a more flexible payout instrument than dividends, since it does not implicitly commit the company to future payouts, it can be used in an attempt to time the equity market or to increase earnings per share (Brav et al, 2005; Skinner, 2008). A glance at figure I confirms that share repurchases are more flexible when looking at the financial crisis of 2008. The amount of share repurchases drops while the amount of dividends remains relatively stable. Figure I supports Stephens and Weisbach (1998) who state that stock repurchases are very pro-cyclical and in addition Dittmar and Dittmar (2008) show that the growth of share repurchases increases during economic expansion. Lastly, Boudoukh et al. (2007) find that adding repurchases to dividends as measurement for payout has statistically and economically significant predictability in time series, instead of using dividends only. The authors find that share repurchases play an important role in explaining the variation in stock returns.

The purpose of this paper is to look at the effects of share repurchases on stock returns and in particular this paper will look deeper into the effects of share repurchases on stock returns in declining markets. In other words, the timing of the share repurchases is questioned. Fuller and Goldstein (2011) find that paying dividends in a declining market matter to shareholders. Dividend paying stocks outperform non-dividend paying stocks by 1 to 2% per month more in a declining market than an advancing market. The subject has not been researched for share repurchases yet. This paper has several contributions to the current literature. First, I develop a share repurchase factor following the methodology of Boudoukh et al. (2007) and use this in combination with the Fama and French’s (1993) three factor model. Second, I specifically research the effect of share repurchases on portfolios of stock returns in declining markets and the difference with advancing markets. Third, I measure the sensitivity of the share repurchase factor with respect to declining markets. Lastly, I research what the effects are of the dividend and share repurchase tax change in 2003 and more in specific the sensitivity of the share repurchase factor before and after this tax change in declining markets.

(3)

3

influence of share repurchases on stock returns and what are the effects in declining markets?” The

result of this research complement the study of Fuller and Goldstein (2011) by adding the effect of share repurchases in declining markets.

Figure I. Cash Flows received by shareholders and the S&P 500 index 2000-2012

This graph includes the S&P 500 firms which have data available for common dividends (Datastream code WC04551) and share repurchases (year to year change treasury stock WC03499). The figure presents the cash flow received by the shareholders with the graph of the S&P 500 index. The left vertical axis is the cash flow per year in billions. The right vertical axis is the S&P 500 historical price index on January 1st in year t.

In the past decades, share repurchases has become an increasingly more popular tool to distribute funds to shareholders. However Jagannathan, Stephens and Weisbach (2000) point out that actual share repurchases are surprisingly difficult to measure. For example, one can measure the share repurchases by the announcement made by the repurchasing company. However the pitfall here is that companies are not obligated to actually do the repurchase. So these announcements do not accurately measure the total amount repurchased. Previous research often uses two kinds of methodologies in measuring share repurchases; the net or gross repurchases method. In this study the focus will be on the Fama and French’s (2001) net repurchases method (yearly change in treasury stock) to calculate the share repurchases and uses the methodology of Boudoukh et al. (2007) to measure the effect of share repurchases on stock returns.

I find that a portfolio of the top 30 percentile share repurchasing firms outperforms portfolios of zero and the lowest 30 percentile share repurchasing firms by more in declining markets than they do in advancing markets. My results show a monthly outperformance of approximately 0.4 to 0.65% in declining markets. Following Boudoukh et al. (2007), the three factor model with an additional factor is estimated to test the influence share repurchases have on stock returns in declining markets. This is

(4)

4

tested over a sample consisting of non-financial S&P 500 firms over the period 2000-2012. Furthermore, the effect of the tax cut on dividends in 2003 is tested. The results of the regressions are not significant and therefore the empirical evidence of the hypotheses remains inconclusive.

The remainder of this paper is organized as follows. Section 1 will highlight the relevant previous literature on share repurchases. What is the expected effect of share repurchases on stock returns and what are the unique characteristics of repurchases. Section 2 and 3 present the methodology and the data used. Furthermore section 4 describes the empirical results, while section 5 concludes.

I. Previous literature

This chapter reviews the underlying theories on the influence of share repurchases on the stock returns. Miller and Modigliani’s (1961) dividend irrelevance theorem indicates that in perfect markets investors would be indifferent between dividends or share repurchases as a payout mechanism. However, research indicates that investors react different on dividend announcements and share repurchases announcements (Masulis, 1980). Since there is abundant research on dividend, it is interesting to look at the motives and possible reactions of stock prices to share repurchases. The current literature describes several incentives for companies to conduct share repurchases and the effect on stock returns: signaling (Vermaelen, 1981; Zhang 2005), reducing the free cash flow available to managers (Grullon and Michealy, 2002), adjusting the capital structure of the company (Dittmar, 2000; Hovikimian, Opler and Titman, 2001), and personal taxation and regulatory changes.

A. Share repurchases as a credible signal for firm value

(5)

5

the stock is currently undervalued. Considering the positive reaction, the signaling theory implies that firms want to correct mispricing of their securities on the basis of favorable inside information. This argument is consistent with the view of Jagannathan, Stephens and Weisbach (2000) who find that firms repurchase stock following poor stock performance and increase dividends following good performance. On the effects of share repurchases, Grullon and Michealy (2004) find that there is a reduction of systematic risk after repurchases and the costs of capital decreases as well. More recent research on share repurchases and firm value from Zhang (2005) shows that Hong Kong firms making actual share repurchases have superior long term outperformance when benchmarked to their peer market-to-book firms the three year buy-and-hold abnormal return is 20% higher. Ikenberry, Lakonishok and Vermaelen (1995) find a similar result where they draw the conclusion that the market appears to ignore much of the information conveyed through share repurchases over the long run.

Brav et al. (2005) report survey evidence that managers of dividend-paying firms would eliminate their firms’ dividends if they could. Acharya et al. (2011) suggests that during a crisis dividends are important for a bank to signal its financial health. Most banks rely heavily on their reputation in capital markets and are therefore reluctant to cut dividends. This might be a reason why share repurchases could be a more credible signal for firm value when they continue or start repurchasing in declining markets. The signaling theory during declining markets would imply that the amount share repurchases should rise when stock prices are low and second, that the stock returns of the top repurchasing firms would outperform the benchmark. The first aspect does not seem to hold, because share repurchases seems to follow a pro-cyclical pattern over the period 2000-2012 (figure I); decreasing the amount of repurchases when the market is in a decline and increasing when the market is advancing. Dittmar and Dittmar (2008) find the same correlations with share repurchases and cyclical pattern in their research. Growth in share repurchases tends to follow the increase in stock market valuation. Therefore they conclude that it is unlikely that undervaluation explains repurchase activity. However, the question remains if stocks with high repurchase yield will outperform the zero repurchasing stocks in declining markets.

B. Share repurchases reduce the free cash flow available to managers

(6)

6

costs. Dittmar (2000) investigated how a firms’ cash flow affects its decision to repurchase stock. The author finds that in many periods share repurchases are used to distribute excess cash. Also share repurchases increase when the firm’s level of excess cash rises. The effect is that it reduces the probability that managers spend the excess cash in negative NPV projects. The result of Stephens and Weisbach (1998) supports the positive correlation between repurchases and the levels of cash flow. This suggests that firms actively adjust their repurchases to their cash position. In addition, Grullon and Michealy (2004) find that the stock market reaction to shares repurchase announcement is more positive among those firms who have the tendency to overinvest.

The existing findings provide strong evidence that support the free cash flow hypothesis for share repurchases. The application of this hypothesis is more difficult for dividends, since dividend payments are relatively stable over time and give little flexibility to adjust for an increase or decrease in levels of cash flow. This is also confirmed in the research of Fuller and Goldstein (2011) who rule out the free cash flow hypothesis as a possible explanation why dividend paying firms outperform non-dividend paying firms. Furthermore, in the Brav et al. survey (2005) nearly 90% of the managers view cuts of dividends as having negative consequences, whereas Dittmar and Dittmar (2008) find that only approximately 20% of the managers view share repurchases as having negative consequences. Therefore managers are more likely to cut in share repurchases than dividends. Much of the variation in the shareholder payout is observed in share repurchases and a determinant of the repurchases is excess cash. Thus an increase in aggregate repurchases reflects an increase in funds available to distribute to the shareholders. During declining markets this could be credible signal for investors since most firms cut in the amount share repurchases. Continuing or increasing the share repurchases could be a good indicator of a company’s view on their (future) cash flow and therefore the returns of these stocks could outperform the zero repurchasing stocks.

C. Leverage and capital structure

(7)

7

tax expenses. This gain of welfare is passed on to the shareholder and as a result the stock price is positively affected.

CEO’s sometimes state that they want to optimize their capital structure by funding a self-tender offer with newly issued debt (Lie, 2002). In addition, Dittmar (2000) also find that companies who repurchase, usually have a lower debt ratio than their industry peers. However, Vermaelen (1981) test this hypothesis and find that tax effects due to change in capital structure are not a predominant explanation for abnormal returns following a tender-offer. Though, he cannot reject the existence of tax effects of a tender-offer either. Hovakimian, Opler and Titman (2001) show the impact of repurchasing stock on leverage. The results of the paper indicate that firms may repurchase to increase their leverage ratio. There can also be negative effects of debt financed share repurchases; when earnings are less than expected, a firm can experience difficulties in repaying its debt and this could lead to a debt overhang problem (Myers, 1977). This is a situation when firms have to postpone positive net present value projects because the firm is unable to fund the project with neither internal financing nor debt financing. When this is the case, it could imply that share repurchases have a negative effect on stock returns in declining markets. Firms who use share repurchases to increase their leverage ratio could experience this debt overhang problem in declining markets.

D. Personal taxation and regulatory consideration

Another determinant of the corporate payout policy is the taxation system on dividends and share repurchases. In a perfect market with no taxes on both, the investors should be indifferent about the form of payout and the two are perfect substitutes to each other (Miller and Modigliani, 1961). However, in the real world there is a different taxation system on both forms of pay out. The difference in the U.S. is that dividends are taxed as ordinary income and share repurchases are taxed as capital gains. Therefore, the personal taxation could have an influence on the choice of disbursement method.

(8)

8

investor. For example, they do not change the type of payout when the majority of clientele consist of institutional investors, who enjoy tax advantages regarding cash dividends.

According to a dividend and share repurchase report of J.P. Morgan (2011) share repurchases provide greater tax efficiency for investors, but this difference is not as large as it used to be. In 2003 the “Jobs and Growth Tax Relief Reconciliation Act” was introduced in the U.S. This act lowered the taxes on cash dividends from 38.1% to 15%, removing the tax disadvantages of dividends2. Based on the change in tax regulation, one would expect that the amount of dividends paid would increase on a larger scale than share repurchases and that there would be a substitution effect away from share repurchases. A look at Figure I does not show evidence of this; the amount of share repurchases remained to increase and eventually overtook dividends in 2006. However, Julio and Ikenberry (2004) find that after the tax cut the fraction of dividend paying companies increases. Though the authors question the causality of the tax cut and they take other factors into account for the dividend increase. When we would apply this tax change in 2003 to share repurchases effects on stock returns in declining markets, one would expect that the share repurchases would have a larger quality signaling effect after 2003 than before 2003. I argue that share repurchases are now a relatively more expensive signal for the shareholders when compared to dividends. Therefore, I expect a higher sensitivity of stock returns to share repurchases in the declining years after the tax change. This implies that the difference between the monthly returns of top repurchasing firms and zero repurchasing firms will be larger in the declining markets in the years 2008 and 2011 than the declining markets of 2001-2002.3

E. Change in amounts of share repurchases in declining markets

There has been no research on the effect of share repurchases on stock returns in declining markets. All extant literature often write about share repurchase programs, the motive behind the repurchase or the market’s reaction to repurchases which is measured in firm value or stock returns. Former literature about share repurchases often does not research the effect on stock returns in different periods. Though, Stephens and Weisbach (1998) show that share repurchase activity is negatively correlated with prior stock returns, indicating that firms purchase their stock prices are perceived as low. This seems to

2

Statutory rates for dividends and capital gains are the same; however these taxes on share repurchases are imposed only when investors who choose to sell. Therefore share repurchases still have a slight tax advantage over dividends. (Julio and Ikenberry, 2004)

3

(9)

9

be in contrast with the findings of Dittmar and Dittmar (2008) who find that share repurchases waves are positively correlated with increasing market valuations. In figure I the amount of share repurchases experience large fluctuations in declining markets, while the dividends often remain stable in declining markets. Fuller and Goldstein (2011) did write a paper about dividends and its effects on stock returns in declining markets. They find that that dividend paying stocks outperform non-dividend paying stocks with 1 to 2% more per month in declining markets than in advancing ones. However, taking only dividends to describe the pay out, biases the results as share repurchases are a significant part of the payout policy of the past decade. Boudoukh et al. (2007) study the combination of both dividends and share repurchases combined and find that this combined payout has far more explanatory power than looking at dividends alone.

In the previous sections I indicated there are reasons why there could be positive or negative effects on stock returns of firms that continue to repurchase or do large repurchases in declining markets. Overall the share repurchases effect on stock returns is expected to be positive. When this positive effect on stock returns exists in practice than the returns of a portfolio of top repurchases should yield higher returns than a portfolio of firms with zero or low amount of repurchases.

F. Hypotheses

Based on my literature review I expect that there will be a positive relationship between share repurchases and stock return over the whole sample period. My first hypothesis will be based on the signaling theory, I hypothesize:

HYPOTHESIS 1: There is a positive relationship between share repurchases and stock returns.

My second hypothesis will look into the monthly returns of a portfolio formed on share repurchase yield and more in specific the monthly returns in declining markets. Based on the dividend research of Fuller and Goldstein (2011), my expectation is that there will be a stronger positive effect of share repurchases on stock return in declining markets than in advancing ones.

HYPOTHESIS 2: There is a stronger positive relation between share repurchases and stock

(10)

10

A third hypothesis will look into the effects of the tax change in 2003. After 2003, share

repurchases are a relatively more expensive signal when compared to dividends, so I expect that there is an increase in the quality of the signal of share repurchases. Therefore I hypothesize:

HYPOTHESIS 3: There is a stronger positive relation between share repurchases and stock

returns in declining markets after the tax change than in declining markets before the tax change.

II. Methodology

According to Fuller and Goldstein (2011) there are multiple interpretations and definitions of declining and advancing markets. An advancing market could be defined when the S&P 500 return exceeds the risk-free rate (Ang and Chen, 2002). Second, bull (bear) markets could be used as an indicator for advancing (declining) markets. Or third, business cycle classifications can be used like expansions (contractions) to define advancing (declining) markets. In all cases Fuller and Goldstein (2011) find substantially similar results. Therefore I select one measurement of declining markets. In this research I define a declining market where the S&P 500 index has a negative return on a yearly basis, measuring the return from January 1st in year t to December 31st in year t. When the S&P 500 index yields a positive return, this will be classified as an advancing market. Over the period 2000-2012 there were five years where the S&P 500 index had a negative return and are defined as declining markets: 2000, 2001, 2002, 2008 and 2011.

Fuller and Goldstein (2011) use the Fama and MacBeth (1973) style regressions to determine if dividend paying firms outperforms non-dividend paying firms in declining markets. While in this research I will use a modified Fama and French (1993) three factor model. Boudoukh et al. (2007) use this methodology and compute different factors for the dividends and shares repurchases and find that these factors are a significant predictor for portfolio of stock returns.

In equation 1 the traditional Fama and French (1993) three factor model is shown.

(1)

Where represents the excess return of a portfolio of stocks when the risk free rate is

(11)

11

market returns and is the market premium. The term is used to mimic the risk factor in returns related to size. It is the difference between the returns on small- and big-stock portfolios with about the same weighted average book-to-market equity. The term is used to mimic the risk factor in return related to book-to-market equity. It is the difference between the returns on high B/M portfolios and low B/M portfolios4.

Similar to the research of Boudoukh et al. (2007), this research will also compute and add another factor to the original Fama and French three factor model, namely share repurchases. This enables me to test whether share repurchases has a positive relationship with stock returns over the period 2001-2012 and draw a conclusion with reference to hypothesis 1. I will estimate the following equation:

(2)

Where the term (Share Repurchase Yield High Minus Zero) is the factor5_that

represents the return of the portfolio with a high share repurchase yield minus the return of the portfolio with zero repurchase yield. The portfolios are formed in July in year t and are reformed in July

t+1 based on their share repurchases. This is in line with the formation of the of the Fama and French (1993) three factor model as well as the dividend and share repurchases yield factor of Boudoukh et al. (2007). The factor is computed as the average monthly return of the top 30% high share repurchase yield group minus the average monthly return of the zero share repurchase group. This is a deviation6 from the approach of Boudoukh et al. (2007) who calculate the difference between the top 30% net payout minus the lowest 30% net payout. This second equation will research the difference between top repurchasing stock returns and non-repurchasing stock returns.

In order to allocate the stocks in the different portfolios I need to calculate the share repurchase yield of the firm’s stock. The amount of share repurchases can be calculated in various ways. The Fama and French (2001) and von Eije and Megginson (2008) method can be used to calculate the share

4

The right hand side variables are obtained from the website of Ken French. The

descriptive statistics are in Table AI, together with the factor. 5

Appendix A present the summary statistics of the four factors: market premium, SMB, HML and SRYHMZ. Table AI shows relatively high variance of the SMB, HML and SRYHMZ portfolios and low correlations with each other. This suggests that the 4-factor model can explain sizeable time-series variation and low cross-correlation imply that multicollinearity does not substantially affect the estimated 4-factor model loadings.

6

(12)

12

repurchase yield. Fama and French (2001) measures the so-called net share repurchases. They define share repurchases as the change in the firm’s Treasury stock or as the differences between repurchases and the stock issuance which is taken of the statement of cash flows (the latter is the retirement method7). Von Eije and Megginson (2008) use the cash flow approach to measure the share repurchases. This is called the gross repurchase method and they measure the reported amount of share repurchases on the firm’s cash flow statement. The use of net repurchases has the advantage that it nets out effects of employee stock options or repurchased stocks to finance a merger and this approach is applicable in an analysis on the stock returns of S&P 500 listed firms. For these reasons I will use the Fama and French (2001) method8.

The third equation uses dummies for declining markets. It contributes in reaching a conclusion regarding hypothesis 2 which researches the effect of share repurchases on stock returns in declining markets. I will apply slope dummy variables which will explain the interaction between the continuous variables and the dummy variable. This leads to equation 3:

)+ 2( )+ 3( )+ 4( )+ , (3)

Where are the different portfolio returns over the period 2001-2012. = 0 for the advancing

markets (2003-2007, 2009, 2010, 2012) and = 1 for the declining markets (2001, 2002, 2008, 2011) Lastly, a fourth equation will be used to test the sensitivity of the share repurchase factor to the tax change in 2003 in declining markets and enables me to draw conclusions regarding hypothesis 3:

)+ 2( )+ 3( )+ 4( )+ , (4)

7_{The retirement measurement is used when the treasury stock calculation has a negative outcome resulting from} subtracting the treasury stock of prior year with current year (Skinner, 2008). The repurchases are measured as the difference between stock repurchases and stock issuances from the statement of cash flows. Share repurchases are set to zero when the outcome is negative and also when the purchase of common stock is unavailable. 8

(13)

13

Where are the different portfolio returns of the declining markets in the years 2001, 2002, 2008,

2011. = 0 for the years before the tax change (2001, 2002) and = 1 for the years after the tax

change (2008, 2011).

III. Data selection and collection

The empirical evidence of this research is taken from Thomson Datastream. I extracted the return index (RI) of the S&P 500 listed firms over the period 2000-2012. The return index assumes that dividends are reinvested in the same equity. Before analyzing the data I transform the return index to a log return as follows:

Where is the return index of stock i in month t and is the return index of stock i in the

previous month. Furthermore, other items from Datastream and used in this research are: the stock’s corresponding Treasury stock for the net repurchase method (WC03499), common dividends (WC04551), common/preferred purchased, retired, converted and redeemed item which measures the repurchases and issuance from cash flow statement for the gross repurchases method/retirement measurement (WC04751), market capitalization (WC08001), S&P 500 Return Index (RI) for the identification of the declining/advancing markets.

Following standard practices, financial firms are excluded from the sample. Floyd, Li and Skinner (2008) find that the payout policies between industrials and financials differ greatly. Financials and banks continued to aggressively repurchase during the financial crisis. Increasing the leverage and shifting the prioritization of the transfer of wealth from debt holder to equity holder. Furthermore, the data was filtered from companies which had crucial missing data, such as market capitalization or returns during the sample period. This yields a sample of 351 firms with 54756 monthly observations of returns over the period 2000-2012. I order the stocks on the basis of the average share repurchase yield (yearly change treasury stock divided by the market capitalization) or when this is not available I apply the retirement measurement method. The top 30 percent of the share repurchasing firms, lowest 30 percent of the share repurchasing firms and the zero share repurchasing firms will be put in a separate portfolios9. Each portfolio’s monthly equal-weighted returns for July of year t to June of year t+1 are

9

(14)

14

calculated, and then the portfolios are reformed in July of year t+1. The summary statistics of these portfolios are displayed in Table I. The factor is computed as the average return across the yearly reformed top thirty percentile of the repurchasing firms minus the average return across the zero repurchase yield portfolio. The result of this procedure is monthly time-series for based on the net repurchase method.

The portfolios in Table I all have a higher average return than the market index S&P 500. There is a substantial difference between the mean return of the S&P 500 index and the top and zero

repurchases portfolios. An investment in the S&P 500 index would have yielded an average of 0.209% per month over the period 2001-2012, while the corresponding return of the top repurchases portfolio is 0.547% per month (neglecting the transaction costs for yearly rebalancing the portfolio). There is a monthly difference of 33 basis points per month between the returns of the S&P 500 index and the

Table I Summary statistics of the portfolios and S&P 500 index over the period 2001-2012

Portfolios Market Whole sample n=351 Top repurchases n=5410 Low repurchases n= 54 Zero repurchases n=171 S&P 500 index n=500 Mean 0.484% 0.547% 0.347% 0.518% 0.209% Median 0.986% 0.939% 0.905% 1.324% 0.883% Stdev 5.792% 5.689% 6.155% 5.997% 5.133% Min -22.11% -22.93% -24.57% -20.83% -18.20% Max 15.69% 17.55% 16.37% 15.89% 14.83% Skewness -0.827 -0.803 -0.924 -0.733 -0.765 Kurtosis 2.266 3.058 2.809 1.805 1.813 N Obs. 144 144 144 144 144

top repurchases portfolio. This would lead to an outperformance of approximately 4.1% on a yearly basis if one would invest in the latter. The difference lies in the fact that the S&P 500 index is a value weighted index while the portfolios are constructed on an equally weighted basis. The stocks included in the S&P 500 index are weighted according to the total market value of their outstanding stocks. This implies that the companies with a higher market capitalization have more influence on the average return of the S&P 500 index. Stocks with a relative larger market capitalization have been underperforming relative to stocks with a relative smaller market capitalization over the period

10

(15)

15

2012. This is in line with the findings of Fama and French (1993) who find that smaller firms outperform the large firms and measure this in their Small Minus Big (SMB) factor11. Because of this difference between the equally and value weighted portfolio, I have also included the average monthly returns of an equally weighted index with of the remaining S&P 500 listed firms (whole sample, N=351). Now the difference with the zero and top repurchases is smaller.

Furthermore, the low repurchases portfolio has a lower average return and higher standard deviation in comparison with the top and zero repurchasing portfolios. Looking at the median returns, the zero repurchases portfolio is the best performing portfolio which is not in favor of hypothesis 1.

To investigate how investor preferences for share repurchase vary across market conditions, I examine the returns of repurchasing and non-repurchasing stocks in declining and advancing markets separately. This is to research if the positive dividend effects of Fuller and Goldstein (2011) also exist for share repurchases. The forward returns were taken into account to measure these effects. For example, the top repurchases portfolio is formed on the basis of share repurchases done in 2003; subsequently the returns for 2004 are taken to measure the effects of the share repurchases in 2003.

Table II presents the monthly and median returns of the different portfolios in declining markets and advancing markets. I observe that the performance of the top repurchases portfolio is substantially better than the zero repurchases portfolio in declining markets when there is a difference between the average returns of 0.286% per month (-1.146% minus -1.432%). In advancing markets there are relatively smaller differences between the portfolio’s monthly returns. The zero repurchases portfolio has an average monthly higher return of 0.109% (1.503% minus 1.394%) in advancing markets. Looking at Table II, these descriptive statistics offer reasonable support for hypotheses 2. In addition, Appendix B presents the spread between the mean return of the top repurchases portfolio and lowest/zero repurchases portfolio. The results indicate that top repurchasing stock outperforms zero repurchasing stock with approximately 0.41% more per month in declining markets than in advancing markets. The spread between the top and lowest repurchase portfolio is even 0.65% per month.

(16)

16

Table II Monthly return in declining markets and advancing markets in percentages

This table presents the average and median (parentheses) monthly returns of the whole sample, top, zero and low repurchases portfolios and the S&P 500. The top, zero and low portfolios are ranked based on their share repurchase yield (net repurchases/market capitalization) and are formed in July in year t and the portfolios are reformed again in July t+1. Subsequently, the forward monthly returns are taken for two different periods: advancing markets (n obs. 96) and declining markets (n obs. 48). Advancing markets are 2003-2007, 2009, 2010, 2012 and declining markets are 2001-2002, 2008, and 2011.

Whole sample N=351 Top repurchases N=54 Low repurchases N=54 Zero repurchases N=171 S&P 500 N=500 Advancing markets 1.442 1.394 1.409 1.503 1.176 N Obs. = 96 (1.182) (1.138) (1.188) (1.694) (1.209) Declining markets -1.432 -1.146 -1.777 -1.452 -1.725 N Obs. = 48 (0.306) (0.353) (0.287) (0.055) (-0.891) IV. Results

In Tables III-VII, the four factor model regression results are presented of the different portfolios and S&P 500 index. The regressions are run over three different periods: column (1) the average excess return over the period 2001-2012, column (2) average excess return in advancing markets and column (3) average excess return in declining markets. Table III shows the regression results and monthly return over the whole sample. Tables IV-VI display the results for the different top, low and zero portfolios based on the yearly share repurchase yield. Table VII display the same for the S&P 500 index.

Turning to the results of the four factor model12, the columns (1-3) of Tables III-VI show the regressions with the market risk premium, size and value factors and the addition of the share repurchase factor. The addition of the fourth factor does not substantially change the factor loadings of the market, value and size factors when we compare this to the three factor regressions. Looking at the factor in Tables III-VI columns (1) we see that only the zero repurchases portfolio has a negative factor loading. Indicating that for the rest of the portfolios, the fourth factor has a positive influence on the portfolio’s return (factor loadings ranging from 0.084 to 0.609). This offers support for hypothesis 1, which state that share repurchases have a positive effect on stock return. However only for the top repurchases portfolio this result is significant (t-statistic = 6.209). Thus the result for hypothesis 1 over the whole sample is not supported. Furthermore, with the exception of the top repurchases portfolio (Table VI), the explanatory power (adjusted ) of the models has hardly changed or even decreased in comparison with the three factor model. Adding the fourth factor does not seem

12

(17)

17

to capture any extra variation in excess returns which was not already captured by the Fama and French three factor model. Continuing with the results of the advancing markets in Tables III-VI column (2), we see that the factor loadings of the factor have all decreased when compared to the whole period in column (1). For the whole sample (Table III) the sign for the factor loading over the period 20012012 and advancing markets goes from positive to negative (the difference between 0.084 and -0.044 is a decrease of 0.128). The other portfolios (Table IV-VI) experience a decrease of 0.128 to 0.160. This implies that the sensitivity to the factor is lower in advancing markets than in the whole sample and therefore I expect higher factor loading for the factor in declining markets. When we look at the results of the declining market in Tables III-VI column (3) this implication is supported. Looking at Tables III-VI column (3), we see that the factor loading of in declining markets is greater than the factor loading over the period 2001-2012 inn column (1) and the advancing markets in column (2). However, again only the factor loading of the top repurchases portfolio in Table IV is significant (t-statistic = 5.633).

***Insert Table III here*** ***Insert Table IV here*** ***Insert Table V here*** ***Insert Table VI here***

Next, I will test equation (3) described in the section methodology. In this dummy variable regression I measure the extra sensitivity to the factor in declining markets in relation with the whole sample period (2001-2012)13 and enable me to reach a conclusion regarding hypothesis 2. Tables III-VI, column (4) presents the results for the whole sample and the portfolios top, low and zero repurchases. For Tables III-VI I observe relatively high extra sensitivity in declining markets for the factor which has a factor loading between 0.304 and 0.344. This would indicate that the factor is positively related to the stock returns in declining markets, offering support for

(18)

18

hypothesis 2. However the results are not significant for a p-value < 0.05 (t-statistics between 1.520 and 1.633). Therefore, the results of the slope dummy variable regressions are inconclusive for hypothesis 2.

Furthermore, when we look at the results for S&P 500 index in Table VII, these results differ from the ones I find in Tables III-VI in the equally weighted portfolios. One example is the negative sign for the factor loading of the factor (-0.071) in column (4), which implies that the share repurchase factor has a negative influence on the share prices in a declining market. I conclude that the factor used in this research adds little explanatory power in capturing the variation in excess return of this value weighted index. The coefficients of the factor are close to zero (0.087 to 0.016) and are not significant (t-statistics range from 0.127 to 0.758) in all periods. Adding the fourth factor in an attempt to explain the returns even leads to a decrease in adjusted for the value weighted S&P 500 index.

***Insert Table VII here***

In sum, only the four factor model in Table IV columns (1-3) (top repurchases portfolio) shows that including the factor has significant additional explanatory power for excess returns which were not already captured by the Fama and French three factor model. This offers some support for hypothesis 1, which states that high share repurchases have positive influence on share prices. However, the factor is calculated with the returns of the top repurchases portfolio and therefore a high significance for the variable in the top repurchases portfolio is not counterintuitive. Furthermore, looking at the monthly returns in Table II, I observe an outperformance of the top repurchases portfolio over the whole sample period 2001-2012 and this difference with the other portfolios is even larger in declining markets. This offers support for hypothesis 2 which state that high share repurchases have positive effect on stock return in declining markets. I test hypothesis 2 with equation (3) which include slope dummy variables, the results are positive coefficients for the factor, indicating an extra sensitivity in declining markets. However, the difference between declining and advancing markets is not significant for all the portfolios.

Lastly, I want to measure the effect of the tax relief act introduced in 200314. This tax cut has eliminated the tax advantage of share repurchases over dividends. In Tables III-VII column (4) I already

14

(19)

19

find a positive relation between the factor and portfolio returns in declining markets. With equation (4) described in the section methodology, I test if there is a difference in sensitivity to the factor between the declining markets years before (2001-2002) the “Jobs and Growth Tax Relief Reconciliation Act 2003” was introduced and the declining markets years after (2008 and 2011) the tax change. Table VIII presents the results of the regressions with the slope dummy variables. I observe that in the years after the tax change the coefficient of the slope dummy variable is positive for all the portfolios (ranging from 0.054 to 0.266). This indicates that the returns of the portfolios have a stronger sensitivity to the factor after the tax regulation change. This offers support for hypothesis 3 which state that there is a stronger positive relation between share repurchases and stock returns in declining markets after the tax change than in declining markets before the tax change. With the dividend tax rate lowered to the same level as to capital gains, share repurchase have become a more expensive and higher quality signal for shareholders. Though, the results are not significant for a p-value < 0.05 and therefore the result for hypothesis 3 is inconclusive.

(20)

20 V. Conclusion

This paper studies the effect of share repurchases on stock returns and in particular the effects in declining markets are researched. The empirical evidence is based on information of S&P 500 listed non-financial firms over the period 2000-2012. In my sample, I document an increasing trend in the amount share repurchases done by S&P 500 listed companies. Just before the financial crisis in 2008, share repurchases overtook dividends as payout form, thereby underlining the importance of this kind of cash distribution to shareholders. Though, Dittmar and Dittmar (2008) indicate that share repurchases are correlated to business cycles and this is confirmed after 2008 when the amount of share repurchases experienced a large decline during the financial crisis, while the amount of dividends remained stable.

Research of Fuller and Goldstein (2011) find that dividend paying stocks outperform non-dividend paying stocks with 1-2% per month more in a declining market than in an advancing market. The question about the other form of payout, share repurchases, is left unanswered. To find if a same relation exists for repurchases an extended version of the three factor asset pricing model of Fama and French (1993) is used to measure the sensitivity of the share repurchase factor. Together with the Fama-French three factors (market risk premium, size and value) and a fourth factor I measure the effect of repurchase yield on a portfolio’s excess return.

(21)

21

Due to the change in tax rate for dividends in 2003, share repurchases have become a relatively more expensive payout method; as a consequence this change increases the quality of the signal. Implying that there should be a stronger relationship between stock return and share repurchases after the tax change. I find a stronger sensitivity to the SRYHMZ factor after the tax change in declining markets than before 2003 in declining markets. However the sensitivity to the factor is not significant for all the portfolios.

In sum, I conclude that shareholders are not indifferent to payout policy as suggested by Miller and Modigliani (1961). Instead, a portfolio with top share repurchasing stocks outperforms portfolios with zero and low repurchases, as well as the S&P 500 index in declining markets. Though, this difference is not as large as in the dividend research of Fuller and Goldstein (2011). Additionally, the tax change in 2003 has a positive effect on the relationship between share repurchases and stock returns. However, I do not find statistical significant results for the tests and the required support for my hypotheses. Different methodologies other than the four factor model based on the paper of Boudoukh et al. (2007) should be used to measure the effect of share repurchases on stock returns.

(22)

22 References

Acharya, V. V., Gujral, I., Kulkarni, N., and Shin, H. S. (2011) “Dividends and bank capital in the financial crisis of 2007-2009” NBER working paper. Available at SSRN 1789477.

Aharony, J., and Swary, I. (1980) “Quarterly dividend and earnings announcements and stockholders' returns: An empirical analysis.” The Journal of Finance. Vol. 35, 1, pp.1-12.

Ang, A., and Chen, J. (2002) “Asymmetric correlations of equity portfolios.” Journal of Financial

Economics. Vol. 63, 3, pp.443-494.

Asquith, P., and Mullins Jr, D.W. (1983) “The impact of initiating dividend payments on shareholders' wealth.” Journal of Business. Vol. 56, 1, pp.77-96.

Banyi, M. L., Dyl, E. A., and Kahle, K. M. (2008) “Errors in estimating share repurchases.” Journal of

Corporate Finance. Vol.14, 4, pp.460-474.

Boudoukh, J., Michealy, R., Richardson, M., and Roberts, M.R. (2007) “On the importance of measuring payout yield: Implications for empirical asset pricing.” The Journal of Finance. Vol. 62, 2, pp. 877-915.

Brav, A., and Graham, J.R., Harvey, C.R., and Michaely, R. (2005) “Payout policy in the 21st century.”

Journal of Financial Economics. Vol. 77, 1, pp.483-527.

Comment, R., Jarrell, G.A. (1991) “The relative signaling power of Dutch-auction self-tender and fixed- price self-tender offer and open-market share repurchases.” The Journal of Finance. Vol. 46, pp.1243-1271.

Dittmar, A.K. (2000) “Why do firms repurchase stock?” The Journal of Business. Vol. 73, 3, pp.331-355. Dittmar, A. K., and Dittmar, R. F. (2008) “The timing of financing decisions: An examination of the

correlation in financing waves.” Journal of Financial Economics.Vol. 90, 1, pp.59-83. Fama, E. F., and French, K. R. (1992) “The cross‐section of expected stock returns.” The Journal of

Finance. Vol.47, 2, pp.427-465.

Fama, E. F., and French, K. R. (1993) “Common risk factors in the returns on stocks and bonds.” Journal

of financial economics. Vol.33, 1, pp.3-56.

Fama, E. F., and French, K. R. (2001) “Disappearing dividends: changing firm characteristics or lower propensity to pay?” Journal of Financial economics. Vol.60, 1, pp.3-43.

Fama, E. F., and MacBeth, J. D. (1973) “Risk, return, and equilibrium: Empirical tests.” The Journal of

Political Economy, pp.607-636.

(23)

23

Fuller, K.P., and Goldstein, M.A. (2011) “Do dividends matter more in declining markets?” Journal of

Corporate Finance. Vol. 17, 17, pp.457-473.

Grullon, G., and Michaely, R. (2004) “Dividends, Share Repurchases, and the Substitution Hypothesis.”

The Journal of Finance. Vol. 57, 4, pp.1649-1684.

Hovakimian, A., Opler, T., and Titman, S. (2001) “The debt-equity choice.” Journal of Financial and

Quantitative analysis. Vol 36, 1, pp. 1-24.

Ikenberry, D., and Lakonishok, J., and Vermaelen, T. (1995) “Market underreaction to open market share repurchases.” Journal of Financial Economics. Vol. 39, 1, pp.181-208.

Jagannathan, M., and Stephens, C.P., and Weisbach, M.S. (2000) “Financial flexibility and the choice between dividends and stock repurchases.” Journal of Financial Economics. Vol. 57, 3, pp.355-384.

Jensen, M.C. (1986) “Agency costs of free cash flow, corporate finance, and takeovers.” American

Economic Review. Vol. 76, 2, pp.323-329.

Julio, B., and Ikenberry, D. L. (2004) “Reappearing dividends.” Journal of Applied Corporate Finance. Vol. 16, 4, pp. 89-100.

J.P. Morgan report (2011) on Dividends and Share repurchases. Available at:

http://www.jpmorgan.com/cm/BlobServer?blobkey=id&blobwhere=1158618771512&blobhead er=application%2Fpdf&blobcol=urldata&blobtable=MungoBlobs

Lie, E. (2002) “Do firms undertake self-tender offers to optimize capital structure?” The Journal of

Business. Vol. 75, 4, pp.609-639.

Lie, E., and Lie, H.J. (1999) “The role of personal taxes in corporate decisions: An empirical analysis of share repurchases and dividends.” The Journal of Financial and Quantitative Analysis. Vol. 34, 4, pp. 533-552.

Masulis, R.W. (1980) “Stock repurchases by tender offer: An analysis of the causes of common stock price changes.” The Journal of Finance. Vol. 35, 2, pp.305-319.

Miller, M.H. and Modigliani, F. (1958) “The cost of capital, corporate finance and the theory of investment.” The American economic review. Vol.48, 3, pp.261-297.

Miller, M. H., and Modigliani, F. (1961) “Dividend policy, growth, and the valuation of shares.” The

Journal of Business. Vol.34, 4, pp.411-433.

Myers, S. C. (1977) ”Determinants of Corporate Borrowing." Journal of Financial Economics. Vol.5, 1, pp.147-176.

Skinner, D. J. (2008) “The evolving relation between earnings, dividends, and stock repurchases.” Journal

(24)

24

Stephens, C.P., and Weisbach, M.S. (1998) “Actual share reacquisitions in open-market repurchase programs.” The Journal of Finance. Vol. 53, 1, pp.313-334.

Vermaelen, T. (1981) “Common stock repurchases and market signaling: An empirical study.” Journal of

Financial Economics. Vol. 9, 2, pp. 139-183.

Von Eije, H., and Megginson, W. (2008) “Dividends and share repurchases in the European Union”

Journal of Financial Economics. Vol. 89, 2, pp.347-374.

Von Eije, H., and Megginson, W. (2009) “Flexibility of dividend policies and shareholders’ returns in the European Union.” Available at SSRN 1342671

Zhang, H. (2005) “Share price performance following actual share repurchases.” Journal of Banking &

(25)

25

Table III Four factor model regression results (Whole sample)

The whole sample (n=351) consists of S&P 500 firms; the financial firms and firms with incomplete data are filtered. are equal weighted monthly returns for January in year t until December in year t. Advancing markets

are the years 2003-2007, 2009, 2010 and 2012. Declining markets are the years 2001, 2002, 2008, and 2011. N Obs. is related to the number of monthly (portfolio) returns in a period. Columns (1-3) present the four factor regression results. In column (4) the results are presented for the equation which includes a slope dummy variable for the declining years, where = 1 for declining markets and = 0 for advancing markets and the returns are over the period 2001-2012. The market risk premium, size and value factors can be obtained from Ken French’s website. The fourth factor is computed as the average monthly return across the yearly rebalanced top 30 percentile repurchase group minus average monthly return of the zero repurchases group. Asterisks * and ** denote significance at 5% and 1%, respectively. For comparison the adjusted for the Fama and French three factor model is added, these regressions are in Appendix D Table DI.

Columns (1-3) Four factor model:

Column (4) Slope dummy model:

Whole sample n=351 2001-2012 (1) Advancing Markets (2) Declining Markets (3) Declining Markets Slope Dummy (4)

Mean Monthly return 0.484% 1.442% -1.432%

(26)

26

Table IV Four factor model regression results (Top repurchases portfolio)

The top repurchases portfolio (n=54) is formed in July year t and rebalanced in July year t+1, this portfolio includes the yearly top 30 percent repurchasing firms. The forward returns (t+1) of this portfolio are taken for . The

forward returns are equal weighted monthly returns for January in year t+1 until December in year t+1. Advancing markets are the years 2003-2007, 2009, 2010 and 2012. Declining markets are the years 2001, 2002, 2008, and 2011. N Obs. is related to the number of monthly (portfolio) returns in a period. Columns (1-3) present the four factor regression results. In column (4) the results are presented for the equation which includes a slope dummy variable for the declining years, where = 1 for declining markets and = 0 for advancing markets and the returns are over the period 2001-2012. The market risk premium, size and value factors can be obtained from Ken French’s website. The fourth factor is computed as the average monthly return across the yearly rebalanced top 30 percentile repurchase group minus average monthly return of the zero repurchases group. Asterisks * and ** denote significance at 5% and 1%, respectively. For comparison the adjusted for the Fama and French three factor model is added, these regressions are in Appendix D Table DII.

Top repurchases portfolio n=54 2001-2012 (1) Advancing Markets (2) Declining Markets (3) Declining Markets Slope Dummy (4)

(27)

27

Table V Four factor model regression results (Low repurchases portfolio)

The low repurchases portfolio (n=54) is formed in July year t and rebalanced in July year t+1, this portfolio includes the yearly lowest 30 percent repurchasing firms. The forward returns (t+1) of this portfolio are taken for . The

forward returns are equal weighted monthly returns for January in year t+1 until December in year t+1. Advancing markets are the years 2003-2007, 2009, 2010 and 2012. Declining markets are the years 2001, 2002, 2008, and 2011. N Obs. is related to the number of monthly (portfolio) returns in a period. Columns (1-3) present the four factor regression results. In column (4) the results are presented for the equation which includes a slope dummy variable for the declining years, where = 1 for declining markets and = 0 for advancing markets and the returns are over the period 2001-2012. The market risk premium, size and value factors can be obtained from Ken French’s website. The fourth factor is computed as the average monthly return across the yearly rebalanced top 30 percentile repurchase group minus average monthly return of the zero repurchases group. Asterisks * and ** denote significance at 5% and 1%, respectively. For comparison the adjusted for the Fama and French three factor model is added, these regressions are in Appendix D Table DIII.

Low repurchases portfolio n=54 2001-2012 (1) Advancing Markets (2) Declining Markets (3) Declining Markets Slope Dummy (4)

(28)

28

Table VIFour factor model regression results (Zero repurchases portfolio)

The zero repurchases portfolio (n average=171) is formed in July year t and rebalanced in July year t+1. The forward returns (t+1) of the portfolio is taken for . The forward returns are equal weighted monthly returns for

January in year t+1 until December in year t+1. N Obs. is related to the number of monthly (portfolio) returns in a period. Columns (1-3) present the four factor regression results. In column (4) the results are presented for the equation which includes a slope dummy variable for the declining years, where = 1 for declining markets and = 0 for advancing markets and the returns are over the period 2001-2012. The market risk premium, size and value factors can be obtained from Ken French’s website. The fourth factor is computed as the average monthly return across the yearly rebalanced top 30 percentile repurchase group minus average monthly return of the zero repurchases group. Asterisks * and ** denote significance at 5% and 1%, respectively. For comparison the adjusted

for the Fama and French three factor model is added, these regressions are in Appendix D Table DIV.

Zero repurchases portfolio n=171 2001-2012 (1) Advancing Markets (2) Declining Markets (3) Declining Markets Slope Dummy (4)

Mean Monthly return 0.518% 1.503% -1.452%

(29)

29

Table VIIFour factor model regression results (S&P 500 index)

The term are the returns of the S&P 500 index, the returns were taken over the period 2001-2012. These

returns also include the returns of the financial firms included in the S&P 500. N Obs. is related to the number of monthly returns in a period. Columns (1-3) present the four factor regression results. In column (4) the results are presented for the equation which includes a slope dummy variable for the declining years, where = 1 for declining markets and = 0 for advancing markets and the returns are over the period 2001-2012. The market risk premium, size and value factors can be obtained from Ken French’s website. The fourth factor is computed as the average monthly return across the yearly rebalanced top 30 percentile repurchase group minus average monthly return of the zero repurchases group. Asterisks * and ** denote significance at 5% and 1%, respectively. For comparison the adjusted for the Fama and French three factor model is added, these regressions are in Appendix D Table DV.

S&P 500 n=500 2001-2012 (1) Advancing Markets (2) Declining Markets (3) Declining Markets Slope Dummy (4)

(30)

30

Table VIII Tax effect of share repurchases in declining markets

This table presents the returns of the different portfolios in declining markets. The four factor model regressions are run with a dummy for the years before the tax change. N Obs. is related to the number of monthly (portfolio) returns in a period, in all portfolios these are 48 months. These are the years before 2003 when the Tax relief act was signed. = 0 in the years 2001-2002 and = 1 in the years 2008 and 2011. Asterisks * and ** denote significance at 5%

and 1%, respectively. Declining markets (2001,2002,2008,2011) Whole sample (1) Top repurchases portfolio (2) Zero repurchases portfolio (3) Low repurchases portfolio (4) S&P 500 (5)

Mean Monthly return -1.432% -1.146% -1.452% -1.777% -1.725%

(31)

31 Appendix A

Table AI Descriptive statistics

This table reports the summary statistics of all the independent variables for the period July 2000-December 2012.

Mean (%) 0.140 0.357 0.570 0.235 Median (%) 0.780 0.100 0.275 0.332 stdev (%) 4.751 2.635 3.141 2.185 Max (%) 11.340 7.010 13.840 7.991 Min (%) -17.230 -6.610 -9.780 -5.854 Kurtosis 0.679 -0.173 3.348 1.992 Skewness -0.532 0.787 0.382 0.354 N Obs. 150 150 150 150

Table AII Correlations Matrix This table reports the correlations for the independent variables.

Variables (1) (2) (3) (4)

(1) 1 -0.336 0.383 -0.336

(2) -0.336 1 -0.22 0.152

(3) 0.383 -0.22 1 -0.207

(32)

32 Appendix B

Table BI Monthly returns of the portfolios and S&P 500 over 2000-2012, advancing markets and declining markets.

This table represents the average and median (parentheses) returns (1) in percentages of different portfolios, the whole sample (n=351) and the S&P 500 over the periods 2000-2012, advancing markets (2003, 2004, 2005, 2006, 2007, 2009, 2010, 2012) and declining markets (2000, 2001, 2002, 2008, 2011). The portfolios top repurchases (TR), lowest repurchases (LR) and zero repurchases (ZR) are formed on July in year t and are reformed in July year t+1 ranked on their share repurchases. The subsequent portfolios returns are taken to calculate the average and median (parentheses) over the different periods. Next I subtract the returns of the lowest repurchase (LR) portfolio, zero repurchases (ZR) portfolio, whole sample (WS) and S&P 500 (SP) from the top repurchases (TR) portfolio to find the differences (2) in returns.

(1) Returns (%) (2) Differences (%) Top repurchases (TR) n=54 Lowest repurchases (LR) n=54 Zero repurchases (ZR) n=171 Whole sample (WS) n=351 S&P 500 (SP) n=500 TR-LR TR-ZR TR-WS TR-SP All markets 0.547 0.347 0.518 0.484 0.209 0.200 0.029 0.063 0.338 (0.939) (0.905) (1.324) (0.986) (0.883) (0.034) (-0.385) (-0.047) (0.056) Advancing markets 1.394 1.409 1.503 1.442 1.176 -0.016 -0.109 -0.048 0.218 (1.138) (1.188) (1.694) (1.182) (1.209) (-0.05) (-0.557) (-0.045) (-0.071) Declining markets -1.146 -1.777 -1.452 -1.432 -1.725 0.630 0.306 0.285 0.579 (0.353) (0.287) (0.055) (0.306) (-0.891) (0.066) (0.298) (0.047) (1.244)

Table BII Difference between advancing markets and declining markets

In this table the difference of differences in Table BI (2) are presented. The difference in average (median) returns of the declining markets (DM) are taken and subtracted with the difference in average (median) return of the advancing markets (AM).

Differences AM-DM (%)

TR-LR TR-ZR TR-WS TR-SP

Average 0.646 0.414 0.334 0.361

Median (0.117) (0.855) (0.091) (1.315)

Table BIII Differences between before and after tax regulation (2003)

In this table the differences between the average and median monthly returns of the top and zero repurchases portfolio are presented. In 2003 a new regulation ended the tax advantage of share repurchases. To see the formation method of the portfolios look in the description of Tables IV and VI.

Declining markets 2001-2002 2008 and 2011

Mean Median Mean Median

Top repurchases portfolio 0.245% 1.088% -2.241% 0.325%

Zero repurchases portfolio -0.490% 0.032% -2.117% 0.380%

(33)

33 Appendix C

Table CI Share repurchases 2000-2012.

This table represents the share repurchase yield over the period 2000-2012. The share repurchase yields are calculated by taking the net share repurchases and dividing this amount by the firm’s market capitalization.

(34)

34 Appendix D

The Tables DI-DV presents the results of the three factor model regressions of Fama and French (1993) over the whole sample, three different portfolios and the S&P 500 index. The market premium, size and value factor can be obtained from Ken French’s website. There are three different periods over which the model is tested: the whole sample period (column 1), advancing markets (column 2) and declining markets (column 3). Advancing markets are the years 2003-2007, 2009, 2010, 2012 and declining markets are the years 2000-2002, 2008 and 2011.

Table DI Three factor model regressions (Whole sample)

Three Factor model:

Whole sample n=351 2000-2012 (1) Advancing market (2) Declining Market (3)

Factor loading t-statistic Factor loading t-statistic Factor loading t-statistic Intercept 0.001 0.433 0.003 1.372 -0.002 -0.436 1.030 22.537** 0.971 13.312** 1.049 14.438** -0.281 -3.092** -0.333 -2.898** -0.269 -1.549 0.233 2.946** 0.135 1.314 0.367 2.717** Adjusted 0.843 0.811 0.855 N Obs. 144 96 48

Table DII Three factor model regressions (Top Repurchases portfolio)

Top repurchases portfolio

n=54 2000-2012 (1) Advancing market (2) Declining Market (3)

Mean Monthly return 0.547% 1.394% -1.146%

(35)

35 Appendix D - Continued

Table DIII Three factor model regressions (Low Repurchases portfolio)

Table DIV Three factor model regressions (Zero Repurchases portfolio)

Table DV Three factor model regressions (S&P 500)

Low repurchases portfolio

n=54 2000-2012 (1) Advancing market (2) Declining Market (3)

Factor loading t-statistic Factor loading t-statistic Factor loading t-statistic Intercept 0.000 0.014 0.003 1.154 -0.003 -0.420 1.093 18.698** 0.954 11.863** 1.126 10.589** -0.253 -2.181** -0.448 -3.537** -0.102 -0.399 0.086 0.847 0.103 0.908 0.149 0.751 Adjusted 0.772 0.785 0.735 N Obs. 144 96 48

Zero share repurchase portfolio n=171 2000-2012 (1) Advancing market (2) Declining Market (3)

Mean Monthly return 0.518% 1.503% -1.452%

Factor loading t-statistic Factor loading t-statistic Factor loading t-statistic Intercept 0.001 0.486 0.003 1.481 -0.002 -0.452 1.060 22.509** 0.987 13.086** 1.089 14.942** -0.256 -2.719** -0.308 -2.594** -0.274 -1.572 0.297 3.647** 0.181 1.708 0.449 3.314** Adjusted 0.845 0.807 0.867 N Obs. 144 96 48

S&P 500 n=500 2000-2012 (1) Advancing market (2) Declining Market (3)