Stock Repurchase as an Alternative to Dividend Payout: Evidence. from the Warsaw Stock Exchange

Transcription

1 Stock Repurchase as an Alternative to Dividend Payout: Evidence from the Warsaw Stock Exchange Sebastian Gryglewicz Poznan University of Economics, Poland March 16, 2004 Abstract The paper investigates market reaction to announcements of stock repurchases and dividends and determinants of choice between the two payout methods. The analysis uses data on Polish firms with the sample period encompassing a significant tax regulation change. The average announcement period abnormal returns are relatively high and exceed 6% for stock repurchases and 2% for dividend initiations. The excess returns seem to be driven solely by the amount of cash distribution regardless of the method used. In particular, the tax issues do not prove to have an impact on the market reaction. Firms with higher operating, sustainable cash flows pay out dividends and use stock repurchases to signal information. JEL Classification: G14; G35 Keywords: Payout policy; Stock repurchases; Dividends I thank Bent Jesper Christensen and Waldemar Frackowiak for heplful comments and discussion. Part of this research was carried out when the author was visiting the School of Economics and Management at the University of Aarhus. This research has been supported by a Marie Curie Fellowship of the European Community Programme Improving the Human Research Potential and the Socio-Economic Knowledge Base under contract number HPMT- CT Adderss correspondence to Poznan University of Economics, Al. Niepodleglosci 10, Poznan, Poland. s.gryglewicz@ae.poznan.pl. 1

2 1 Introduction The following two empirical facts are well established in the corporate finance literature on payout policy. First, firms distribute cash to their shareholders using both alternative mechanisms: cash dividends and stock repurchases; despite that dividends are taxed more heavily in the U.S. and in many other markets, dividends persist in these markets. This empirical evidence has been referred as the dividend puzzle. 1 Second, open market stock repurchase and dividend initiation announcements are associated with positive abnormal stock return. The market reaction to both payout methods is of similar magnitude. 2 There are several hypotheses that interpret those observations: information signaling, reduction of principal-agent conflict costs, impact of market trading (in case of stock repurchases) and the tax disadvantage of dividends. The hypotheses do not make a consistent theory of payout policy, they are in some cases competing and contradictory; cash distribution to shareholders still remains a puzzling issue. The evidence and its theoretical rationalization are received mainly from U.S. data and U.S. companies. However, corporate payout policy is very sensitive to the legal and economical environment. For example, Grullon and Michaely (2002) argue that the growing importance of stock repurchases in the U.S. in the last 20 years can be explained by regulatory changes. This point makes the following question even more relevant: how can the U.S. results be generalized to other markets under different regulations? Answering this question will also help to predict and understand the potential impact on payout policy of tax or corporate law changes in the U.S. (e.g. the presidential tax cut proposal of 2003 including the abolition on taxation on dividend income). In this paper we analyze payout policy of publicly traded firmslistedonthewarsawstock Exchange (WSE) over the 1997 to 2000 period. The interesting feature of the Polish data is the change of the stock repurchase tax interpretation during the study period. Since February 1999 the tax authorities have started treating stock repurchases as regular cash dividends for tax purposes. In effect, the clear tax advantage of stock repurchases stopped to exist. Our objectives are twofold. (1) First, we analyze if the market recognizes any differences between stock repurchases and dividends and reacts more favorably to firms announcements of one of the payout instruments. The results show that the announcement period abnormal returns are 1 Recent studies of Fama and French (2000) and Grullon and Michaely (2002) show how the pattern of payout policy has been changing in the last decades. 2 Compare, e.g., Asquith and Mullins (1983) and Ikenberry, Lakonishok, and Vermaelen (1995). The former paper reports the initial reaction to dividend initiations of +3.7% and in the later paper the reaction to stock repurchases is +3.54%. 2

3 significantly higher in case of stock repurchases. Market reaction should be explained by the known hypotheses of payout policy. In principle, different characteristics of both cash distribution methods can make one of them a better signal of information or a more successful instrument in mitigating agency problems. We also directly investigate whether the tax change during the study period had any impact on market valuation of payout announcements. (2) Second, we test the hypothesis that different firms in different situations will prefer one of the two alternative means of payout. For example, undervalued firms would choose a payout instrument that more effectively signals this information to the market, or similarly firms with excess cash flow would use a method that is batter at mitigating agency conflicts. We utilize the event study procedure to reach the first of the abovementioned objectives. For the second objective, a logit-regression analysis estimates what factors determinate corporate payout policy decisions. The results prove that the market reacts positively to announcements of both payout instruments. The average abnormal returns in the announcement period are relatively high and exceed 6% for stock repurchases and 2% for dividend initiations. The results do not confirm any of the known hypotheses that could explain this pattern of the market reaction. The excess returns upon payout announcements seem to be driven solely by the amount of cash distribution regardless of the method used. In particular, the tax issues do not prove to have an impact on the market reaction. This is despite the fact that the specific tax environment (the differential taxation of individual and institutional investors and the regulatory change during the study period) ex ante seemed to provide a good background for testing the taxation hypothesis of payout policy. The paper shows a clearer picture on the side of corporate payout choice between stock repurchases and dividends. Dividends are paid out when firms have higher operating cash flows. This can be interpreted in line with the results from U.S. firms (Jagannathan, Stephens, and Weisbach (2000)), that operating and more sustainable cash flows are used for dividend payouts and stock repurchases are paid from more temporary cash flows. Firms also seem to use stock repurchases to signal information. As the evidence shows smaller, so less observed, and undervalued firms prefer stock repurchases. There are several interesting studies that tested for the various payout policy hypotheses in an international context. Rau and Vermaelen (2002) using U.K. data document the impact of regulatory provisions on stock repurchase activity. Several papers used Canadian repurchases to confirm or sometimes challenge the U.S. results (e.g. Ikenberry, Lakonishok, and Vermaelen (2000) and de Jonga, 3

4 Dijkb, and Veld (2003)). A number of studies of the Polish capital market data have been appearing in the international finance literature 3, however there are very few papers on corporate finance issues in Polish firms. In fact, we are aware of only one study connected with corporate finance by Lyn and Zychowicz (2003) on IPO performance. The remainder of the paper is organized as follows. Section 2 reviews related literature that motivates the empirical models analyzed in subsequent sections. Section 3 describes the sample. Sections 4 and 5 present empirical analyses of sources of announcement period abnormal returns and determinants of payout policy respectively. Conclusions are in Section 5. 2 Related literature Abnormal payout announcement returns Essentially, from the financial policy indifference theorem of Miller and Modigliani (1961), in perfect and complete markets there should be no positive market reaction to any payout announcements, and stock repurchases and dividends would be interchangeable. In fact, payout policy would not create any value and any cash distributions to shareholders would be obsolete. Two questions relevant to this paper arise in light of the Miller and Modigliani s (1961) payout policy indifference theorem: What factors create the positive market reaction to payout announcements? What explains the observed cross-sectional differences in the market reaction? In the corporate finance literature some of the assumptions of perfect and complete markets have been lifted to explain empirical evidence which is often at odds with the dividend policy indifference theorem. In signalling models of Bhattacharya (1979) and Miller and Rock (1985) cash distributions to shareholders convey positive information about firm s future earnings. However, in these settings dividends and stock repurchases are perfect substitutes. There is mixed empirical evidence of cash disbursements information signalling. Bartov (1991) find (weak) increase of earnings after open market stock repurchase announcements, but a more recent study of Grullon and Michaely (2003) does not confirm this result. 5 The evidence for dividends announcements varies in similar manner 3 Bohl and Henke (2003) study stock market volatility and returns. Charemza and Majerowska (2000) analyze how stock market regulations infuence portfolio risk. Scheicher (2001) studies the regional integration of stock markets including Poland. 4 For a comprehensive recent survey of the literature on payout policy refer to Allen and Michaely (2002). 5 The information signalling of tender offer and Dutch auction stock repurchases is documented more convincingly (see Vermaelen (1981), Comment and Jarrell (1991), and Lie and McConnell (1998)), but in this paper the focus is on the most popular form of repurchases, open market repuchsases. 4

5 but more often then not rejects earnings signalling of dividends (e.g. Benartzi, Michaely, and Thaler (1997) and DeAngelo, DeAngelo, and Skinner (1996)). Stock repurchases announcements can also signal stock price undervaluation. Ikenberry, Lakonishok, and Vermaelen (1995) show that stock prices experience a long-term upward drift after repurchase announcements. Firms seem to decide to repurchase when they are undervalued and the market discounts only part of this information at the announcement. If we assume that stock price is for some reason undervalued before a repurchase announcement and remains so afterwards then repurchasing can create positive value to shareholders analogous to that of Porter, Roenfeldt, and Sicherman (1999). They show that undervalued closedend funds gain in value after repurchasing their shares. It is even if the discount factor stays at the same level after repurchases. 6 It remains to be proved whether and to what degree the market reacts to the information signals of both cash distribution methods announcements, and it is an important research purpose of this paper. The second strand of literature that motivates payouts to shareholders and explains announcement period abnormal returns stems from the principal-agent conflicts and the free cash flow hypothesis (Jensen (1986), Easterbrook (1984)). Cash distributed in dividends and stock repurchases reduces the likelihood that managers will overinvest (invest in negative net present value projects). Usually dividends are permanent commitments to distribute cash to shareholders and stock repurchases are much more volatile and more discretionary method of cash payout (see Jagannathan, Stephens, and Weisbach (2000)). This could imply that dividends are a better instrument to monitor managers. But on the other hand stock repurchases are on average much larger disbursements and as such serve better to decrease excess funds available to managers. There is empirical support to the free cash flow hypothesis in case of dividends increases (Lang and Litzenberger (1989), Lie (2000)) and the evidence is even firmer for stock repurchases (Nohel and Tarhan (1998), Lie (2000), and Grullon and Michaely (2003)). The option to repurchase as described by Ikenberry and Vermaelen (1996) could be another source of the positive abnormal returns after a stock repurchase announcement. After announcing an open market repurchase program the timing and amount of purchased stocks is left to a large extent to managers discretion. If they act in favor of existing shareholders and repurchase when stocks are undervalued and refrain from repurchasing otherwise, then the option to repurchase will create a positive value. The more volatile is the stock price the higher is the potential value of this option. Market transactions associated with stock repurchase programs can have positive effects on prices. 6 Undervaluation of closed-end funds is clearly visible and measureable. 5

6 This aspect of repurchasing may also generate value of participating firms. Transactions within stock repurchase programs are often substantial even over 20% of outstanding shares are sometimes purchased (10% on average in this study). But if the price effect is to be beneficial to long-term shareholders it has to be lasting. Otherwise only selling shareholders would gain from price increase. In studies of the price impact of large transactions it is documented that the price increase after large purchases does not fully reverse (e.g. Chan and Lakonishok (1995)). It gives some warranty to the hypothesis that repurchase trades can boost prices and add value to shareholders. 2.2 Determinants of payout policy with dividends and stock repurchases There has been vast research explaining patterns of payout policy beginning at least from the classic study of Lintner (1956). However it was not until the 1980s, when stock repurchases became important substitutes of cash dividend disbursements in the U.S., that the finance literature started incorporating stock repurchase possibility into analyses. Ofer and Thakor (1987) present a theoretical model based on information signalling of payout policy where managers can signal undervaluation using either stock repurchases or dividends. In their model it is more cost effective to pay dividends when undervaluation is relatively small and to repurchase stocks to signal larger stock price misvaluation. This result hinges on an observation that the stock repurchases are more costly signals as they increase risk exposure of managers. In a similar information asymmetry setting Persons (1997) also shows that stock repurchases are better signals. In a model of Brennan and Thakor (1990) the information asymmetry is not between managers and shareholders but among different shareholders. As information is costly and it is not necessarily profitable for every shareholder to acquire information, stock repurchases lead to redistribution of wealth from uninformed to informed stockholders. The larger is the intended cash distribution the more shareholders will be willing to get informed. Therefore Brennan and Thakor argue that, if the decision about payout method is made by majority vote of shareholders, smaller distributions are made by dividends and larger by repurchases. Chowdhry and Nanda (1994) and Lucas and McDonald (1998) propose dynamic models of payout policy with better informed managers who maximize value of existing shareholders. In both papers authors interestingly analyze the tradeoff between advantages and costs of dividends and repurchases. The general results are similar. Managers are reluctant to repurchase when shares are above the true value despite lower taxes on repurchases. Hence, firms redistribute small amounts of cash with dividends and accumulate the rest to repurchase later when shares are undervalued. 6

7 Recent empirical studies of Guay and Harford (2000) and Jagannathan, Stephens, and Weisbach (2000) show interesting relations between the two methods of payout policy and cash-flows. In essence, they find that dividends are paid by firms with higher permanent operating cash flows, whilerepurchasesareusedbyfirms with higher temporary, non-operating cash flows. 3 Data The empirical analysis is focused on comparison of stock repurchases and dividends as two alternatives of payout policy. This restricts the beginning of the sample period to 1997 since that is when Polish exchange-listed firms first started repurchasing stocks. The data end with The sample includes announcements of initiations of stock repurchase programs, dividend initiations and substantial dividend increases. The latter we define as an increase of more than 25% from the previous year dividend. The exact announcement dates were not readily available. The data ware drawn from the database of the major Polish financial newspaper Parkiet. The database provided: the Polish Press Agency news, public announcements of listed firms and financial news from Parkiet. Of all these information we chose the earliest one for each payout announcement. Initially 43 stock repurchases made by 33 firms are identified in the study period. Non-open market repurchases, firms with missing date of announcement or size of payout and not actively traded firms are excluded. In the empirical analyses, additional three cases of firms announcing at the same time a stock repurchase and a dividend are not included. We repeated the same procedure for dividend initiations and dividend increases. In the final sample we study announcements of 33 repurchases, 54 dividend initiations and 33 dividend increases. Table 1 presents descriptive statistics of the sample. 7 Stock repurchases were the most popular in 1998 with 18 announcements. In year 2000 there were only 5 programs announced, however the value of the programs exceeded the value of 11 repurchases initiated in [Insert TABLE 1 around here] From Table 1 the relation between cash disbursements using alternative methods is not clear. However when payouts are scaled by size of a firm the differences are large and statistically significant. Table 2 shows these results. An average repurchase announcement amounts to almost 10% of 7 Table 1 provides statistics of 36 stock repuchase programs including those that coincided with dividends. This is to show a bigger picture of repurchase activities of Polish firms, that - as to my knowledge - have not been researched before. 7

8 outstanding stocks. For an average dividend initiation and dividend increase firms spend respectively 3.78% and 1.86% of their market value. [Insert TABLE 2 around here] The accounting data are drawn from the Notoria Service Database. A database of stock market data is obtained from the Warsaw Stock Exchange. 4 Empirical analysis of announcement period abnormal returns In this section, a multivariate regression is used to examine sources of market reaction to announcements of stock repurchases and dividend initiations. InSection4.1,wepresentandmotivatethe regression equation. Section 4.2 shows announcement period abnormal returns. Section 4.3 presents the results of regression analysis. Section 4.4 explores a bit further some tax issues with respect to market reaction to payout announcements. Throughout this section, the analysis focuses on market reaction to a new instrument. We assume, consistently with other studies (Guay and Harford (2000) and Jagannathan, Stephens, and Weisbach (2000)), that dividends are implicitly permanent commitments whereas stock repurchases are single year cash disbursements. If a repurchase program is repeated in a following year it is usually treated by managers and the market as a new independent decision. Therefore to understand how investors perceive dividends as a method of cash distribution in comparison with stock repurchases the analysis includes the repurchases and dividend initiations. 4.1 Model 1 The multivariate regression presented here aims at explaining the sources of market reaction upon announcements of dividend payouts and stock repurchases. The dependent variable is abnormal return around each announcement in the sample. We employ the standard event-study methodology of Brown and Warner (1985) to compute abnormal returns (see also the following section). A set of explanatory variables serve as proxy of hypothesized causal factors suggested in the theoretical and empirical literature (see Section 2.1). A dummy variable, SRD, takes the value one if a firm announces a stock repurchase program and zero if a firm initiates dividends. The variable indicates what part of the market reaction is due to the method of cash distribution after controlling this reaction for size of the disbursement and other explanatory variables. In other words, a positive (negative) coefficient of this variable will mean 8

9 that stock repurchase (dividend) announcements represent some added value to shareholders relative to the alternative payout method that is not described by other hypotheses (or more precisely variables). The regression includes a variable of relative payout to capture differences in the size of disbursements among individual announcements and also between the alternative instruments. The variable, denoted PAY, is measured as the value of announced disbursement scaled by the firm s market capitalization five days before the announcement. Market price to book value ratio (denoted MTB) can show the level of under- and overvaluation and in this manner can test the information signalling hypothesis. The ratio is also frequently used as a proxy of future growth opportunities. Firms with low market-to-book ratio have less investment opportunities, less positive net present value projects. And as managers have a tendency to overinvest, these firms suffer from higher agency problems. The predictions on market-to-book ratio are ambiguous. It can be interpreted in favor of different hypotheses. Hence the regression function also includes other variables connected with agency costs. The operational cash flows variable (denoted OCF) is calculated by dividing operational cash flows from the previous year before the announcement and asset value at the end of the previous year. Debt ratio (denoted DEBT) approximates marginal costs of external financing. Firms with more leverage face higher financial distress costs and have higher marginal costs of external financing. According to the agency conflict hypothesis the expected relation between the foregoing three variables and abnormal announcement returns is as follows: firms with more cash (higher OCF), lower future investment opportunities (lower BTM) and lower cost of additional external financing (lower DEBT) have higher abnormal returns around payout announcements. In February 1999 the Polish tax authorities announced their intent to start treating stock repurchases as regular cash dividends for tax purposes. As capital gains were and remained not taxed for individual investors, in effect, the marginal tax on repurchases was brought up to the level of dividend tax (see Section 4.4 for further discussion of tax issues). Dummy variable TAX (equal to 1 before and 0 after February 1999) captures the two states of the tax regulations during the study period. Stock prices with low trading volume could be more vulnerable to repurchase transactions (see Section 2.1 for discussion). If repurchase trading could have positive and lasting impact on stock prices then firms with a lower ratio of volume to stocks outstanding (denoted VLM) would benefit more from stock repurchases. 9

10 The regression equation also includes a proxy of the value of the option to repurchase (Ikenberry and Vermaelen (1996)), namely, the market price volatility of a stock calculated as the annualized standard deviation of daily returns (denoted VOL). The higher the volatility, the bigger are managers possibilities to create value by exercising this option. This value should be discounted by the market at the announcement. The final regression equation of Model 1a is thus given as AR it = α + x 0 itβ + ε it, (1) where the vector of explanatory variables is x it = [SDR it,pay it,mtb it,ocf it,debt it,srd it TAX it, SRD it VLM it,srd it VOL it ] 0. The dependent variable, AR it, represents abnormal return around payout announcement of firm i in year t. The regressors are defined as in the foregoing paragraphs. Table 3 summarizes the variables, the hypotheses they test, and expected signs of coefficients. [Insert TABLE 3 around here] 4.2 Announcement abnormal returns Most event studies use the market model to estimate normal performance of a given stock. However, the sample of this research is not typical. The main focus of this study is on stock repurchases, and the sample of repurchase announcements is already relatively thin. Firms that repurchase stocks are on average small and often newly exchange-listed. If the market model parameter estimation window is of the usual minimum of 120 trading days, it turns out that the estimation can be done for only 23 announcements, shrinking the sample by almost one-third. Campbell, Lo, and MacKinlay (1997) advocate that with limited data the market-adjusted-return model can be used to estimate abnormal returns. Correspondingly, Brown and Warner (1985) find that the estimation method does not have significant impact on the abnormal return estimates (in short-term event studies). In order to keep the sample size reasonably large and to increase statistical validity of the analysis, the presented results are obtained with the market-adjusted-return model and the sample of 33 10

11 repurchase announcements. 8 The market index used to estimate the normal rate of return is WIG, the most comprehensive index of the Warsaw Exchange. Consistently with the expectations, the market reacts positively to announcements of both dividend initiations and stock repurchases. The reaction to repurchase announcements is significantly larger. Table 4 presents cumulative average abnormal returns in 3-, 5-, 7-, 9- and 11-day windows centered on the announcement dates. [Insert TABLE 4 around here] Abnormal returns around stock repurchase announcements are significant at standard levels of confidence in all time windows. Already in the 3-day window the abnormal returns amount to 2.11%. As the analyzed time period widens, the abnormal returns increase and reach the most statistically significant level of 6.23% in the 7-day window. When compared to other studies, these excess returns are relatively high. Ikenberry, Lakonishok, and Vermaelen (1995) show 3.54% abnormal returns in the 5-day window around stock repurchase announcements. In Rau and Vermaelen (2002) study of British repurchases the excess market reaction is only 1.14% within the 11-day window. Dividend initiations have the most significant abnormal returns also in the 7-day window. The returns amount then to 2.25%. In all other time periods the returns are not statistically significant at any standard confidence levels. All the following analyses use the strongest abnormal returns of the 7-day announcement period. Figure 1 plots cumulative average abnormal returns for both samples stock repurchases and dividend initiations. The upper part of the figure represents a narrower time period, from 10 days prior to announcements to 10 after. It is evident that stock repurchases have higher abnormal returns upon announcement then do dividend initiations. In the lower part of the figure the plots start 50 trading days before each instrument initial announcement and continues 50 days after. There is an interesting pattern in the wider time window. Announcements of stock repurchases are preceded by a significant decrease in stock price. Cumulative abnormal return from the day 50 reaches the minimum of 8%, and five days before the announcement is below 6%. It is opposite in case of dividend initiations. On average they are announced after positive excess returns. Cumulative abnormal return is 2.26% on the fifth day before announcements. [Insert FIGURE 1 around here] 8 The abnormal returns estimated with the market model and the smaller sample not presented here are of the same magnitude. 11

12 4.3 Multivariate regression results Table 5 provides results of the empirical specification in Equation (1) estimated for the sample of stock repurchase and dividend initiation announcements (columns Model 1a). The regression model is dominated by one explanatory variable the only significant variable is the announced amount of cash disbursement (PAY). The coefficient of can be interpreted as follows: 1% of a market capitalization spent for distribution to shareholders leads to 0.391% positive market abnormal return. [Insert TABLE 5 around here] In order to examine if the dummy variable alone can show any differences in the market reaction to the two methods of distribution, in Model 1b all other variables ware dropped. The reasoning is that the dummy variable could potentially capture the relations which the remaining variables were to examine in details. The estimation presented in Table 5 (column Model 1b) shows that the market reacts solely to the level of payout regardless of the method of distribution. One way to interpret this result is to infer that a payout by a dividend initiation is undervalued by the market when compared with a similar stock repurchase distribution. This is because dividends are usually long-term commitments and repurchases are usually one time events. So an announcement of the same cash distribution with a dividend initiation implies a larger payout in total. If we follow that argument, the conclusion could be that stock repurchases have an additional value to shareholders. Unfortunately the known hypotheses or rather the variables proposed in the model do not show where this positive value comes from. 4.4 Taxation of stock repurchases and announcement abnormal returns In general, in the Polish tax system, capital gains are taxed more heavily for institutional investors (individual investors are exempted from the capital gains tax) and income taxes on dividends are higher for individual investors. In February 1999 the tax interpretation have changed and stock repurchases started to be subject to a withholding tax similar to that of dividends. This effectively raised marginal tax burden on repurchases to the level of dividend tax. The change of tax regulations during the sample period provides a unique opportunity to study the tax considerations of market reaction to payout announcements. In the regression Model 1a, the coefficient of the dummy explanatory variable indicating the state of the tax regulations (denoted TAX) is not statistically different from zero (see Section 4.3 and Table 5). Additional analyses show that there is no significant difference in abnormal returns around 12

13 stock repurchase announcements before and after the change of tax regulations in February Prior to February 1999 the average abnormal return upon 19 repurchase announcements amounts to 6.92% in the 7-day window. When stock repurchases became marginally taxed similarly to dividends after February 1999, the average market reaction to 14 repurchase announcements was 5.29%. The difference of 1.63% is statistically insignificant (p=0.617). Elton and Gruber (1970) first argued for the tax clientele effect as an explanation of payout policy. Investorswithhighmarginaldividendtaxeswouldpreferandinvestinfirms not paying dividends or paying out through stock repurchases instead. Some firms in the sample paid dividends when they announced stock repurchase programs. Taking the different tax rates of individual and institutional investors, the tax clientele effect could be observed in the announcement abnormal returns. Investors of firms paying out high dividends would less favorably react to stock repurchases. Without using any more advanced econometric methods, the basic prediction would be that the correlation between dividend yields and repurchase announcement abnormal returns is negative. Quite surprisingly, the correlation is positive (equals to 0.382) and statistically significant (p=0.028). The higher is firm s dividend yield the higher is positive abnormal return upon stock repurchase announcement. The result is similar prior to and after the tax regulation change. The conclusions presented in this section are at odds with the hypothesis that tax considerations play a role in how the market reacts to payout announcements. 5 Empirical analysis of determinants of payout policy The analysis of determinants of payout policy aims at measuring influence of different internal and external factors on the probability of choosing stock repurchase or dividend. Section 5.1 the variables of the logit regression are introduced and motivated. Section 5.2 presents and interprets the results. Throughout this section the analysis is done from a manager s point of view. A firm has to decide what method of cash disbursement to choose for a marginal distribution. The alternative studied in the previous section, i.e. stock repurchase or dividend initiation, is not a firm s decision problem in this context. A number of firms in the sample pay dividends when they announce repurchases. Thus, if a marginal payout choice was dividend, these firms would simply increase their dividends. To capture the real decision alternatives of marginal cash distribution, in this section the analysis includes dividend increases along with dividend initiations. 13

14 5.1 Model 2 Using a logit regression, denoted here Model 2, we estimate the common impact of a set of variables on the probability of choosing a dividend or a stock repurchase as a marginal cash disbursement method. The dummy dependent variable equals to one if a firm decides to repurchase and zero if a firm pays a dividend. The variable is interpreted as the probability of choosing stock repurchase. The explanatory variables characterize the financial and stock market situation of a firm at the time of announcement. The selected variables test for the existent hypotheses of determinants of payout policy as outlined in Section 2.2. Table 6 summarizes all regressors of Model 2. In theoretical models of Brennan and Thakor (1990) and Chowdhry and Nanda (1994) larger disbursements are made by means of repurchases and smaller by cash dividends. The variable PAY (defined as in Model 1) is thus included. Firms size is usually connected with the level of information asymmetry. Larger firms are more observed by investors and thus are less likely to be undervalued. Smaller firms will tend to choose a cash disbursement method that conveys more positive information to the market. In many asymmetric information models, such as Ofer and Thakor (1987), Persons (1997) and Lucas and McDonald (1998), a stock repurchase is a stronger signal than a dividend. The size variable (denoted CAP) is log of market capitalization on the fifth trading day before the announcement. Cumulative abnormal return in 50 days preceding the announcement (variable denoted CAR50) is a proxy of undervaluation. If stock repurchases are better signals then managers may be willing to use them instead of dividends after significant declines of prices. Correspondingly, Chowdhry and Nanda (1994) predict that firms wait with repurchases until prices are low and repurchases benefit long-term shareholders. It is ambiguous which cash disbursement method is more advantageous to mitigate agency costs. Stock repurchases are on average larger payouts so they take away more free cash flow from the managers in a given year. However dividends are usually long term commitments of cash distribution to shareholders. To test for free cash flow hypothesis the regression Model 2 includes the same proxies of agency costs as in the regression Model 1. However, in this case, we do not predict signs of estimated coefficients. The remaining variables (VLM and VOL) are calculated and interpreted as in Model 1. 14

15 The final regression function of Model 2a is thus given as SRD it = exp(α + x0 it β + ε it) 1+exp(α + x 0 it β + ε it), (2) where the vector of explanatory variables is x it =[PAY it,cap it,car50 it,mtb it,ocf it,debt it,vlm it,vol it ] 0. The dummy dependent variable, SRD it,equalsto1ifafirm i in year t announces a stock repurchase and 0 if it announces a dividend. The regressors are defined as in the foregoing paragraphs. [Insert TABLE 6 around here] 5.2 Logit regression results Table 7 (columns Model 2a) shows the estimated coefficients in the logit regression Equation (2). The dominance of one variable, the amount of announced payout (PAY), is not surprising as we could see before that there is a significant difference in the size of distribution in both methods of cash distribution (Table 2). However, the decision of the size of cash distribution can come after the decision of distribution type (as in, e.g., Chowdhry and Nanda (1994)). This relation could bias the estimation in Model 2a. Model 2b excludes the variable of payout amount from the regressors in Equation (2). [Insert TABLE 7 around here] The results of Model 2b suggest that information signalling is behind payout policy choice. The coefficient of the firms size is (significant at 5% level) and the coefficient on cumulative abnormal returns in 50 days prior to announcement of (significant at 10% level). This indicates that the probability to choose stock repurchase decreases with the size of a firm. In the model the size variable is a proxy of the level of information asymmetry. Smaller and less observed firms tend choose stock repurchases. Additionally, firms that experienced negative abnormal returns prior to announcement are more likely to repurchase stocks than to pay a dividend. It can be interpreted in line with the information signalling hypothesis, that undervalued firms signal with repurchases. Thereismoderateevidenceiffavoroffreecashflow considerations in payout policy. The coefficient of the operating cash flows variable is negative and weakly significant in both Model 2a and 15

16 2b. It means that the higher are firm s operating cash flows the higher is the probability of paying dividends. Nevertheless, there is no further confirmation of the agency cost hypothesis from other variables which were used to approximate free cash flow and agency problems. The negative coefficient of the operating cash flows variable is similar the results of Jagannathan, Stephens, and Weisbach (2000). It seems that dividends are financed from more permanent, operational cash flow, while stock repurchases are funded from non-operating, temporary cash flows. 6 Conclusions This paper studies to what extend stock repurchases and dividends are alternatives to each other. The analysis is conducted from two points of view. First, we investigate how the market perceives and reacts to announcements of both cash distribution methods. Second, we look into what determinates corporate payout policy when marginal disbursements could be done with either dividends or stock repurchases. The analysis uses a sample of payout announcements of firms listed on the Warsaw Stock Exchange over the period If stock repurchases and dividends are alternatives but not perfect substitutes, then various financial and stock market characteristics can make some firms more likely to choose one payout method. Our results can be interpreted in favor of the signalling and agency cost hypotheses. In our sample firms seem to use stock repurchases to signal information. Smaller, so less observed, and undervalued firms decide to repurchase stocks instead of paying dividends. On the other hand, dividends are paid out when firms have higher operating sustainable cash flows. The study finds large positive abnormal returns around announcements of dividends initiations (2.25%) and stock repurchases (6.23%). In the proposed empirical model, that regresses the observed abnormal returns on hypothesized factors, the only significant variableistheamountofpayout. The results do not confirm any of the known hypotheses that could explain this pattern of the market reaction. Taking into account the size of distribution and permanence of dividends payouts, we conclude the market reacts more favorably to stock repurchases. Additionally, tax considerations do not seem to have an impact on the observed excess returns. Quite on the contrary, the evidence is against the tax clientele effect in the market reaction to payout announcements. 16

17 References Allen, Franklin, and Roni Michaely, 2002, Payout policy, in Milton Harris George Constantinides, and Rene Stulz, ed.: Handbooks of Economics (North-Holland, Amsterdam). Asquith, Paul, and David W. Mullins, 1983, The impact of initiating dividend payments on shareholders wealth, Journal of Business 56, Bartov, Eli, 1991, Open-market stock repurchases as signals for earnings and risk changes, Journal of Financial Economics 14, Benartzi, Shlomo, Roni Michaely, and Richard Thaler, 1997, Do changes in dividends signal the future or the past?, Journal of Finance 52, Bhattacharya, Sudipto, 1979, Imperfect information, dividend policy, and the bird in the hand fallacy, Bell Journal of Economics 10, Bohl, Martin T., and Harald Henke, 2003, Trading volume and stock market volatility: The Polish case, International Review of Financial Analysis forthcoming. Brennan, Michael J., and Anjan V. Thakor, 1990, Shareholder preferences and dividend policy, Journal of Finance 45, Brown, Stephen J., and Jerold B. Warner, 1985, Using daily stock returns: The case of event studies, Journal of Financial Economics 14, Campbell, John Y., Andrew W. Lo, and A. Creig MacKinlay, 1997, The Econometrics of Financial Markets (Princeton University Press, Princeton). Chan, Louis K. C., and Josef Lakonishok, 1995, The behavior of stock prices around institutional trades, Journal of Finance 50, Charemza, Wojciech W., and Ewa Majerowska, 2000, Regulation of the Warsaw Stock Exchange: The portfolio allocation problem, Journal of Banking and Finance 24, Chowdhry, Bhagwan, and Vikram Nanda, 1994, Repurchase premia as a reason for dividends: A dynamic model of corporate payout policies, Review of Financial Studies 7, Comment, Robert, and Gregg A. Jarrell, 1991, The relative signaling power of dutch-auction and fixed-price self-tender offers and open-market share repurchases, Journal of Finance 46,

18 de Jonga, Abe, Ronald Van Dijkb, and Chris Veld, 2003, The dividend and share repurchase policies of Canadian firms: Empirical evidence based on an alternative research design, International Review of Financial Analysis 146, DeAngelo, Harry, Linda DeAngelo, and Douglas J. Skinner, 1996, Reversal of fortune, dividend signaling and the disappearance of sustained earning growth, Journal of Financial Economics 40, Easterbrook, Frank H., 1984, Two agency-cost explanations of dividends, American Economic Review 74, Elton, Edward, and Martin Gruber, 1970, Marginal stockholders tax rates and the clientele effect, Review of Economics and Statistics 52, Fama, Eugene F., and Kenneth R. French, 2000, Disappearing dividends: Changing firm characteristics or lower propensity to pay?, Journal of Financial Economics 60, Grullon, Gustavo, and Roni Michaely, 2002, Dividends, share repurchases and the substitution hypothesis, Journal of Finance 57, , 2003, The information content of share repurchase programs, Journal of Finance forthcoming. Guay, Wayne, and Jarrad Harford, 2000, The cash-flow permanence and information content of dividend increases versus repurchases, Journal of Financial Economics 57, Ikenberry, David, Josef Lakonishok, and Theo Vermaelen, 1995, Market underreaction to open market share repurchases, Journal of Financial Economics 39, , 2000, Share repurchases in Canada: Performance and strategic trading, Journal of Finance 55, Ikenberry, David, and Theo Vermaelen, 1996, The option to repurchase stocks, Financial Management 25, Jagannathan, Murali, Clifford P. Stephens, and Michael S. Weisbach, 2000, Financial flexibility and the choice between dividends and stock repurchases, Journal of Financial Economic 57,

19 Jensen, Michael C., 1986, The agency cost of free cash flow, corporate finance and takeovers, American Economic Review 76, Lang, Larry H. P., and Robert H. Litzenberger, 1989, Dividend announcements: Cash flow signaling vs. free cash flow hypothesis, Journal of Financial Economics 24, Lie, Erik, 2000, Excess funds and agency problems: An empirical study of incremental cash disbursements, Review of Financial Studies 13, , and John J. McConnell, 1998, Earnings signals in fixed-price and Dutch auction self-tender offers, Journal of Financial Economics 49, Lintner, John, 1956, Distribution of incomes of corporations among dividends, retained earnings, and taxes, American Economic Review 46, Lucas, Deborah J., and Robert L. McDonald, 1998, Shareholder heterogeneity, adverse selection, and payout policy, Journal of Financial and Quantitative Analysis 33, Lyn, Esmeralda O., and Edward J. Zychowicz, 2003, The performance of new equity offerings in Hungary and Poland, Global Finance Journal 14, Miller, Merton H., and Franco Modigliani, 1961, Dividend policy, growth, and the valuation of shares, Journal of Business 34, Miller, Merton H., and Kevin Rock, 1985, Dividend policy under asymmetric information, Journal of Finance 40, Nohel, Tom, and Vefa Tarhan, 1998, Share repurchases and firm performance:: New evidence on the agency costs of free cash flow, Journal of Financial Economics 49, Ofer, Aharon R., and Anjan V. Thakor, 1987, A theory of stock price responses to alternative corporate cash disbursement methods: Stock repurchases and dividends, Journal of Finance 42, Persons, John C., 1997, Heterogeneous shareholders and signaling with share repurchases, Journal of Corporate Finance 3, Porter, Gary E., Rodney L. Roenfeldt, and Neil W. Sicherman, 1999, The value of open market repurchases of closed-end fund shares, Journal of Business 72,

20 Rau, P. Raghavendra, and Theo Vermaelen, 2002, Regulation, taxes and share repurchases in the U.K., Journal of Business 75, Scheicher, Martin, 2001, The comovements of stock markets in Hungary, Poland and the Czech Republic, International Journal of Finance and Economics 6, Vermaelen, Theo, 1981, Common stock repurchases and market signaling: An empirical study, Journal of Financial Economics 9,

21 Table 1: The sample description Table 1 provides descriptive statistics of the sample. The sample consists of announcements of stock repurchases, dividend initiations and dividend increases of Polish firms over the period of To be included in the sample the observation must satisfy the following criteria: the firm must be listed at the Warsaw Stock Exchange; the stock repurchase must be an open market repurchase program; the firm discloses the amount to be spent over the duration of the share repurchase program; the date of initial announcement must be possible to identify from the Parkiet on-line database; the firm initiates dividends if it did not pay a dividend in the previous year; the firm increases dividend if the increase is of more than 25% from the previous year Stock repurchases sample Number of announcements Total payout (in milions of PLN) Mean payout (in milions of PLN) Median payout (in milions of PLN) Dividend initiations sample Number of announcements Total payout (in milions of PLN) Mean payout (in milions of PLN) Median payout (in milions of PLN) Dividend increases sample Number of announcements Total payout (in milions of PLN) Mean payout (in milions of PLN) Median payout (in milions of PLN) Table 2: Comparision of the announced amounts of payout Panel A Announced amounts of payout scaled by firm s market capitalization Mean Median Stock repurchases (N=33) 9.97% 8.45% Dividend initiations (N=54) 3.78% 2.59% Dividend increases (N=33) 1.86% 1.52% Panel B T-statistics of the mean difference test Stock repurchases and dividend initiations 5.266*** Stock repurchases and dividend increases 6.021*** Dividend initiations and dividend increases 2.961*** *Significant at the two-tailed 0.1 level ** Significant at the two-tailed 0.05 level *** Significant at the two-tailed 0.01 level 21

22 Table 3: Explanatory variables of Model 1 Table 3 summarizes the explanatory variables of the regression Model 1 (Equation (1)). The dependent variable is the abnormal return upon payout announcements. The payout method choice (SRD) is a dummy variable equal to 1 if the firm announces a stock repurchase and 0 if it announces a dividend initiation. The amount of payout (PAY) is measured as the value of announced disbursement scaled by the firm s market capitalization five days before the announcement. The market-to book ration (MTB) is from fifth day before the announcement. The operational cash flows variable (OCF) is measured by operational cash flows from the previous year before the announcement scaled by asset value at the end of the previous year. Debt ratio (DEBT) is debt divided by assets from the last quarter before the announcement. The tax regulation change (TAX) is a dummy variable equal to 1 before and 0 after February The trading volume variable (VLM) is measured by trading volume scaled by stocks outstanding in the quarter before the announcement. The volatility variable (VOL) is the annualized standard deviation of daily stock returns. The third column indicates which hypothesis the variable tests and what it is a proxy for. Explanatory variable Abbreviation Hypothesis Expected (variable proxy for) sign Payout method choice SRD Signaling/Agency +/ Cost/Taxation/Other Amount of payout PAY Signaling/Agency Cost + Market-to-book ratio MTB Agency cost (investment opportunities) Operational cash flow OCF Agency cost (financial + slack) Debt ratio DEBT Agency cost (marginal cost of external financing) Tax regulation change TAX Differences in taxation Trading volume VLM Impact of insiders transactions Stock price volatility VOL Option to repurchase (value of the option) Table 4: Cumulative abnormal returns around announcement day Table 4 reports the cumulative abnormal return surrounding payout announcements in different periods centered on the announcement day (announcement day = 0). Cumulative abnormal returns are calculated relative to the value-weighted index, WIG. The differences, reported in lower panel of the table, are between the cumulative abnormal returns surrounding announcements of stock repurchases and dividend initiations. Days 1 to +1 2 to +2 3 to +3 4 to +4 5 to +5 Stock repurchases 2.11%** 4.88%*** 6.23%*** 6.27%*** 4.67%** (t-statistic) (2.050) (3.586) (3.856) (3.359) (2.268) Dividend initiations 0.85% 1.78% 2.25%* 1.05% 0.46% (t-statistic) (0.971) (1.613) (1.688) (0.694) 0.275) Difference 1.25% 3.11%* 3.98%** 5.21%** 4.21%* (t-statistic for difference) (0.830) (1.735) (2.019) (2.234) 1.801) *Significant at the two-tailed 0.1 level ** Significant at the two-tailed 0.05 level *** Significant at the two-tailed 0.01 level 22