THE MARKET REACTION TO DIVIDEND INITIATION AND THE
ROLE OF DIVIDEND TAXATION IN THE NETHERLANDS
E.M.van Tuyll van Serooskerken1 1435310
Under supervision of dr. P.P.M. Smid
University of Groningen Faculty of Economics and Business
MSc Business Administration – Specialization Finance 7 September 2010
Abstract
In 2001 a Dutch tax reform eliminated a substantial tax bias against dividends which, up until that point, had been taxed more heavily than its alternative, capital gains. This study examines stock price reactions to dividend initiation announcements in the Netherlands in the context of this tax reform and finds that dividends initiation announcements result in a positive abnormal return of 2.64%. Subsequently, a conventional regression model based on previous studies was used to explain the dynamics behind the observed reaction and the role of dividend taxation. The results of this regression model were inconclusive.
JEL classification G14, G35
Keywords
Table of Contents
1. Introduction 3
2. Literature Review 5
2.1 Why Do Firms Pay Dividends? 5
2.2 Empirical Evidence 7
2.3 Dividend Taxation 10
3. Hypotheses and Methodology 11
3.1 Hypotheses 12
3.2 Estimating Price Reactions 13
3.3 Cross-sectional Regression 14
4. Data and Descriptive Statistics 17
5. Results 20
5.1 Price Reaction Results 20
1. Introduction
In the past couple of decades the subject of dividends has been hotly debated throughout the field of finance. In their extensive literature review on the topic of dividend policy, Allen and Michaely (2003) observe that the evidence found by studies is so mixed that they are unable to give any dividend policy recommendations. Primarily much of the discord can be traced back to the core question: since Miller and Modigliani (1961) showed that dividends are theoretically irrelevant, as in a perfect market investors can undo or replicate dividends, why do firms pay dividends at all? When adding imperfections such as transaction costs and particularly taxes into the equation the puzzle becomes even more complicated. The role of taxes, to which Miller and Modigliani refer as “undoubtedly the major systematic imperfection in the market” (p.432), is pivotal in the context of dividends. Since from an investors’ perspective the alternative to dividends are capital gains, taxes on dividends should only matter when they differ from capital gains taxes. Historically in most countries, such as the Netherlands (pre-2001) and the United States (pre-2003) dividends were taxed more heavily. As such one would not expect firms to have an incentive to pay dividends at all and those firms that do should be traded at a discount by the market. However markets have demonstrated the contrary, giving rise to what is commonly called ‘the dividend puzzle’: what are the benefits and costs of dividends in comparison to its alternative, capital gain?
observation has given rise to two research questions. Firstly, how does the market react to dividend initiation announcements in the Netherlands? Secondly, how has the Dutch tax reform affected these reactions? Investigating 74 instances of dividend initiation on the Dutch stock exchange from 1991 till 2009 this study will explore these questions.
The motivation for this study is two-fold. Firstly this study will contribute by examining dividend initiation announcements in the Netherlands, providing alternative evidence to a topic dominated by studies focused on the United States. Secondly I want to demonstrate how the impact of dividends on shareholder value is affected by non-neutral dividend taxation. Currently this is a contested issue, with no studies able to present conclusive evidence.
Following the studies done by Mitra and Owers (1995) and Jin (2000) the methodology used is two-staged. First, an event study is done to measure the impact of dividend initiation announcements on stock prices. Dividend initiations are defined as the first time a firm pays dividend or a resumption of payment after a 5-year hiatus. Initiations are preferred over other dividend announcements because they are unanticipated, thus avoiding the problem of investors’ expectations. The second stage involves a cross-sectional multivariate regression to explain the relationship between the observed announcement reaction, the Dutch tax reform and several control variables.
2. Literature Review
2.1 Why Do Firms Pay Dividends?
Paying dividends brings both costs and benefits to shareholders’ value. Costs associated with paying dividends are straightforward. Paying dividends is costly, because it raises a firm’s expected financing costs due to the higher probability of external financing, which is costlier than using internal funds (Black 1976). Additionally paying dividends reduces financial slack, which causes managers to forego profitable investments (Myers and Maljuf, 1984). Furthermore paying dividends involves transaction costs involved. Moreover taxing dividends higher than capital gains, as is done in many countries, also implies an additional economic cost to the dividend-receiving shareholder. Using share repurchases instead of dividends would circumvent the problem of non-neutral taxation. However, despite this obvious advantage, firms have continued to distribute cash through dividends. As such the main body of research has sought to explain the witnessed dividend policy.
stock is undervalued or overvalued. As such Brennan and Thakor’s model is considered the signaling model that fits the dividend payout policies witnessed best.
An alternative strand of literature sees a role for dividends as a mean to mitigate agency problems rather than informational asymmetry. According to the free cash flow hypothesis as put forward by Jensen (1986), managers have the tendency to spend excess cash on low-return investment projects and unnecessary perks. Lang and Litzenberger (1989) argue that paying dividends is valuable to investors because it lowers the amount of funds that managers could waste. Rozeff (1982) suggests that restricting managers’ available cash forces them to seek new funds suppliers that will provide free monitoring of the firm to existing shareholders.
2.2 Empirical Evidence
How the dividend-paying decision affects shareholder value is ultimately the sum of the perceived costs and benefits. The standard methodological framework to measure this net value of dividends was established by Asquith and Mullins (1983). They examined stock returns of 168 US firms in a 20-day window surrounding dividend initiation announcements. Asquith and Mullins’ model thoroughly deals with deficiencies that plagued earlier studies. Firstly, they control for simultaneous sources of information such as earnings announcements. Secondly, and most importantly they isolate and control for investors’ expectations. Expectations are problematic because when investors expect a dividend, its initiation effect could already be included in the stock price. This could lead to distorted results. Investors could, for example, have anticipated a higher dividend than announced, which would then cause a smaller or even negative price reaction. Asquith and Mullins deal with this issue by only taking into account firms that paid dividends for the first time or after a 10-year hiatus and thereby capturing the full extent of the dividend announcement surprise. Their result is both substantial and significant: Asquith and Mullins report that dividend initiations bring investors an average cumulative abnormal return (ACAR) of 3.7% in a 2-day window around the initiation announcement. A 2-day window is used because initiation news usually arrives at the market the day before the newspapers report it.
Table 1 : Average Cumulative Abnormal Returns around Dividend Announcements
Study Data surveyed ACAR at Dividend Event*
Initiation Increase Omission Decrease
Asquith and Mullins (1983) † US (70-79) 3.70% Richardson et al. (1986) † US (69-82) 4.00%
Healy and Palepu (1988) † US (69-79) 3.90% -9.10%
Bajaj and Vijh (1990) US (62-87) 1.04% -0.53% Mitra and Owers (1995) US (76-87) 2.19%
Michaely et al. (1995) US (64-68) 3.40% -7.00%
Yoon et Starks (1995) ††† US (69-88) 1.15% -5.16% Lonie et al. (1996) UK (91) 2.03% -2.15% Amihud and Murgia (1997) Germany (88-92) 0.96% -1.73% Howe and Shen (1998) † US (68-92) 3.32%
Alangar et al. (1999) †† US (76-90) 2.58% 1.03% -5.86% -4.20% Jin (2000) † US (73-93) 2.98%
Gurgul et al. (2003) Austria (92-02) 1.50% -1.78 Capstaff et al. (2004) Norway (93-98) 1.00% -0.59% Dasilas et al. (2009)†† Greece (00-04) 2.14%
* All values reported have a minimal significance α < 0.05
†
Initiations include resumptions of dividends after a minimum 10 year hiatus
†† Initiations include resumptions of dividends after a minimum 3 year hiatus ††† 3-day (instead of 2-day) event window
Taking into account the aforementioned evidence for signaling theory, research has been trying to find out exactly what kind of information dividends actually contain. Healy and Palepu (1988) find evidence that dividends announcements convey information about subsequent future changes in earnings. Venkatesh (1989) finds that the informational value of quarterly earnings announcements diminishes when a firm pays cash dividends. Lonie et al. (1996) demonstrate that price reactions on dividend announcements interact heavily with the sign and value of a firm’s earnings announcement. Benartzi, Michaely and Thaler (1997) find that dividends signal that recent earnings changes are sustainable but contain no information about future growth. Dyl and Weigand (1998) find that dividend initiations signal that a firm’s earnings and cash flows have become fundamentally less risky. Howe and Shen (1998) show that dividend initiations are firm-specific events, that contain no information that affects the share price of intra-industry competitors.
Support for the alternative theories, such as free cash flow and behavioral finance is scarce. Using Tobin’s Q as an indicator of a firm’s investment opportunities, Yoon and Starks (1995) find no evidence that firms with low investment opportunities see higher price reactions at dividend announcements, which is what free cash flow theory would predict. Dong, Robinson and Veld (2005) submitted a questionnaire on dividends to a Dutch investor panel and evaluated which theories were consistent with investors’ responses. Interestingly, evidence was found for signaling theory, but little support for its alternatives. This lack of support is certainly puzzling, especially since the agency theory on the role of dividends appears solid and rather intuitive. Allen and Michaely’s (2003, pp. 828) opinion on the state of dividend policy research is as follows:
“(There are)…two opposing views of why dividends are paid. The first view is that
2.3 Dividend Taxation
As mentioned in the introduction, the whole debate on why firms pay dividends is primarily fueled by the observation that the presence of non-neutral taxation makes it counterintuitive to do so. Firms continue to distribute part of their profits as dividends while many of the receiving investors are in high tax brackets. Rather than solely looking at ‘hidden’ benefits such as those implied by signaling and agency theory, some authors have been investigating the dynamics of dividend taxation instead. Pivotal to this body of literature is the so-called ‘tax effect’: how does dividend taxation affect shareholder value in the context of the dividend paying decision? Naturally this question is only relevant in the cases where dividends are effectively taxed higher than capital gains, which was the case in the Netherlands (pre-2001) the United States (pre-2003) and numerous other countries.
Most dividend models assume non-neutral taxation has a negative impact on the dividend-receiving shareholder’s wealth. In other words, the shareholder had been better off if the dividend had been retained or used to repurchase shares. However, several signaling models such as John and Williams (1985) and Bernheim (1991) imply the contrary. In Bernheim’s model, firms use dividends to signal quality. He shows that with a high dividend tax rate, high-quality firms do not need to pay as large a dividend to deter imitation from low-quality firms. Thus, the firm manages to convey the same information at a lower level of payout, making paying dividends attractive even in the presence of non-neutral taxation. In support of this view Bernheim and Wantz (1995) find evidence that, in periods of higher relative taxes, the abnormal returns witnessed at dividend announcements are greater.
value completely. In my view signaling models such as Bernheim’s indeed have several shortcomings. First they assume that low-quality firms can succeed in mimicking high-quality firms by paying dividends. However Jin (2000) demonstrates that there is a population of firms for which the dividend initiation provokes a negative price response based on specific economic characteristics, which implies investors are hard to fool and can distinguish a good dividend initiating firm from a bad one. Secondly Bernheim’s model is counterintuitive to the observation made by Litner (1956) and Deangelo and Deangelo (1990) that firms are extremely reluctant to reduce dividends once they are committed to them and thus only start paying dividends when they believe they are safely sustainable. Which by definition however, for low-quality firms they are not.
3. Hypotheses and Methodology
3.1 Hypotheses
The question addressed by this paper is how the market’s reaction to the dividend-paying decision is impacted by dividend taxation. In September 1999 a draft for a new Income Tax Act was presented to the Dutch parliament to replace the old Income Tax Act which had been in place since 1964. Under the old income tax act taxpayers would pay a progressive (a sliding scale up to 60%) rate on dividend income and no taxes were paid on capital gains2. Under the new Income Tax Act, which entered into force 1 January 2001, all financial assets are assumed to have a 4% yield over which a flat rate of 30% is levied. This means an effective 1.2% tax on the actual value of assets held and no separate taxes on dividends or capital gains. The Dutch reform is superior to simply evening out the tax rates on both forms of income, because even at equal rates, taxpayers could still defer taxes on capital gains by realizing capital losses. The fact that the reform has completely eliminated this tax bias makes the Netherlands a perfect case study. To accurately measure the market’s reaction, dividend initiations are used. This avoids the problem of investor expectations already included in the stock price when the dividend is announced. Because of the theoretical ambiguity, which was discussed in the literature review, the main hypotheses are specified as follows:
Hypothesis 1:
There is no stock price reaction to dividend initiation announcements in the Netherlands
Hypothesis 2:
The change in tax regime in the Netherlands has had no effect on the price reaction to dividend initiation announcements
To test the hypotheses this study will use a two-stage methodology. Firstly through an event study the price reactions to dividend initiation announcements will be measured.
The second stage consists of the use of a cross-section multivariate regression model to test Hypothesis 2.
3.2 Estimating Price Reactions
To estimate the stock price reaction to the dividend initiation announcement this study will follow a standard event study methodology, see e.g. Mackinlay (1997). Let the day the dividend initiation is announced in the newspaper be defined as τ = 0. Using the market model, stock market abnormal returns of dividend initiating firms are computed in a 40-day window around the announcement day. The abnormal return (ARkτ ) of the security of firm k on day τ is defined as follows:
) ( τ τ τ k αk βk k k R R AR = − + (1)
Where Rkτ is the return of the security of firm k on day τ. α and βk are the intercept
and slope coefficients for the market model that is run over an estimation period of τ = -90 to τ = -31 to predict the unbiased daily return of the security of firm k relative to the dividend announcement event.
To aggregate across events, average abnormal returns (AARτ) are calculated for each relative day. Given N events, the sample average abnormal return for day τ is
∑
= = N k k AR N AAR 1 1 τ τ (2)To aggregate across time the cumulative abnormal return (CARk(τ1,τ2)) of firm k is computed for any interval τ1 to τ2 in the event window:
Initiation announcements made during the day usually appear in newspapers the next morning. When the dividend is announced before the market closes, which happens in most cases, the market reaction actually predates the announcement day by one. Studies such as Asquith & Mullins (1983), Mitra and Owers (1995) and Jin (2000) confirm this and find that the full price reaction to the news of the dividend initiation of firm k,
) 0 , 1 (− k
CAR is usually captured in a two-day window from τ1 =−1 to τ2 =0.
To compute the average price reaction in time interval τ1 toτ2, the average cumulative abnormal return ACAR(τ1,τ2) is computed as follows:
∑
= = 2 1 ) , ( 1 2 τ τ τ ττ
τ
AAR ACAR (4) or alternatively ) , ( 1 ) , ( 1 2 1 2 1 1τ
τ
τ
τ
∑
= = N k k CAR N ACAR (5)Likewise, it is assumed thatACAR(−1,0)fully represents the average price reaction to dividend initiation announcements across the observed events. Its resulting value will be used to test Hypothesis 1.
3.3 Cross-sectional Regression
The draft version of the reform was presented to the Dutch parliament in September 1999. However its content was adjusted five times before it passed legislation. Finally in May 2000 it was announced officially. To simplify, 1 January 2001 will be considered the cut-off date for the main model. In the regression results section, the use of alternative cut-cut-off dates will be addressed.
Additionally there has been controlled for the following effects:
1. Firm Size (SIZE): As shown by Bajaj and Vijh (1990), Mitra and Owers (1995) and Jin (2000) small firms that initiate dividends see higher price reactions at announcement than large firms. This is because small firms have less publically available information and thus suffer more from informational asymmetry. As such the information content of the dividend initiation is more valuable. Firm size is measured as the natural log of the total value of a firm’s equity. The sign of the coefficient is predicted to be negative.
2. Institutional Ownership (INST): Institutional ownership, defined as the percentage of equity held by institutions, is another proxy for informational asymmetry. Institutes monitor firms more closely because their monitoring activities benefit from economies of scale, thus reducing the informational value of dividends. Mitra and Owers (1995) and Alangar, Bathala and Rao (1999) and Jin (2000) all document that the percentage of equity held by institutions is inversely related to the announcement price reaction.
4. Market Anticipation (ANT): To control for the situation where the dividend initiation was anticipated by investors in the days leading up to the announcement, the cumulative abnormal return CARk(−20,−2)from day -20 through day -2 is computed. As can be expected the relationship between CARk(−20,−2) and
) 0 , 1 (−
CAR is inverse. If the market anticipated good news the price reaction at announcement will be smaller. This relationship was established by Jin (2000).
5. Earnings Announcements (EARN): Usually the news of the dividend initiation is brought together with biannual or quarterly earnings announcements. Asquith and Mullins (1983) showed that on average, simultaneous earnings announcement news reduces the price reaction of initiation announcements. Therefore we introduce a dummy variable to control for this effect, where the value of 1 means that there is a simultaneous earnings announcement.
Nowadays most financial regression models control for sector effects in some fashion. Strangely, so far none of the related studies done on dividend announcements has done so. Consequently the cross-sectional regression model (6) is defined as follows:
k k k k k k k
k LN SIZE INST YIELD DTAX ANT EARN
CAR (−1,0)=α+β1 ( )+β2 +β3 +β4 +β5 +β6 +ε
4. Data and Descriptive Statistics
A database of all the dividend initiations between 1991-2009 by firms listed on the Amsterdam Stock Exchange was constructed using Datastream and Bloomberg Professional. The year 1991 was chosen as starting point as some data sources did not go back further. To be included in the sample, the dividend initiation had to meet the following criteria:
(i) The firm is paying dividend for the first time or is resuming payment after a minimum 5-year hiatus. After 5 years the dividend is assumed to be unanticipated.
(ii) Its announcement was published in a Dutch daily newspaper.
(iii) There is no arrival of simultaneous other news in the event window (other than earnings announcements, which are controlled for).
(iv) All other data requirements are met.
Table 2: Descriptive Statistics
Panel A: Break up of Dividend Initiation Announcements by year
Year Number of Firms Year Number of Firms
1991 1 2000 14 1992 4 2001 0 1993 5 2002 2 1994 4 2003 1 1995 5 2004 2 1996 2 2005 2 1997 2 2006 8 1998 3 2007 6 1999 8 2008 4 2009 1 Total 74
Panel B: Break up of Dividend Initiation Announcements by sector
Sector Number of Firms Percentage (%)
Finance 6 8.11 IT/Communication 14 18.92 Manufacturing 27 36.49 Media/Publishing 2 2.70 Property 5 6.76 Shipping/Transport 3 4.05 Other 17 22.97 Total 74
Panel C: Descriptive Statistics of Independent Variables
Mean St. Dev Minimum Maximum
SIZE†‡ 1171.7 3095.5 3.93 21824.6
INST‡ 0.129 0.182 0 0.79
YIELD‡ 0.037 0.055 0.002 0.398 ANT -0.011 0.132 -0.593 0.320
Panel D: Descriptive Statistics of Dummy Variables
Dummies 0 1
N N
DTAX 48 (65%) 26 (35%) EARN 9 (12%) 65 (88%)
† Market capitalization in millions of Euros ‡ Measured at τ = -2
Table 2, Panel A shows the break up of dividend initiation announcements by year. As can be seen from the table, there was a peak in dividend initiations at the turn of the century. At a glance it seems the amount of dividend initiations follows the economic cycle, with a slump in 2001 as a result of the economic downturn then. Panel B shows the break up of dividend initiation announcements by sector.
shows, the distribution of SIZE is very skewed which is why the variable is log liniarized in the regression model. INST is the percentage of equity held by institutions measured two days before the initiation announcement. This information was retrieved from the Public Database Notifications Substantial Holdings, which is published by the Netherlands Authority for the Financial Markets (AFM). Since 1991, the AFM requires all listed companies to disclose shareholders that hold over 5% of issued capital. Unfortunately this means institutional holdings below 5% could not be counted. On average the firms in the sample were 12.9% institutionally owned. This number is comparable to Jin’s (2000) US sample, where on average dividend initiation firms were 11.5% institutionally owned. YIELD, the size of the initial dividend announced divided by the share price two days before the announcement, on average is 3.7%, which is fairly low compared to the yields reported in other studies. ANT is CARk(−20,−2) and measures investor anticipation in the trading days leading up to the dividend announcement. On average, anticipation is pretty volatile with a standard deviation of 13.2%.
5. Results
5.1 Price Reaction Results
Table 3 reports the average abnormal returns and cumulative abnormal returns of the securities of dividend initiating firms in a 40-trading day window around the announcement date. Abnormal returns were calculated using closing prices.
Table 3: Average Abnormal Returns (AAR) and Average Cumulative Abnormal Returns (ACAR) in a 40-day window around the dividend initiation announcement day
τ AAR(%) t-value ACAR(-20,τ)(%) τ AAR(%) t-value ACAR(-20, τ)(%)
-20 0.52 1.16 0.52 1 - 0.07 - 0.17 1.44 -19 - 0.13 - 0.45 0.39 2 - 0.42 - 1.14 1.02 -18 - 0.18 - 0.69 0.21 3 0.23 0.47 1.25 -17 - 0.07 - 0.20 0.14 4 0.27 0.54 1.52 -16 0.05 0.17 0.19 5 0.01 0.03 1.52 -15 - 0.33 - 0.98 - 0.14 6 - 0.06 - 0.20 1.46 -14 0.37 0.93 0.24 7 0.23 0.97 1.69 -13 - 0.17 - 0.76 0.06 8 0.16 0.57 1.86 -12 - 0.39 - 1.61 * - 0.33 9 - 0.21 - 0.83 1.65 -11 0.31 1.13 - 0.01 10 - 0.57 - 2.75 *** 1.08 -10 - 0.42 - 1.52 * - 0.44 11 0.32 0.97 1.40 -9 - 0.11 - 0.47 - 0.55 12 - 0.29 - 1.06 1.11 -8 - 0.33 - 1.21 - 0.88 13 0.10 0.39 1.21 -7 0.07 - 0.18 - 0.94 14 0.12 0.48 1.33 -6 - 0.02 - 0.09 - 0.96 15 0.20 0.74 1.53 -5 - 0.57 - 1.84 ** - 1.53 16 0.37 1.26 1.90 -4 0.11 0.53 - 1.42 17 - 0.20 - 0.90 1.70 -3 0.38 1.33 * - 1.03 18 0.01 0.02 1.71 -2 - 0.10 - 0.37 - 1.13 19 - 0.17 - 0.58 1.53 -1 1.75 3.79 *** 0.62 20 - 0.27 - 0.81 1.27 0 0.89 1.93 ** 1.51
* Statistically significant at the 0.1 level ** Statistically significant at the 0.05 level *** Statistically significant at the 0.01 level
return of 0.89%, with a t-value of 1.93 and significance at the 5% level of confidence. See Appendix 1 for a graph plotting the results from Table 3.
The average cumulative abnormal return for the two-day announcement period,ACAR(−1,0), is 2.64%. This result has a t-value of 4.46 and is significant at the 0.05% level of confidence. This implies that there is a significant and large price reaction to the news of the dividend initiation. Hypothesis 1 is thus rejected. This result is in line with the results of similar studies such as those reported in Table 1.
5.2 Regression Results
Table 4 reports the results of the cross-sectional regression (6) of the dependent variable ) 0 , 1 (− k
CAR , which is the two-day price reaction to the dividend initiation announcement.
Table 4: Results of the cross-sectional regression of CARk(−1,0), full model.
Independent Variable Coefficient t-value LN(SIZE) -0.0067 -0.79 INST -0.0088 -0.25 YIELD 0.0612 0.52 ANT -0.0334 -0.70 DTAX 0.0179 1.21 EARN -0.0121 -0.64 Intercept 0.0589* 2.08 Adj. R² 0.07 F-statistic 0.84 * Statistically significant at the 0.1 level
rejected. However, the insignificance of established control variables suggest that there may be a deeper conflict with the specific nature of the dataset and the theoretical assumptions of the model. The robustness checks, which can be found in Appendix 2, indicate that multicollinearity is not a large problem but heteroscedasticity is. This heteroscedasticity could be attributable to the relatively small amount of observations. As such one of the assumptions of the ordinary least squares method is violated and regression results are is inconclusive. To further explore, an alternative partial regression model is run in which the control variables that proxy for informational asymmetry, SIZE and INST, are left out. Table 5 contains the results of the partial regression model.
Table 5: Results of the cross-sectional regression of CARk(−1,0), partial model.
Independent Variable Coefficient t-value YIELD 0.0776 0.70 ANT -0.0258 -0.56 DTAX* 0.0226 1.71 EARN -0.0116 -0.63 Intercept** 0.0427 2.18 Adj. R² 0.06 F-statistic 1.12 * Statistically significant at the 0.1 level ** Statistically significant at the 0.05 level
As can be seen from Table 5, the biggest difference between the full and partial regression models is that the result of the latter the variable DTAX is found significant at the 10% confidence level. Unfortunately the partial model suffers from similar heteroscedasticity problems as the full model. Subsequently it is impossible to draw conclusions from this result.
Summarizing, the empirical results show that there is a positive price reaction to dividend announcements in the Netherlands, but the regression model used is unable to explain why. Elaborating on this observation, I present several possible explanations.
Firstly one could argue that these results reflect the general state of research on dividends. Bernhardt et al. (2005) argued that the fundamental questions: ‘why do firms pay dividends?’ and ‘why does the market respond favorably to dividend increases?’ remain unanswered. Indeed the power of Mitra and Owers’s (1995) and Jin’s (2000) similar regression models is stronger albeit still marginally. Linking the implications of dividend theories such as signaling and agency theory to empirical evidence thus remains problematic. There is a consensus that the market responds favorably to the dividend initiation news but the dynamics behind that response are still uncertain. How dividend initiation price reactions differ across industry sectors and how they are affected by the economic cycle has not been firmly established yet.
Secondly with respect to dividend taxation one could wonder whether the use of cum-ex-day trading strategies has somewhat absorbed the effects of non-neutral dividend taxation? Investors using these strategies would dampen the ‘tax effect’ that this study was trying to witness.
6. Conclusion
This study investigates the price reaction to dividend initiation announcements in the Netherlands. Specifically the question how the Dutch tax reform of 2001 affected these price reactions is addressed. Previously, dividends were taxed more heavily than its alternative, capital gain. The tax reform eliminated this bias and thus poses a good opportunity to witness the role of dividend taxation. In total 74 instances of dividend initiations on the Dutch stock market in the period 1991-2009 are identified. Subsequently a two-stage methodology is used to study the price reaction and the ‘tax effect’. Firstly the price reaction in a two-day window around the announcement is measured using an event study. Secondly a cross-sectional regression is conducted to measure the dynamics behind the announcement price reaction and the effect of the tax reform.
The market reaction to dividend initiation announcements in the Netherlands is both positive and significant. On average the 2-day cumulative abnormal return found is 2.64%. This result is in line with comparable studies done in other countries. The regression result is inconclusive. The coefficients of the independent variables have the sign that was predicted but they are statistically insignificant. Overall, the results of this study show that the market responds to dividends favorably but fail to demonstrate why. As such, this study confirms some doubts previous authors have had about the power of existing theory to explain why exactly the market responds to the dividend initiating news so favorably. As it remains, the precise role of non-neutral dividend taxation is still unclear.
researchers’ ability to separate the effect of the dividend initiation news from the earnings news.
Future research should be in two directions. First a similar study could be conducted in the United States, where a dividend taxation reform was introduced in 2003. Such research would circumvent some of the data limitations of this study and provide more conclusive evidence. Up until this point such as study has not yet been done. Alternatively research should more thoroughly investigate the dynamics behind price reactions to dividend initiation announcements. Such a study could provide the foundations for more robust regression models that have better explanatory power. For example, the existence of cyclical or sector effects could be examined.
7. References
Alangar, S., Bathala, C. and Rao, R. (1999), “The effect of institutional interest on the information content of dividend-change announcements”, The Journal of Financial Research, Vol. 22, pp. 429-448.
Allen, F., and Michaely, R. (2003), “Payout policy”, in Constantinides, G., Harris, M. and Stulz,R., eds.,
Handbooks of Economics, pp. 793-837.
Amihud, Y and Murgia, M. (1997), “Dividends, taxes, and signaling: evidence from Germany”, Journal of
Finance, Vol. 52, pp. 397– 408.
Asquith, P. and Mullins, D. (1983), “The impact of initiating dividend payments on shareholders’ wealth”,
Journal of Business, Vol. 56, pp. 77-96.
Bajaj, M. and Vijh, A. (1990), “Dividend clienteles and the information content of dividend changes”,
Journal of Financial Economics, Vol. 26, pp. 193-219.
Benartzis, S., Michaely, R. and Thaler, R. (1997), “Do changes in dividends signal the future or the past?”, Journal of Finance, Vol. 52 No. 3, pp. 1007-1034.
Bernhardt, D., Douglas, A. and Robertson, F. (2005), “Testing dividend signaling models”, Journal of
Empirical Finance, Vol. 12, pp. 77-98.
Bernheim, B. D. (1991), “Tax policy and the dividend puzzle”, Rand Journal of Economics, vol. 22, pp. 455-476.
Bernheim, B.D. and Wantz, A. (1995), “A tax-based test of dividend signaling hypothesis”,
American Economic Review, Vol. 85 No. 3, pp. 532-551.
Bhattacharya, S. (1979), “Imperfect information, dividend policy and ‘the bird in the hand’ fallacy”, Bell Journal of Economics, Vol. 10, pp. 259-270.
Black, F. (1976), “The Dividend Puzzle”, Journal of Portfolio Management, vol. 2, pp. 5-8.
Capstaff, J., Klæboe, A. and Marshall, A. (2004), “Share Price Reaction to Dividend Announcements: Evidence on the Signaling Model from the Oslo Stock Exchange”, Multinational Finance Journal, Vol. 24, pp. 115-139.
Dasilas, A., Lyroudi, K. and Ginoglou, D. (2009), “The impact of dividend initiations on Greek listed firms’ wealth and volatility across information environments”, Managerial Finance, Vol. 35, pp. 531-543.
Deangelo, H. and Deangelo, L. (1990), “Dividend Policy and Financial Distress: An Empirical Investigation of Troubled NYSE Firms”, Journal of Finance, Vol. 45, pp. 1415-1431.
Dong, M., Robinson, C. and Veld, C. (2005), “Why individual investors want dividends”, Journal of
Corporate Finance, Vol. 12, pp. 121-158.
Dyl, E. and Weigand, R. (1998), “The Information Content of Dividend Initiations: Additional Evidence”,
Financial Management, Vol. 27, pp. 27-35.
Eades, K., Hess, P. and Kim, H. (1994), “Time-series variation in dividend pricing”, Journal of
Finance, Vol. 49, pp. 1617-1638.
Gurgul, H., Mestel, R. and Schleicher, C. (2003), “Stock market reactions to dividend announcements: empirical evidence from the Austrian stock market”, Financial Markets and Portfolio Management, Vol. 17, pp. 332-350.
Healy, P. and Palepu, K. (1988), “Earnings information conveyed by dividend initiations and omissions”, Journal of Financial Economics, Vol. 21, pp. 149-175.
Howe, S. and Shen, Y. (1998), “Information associated with dividend initiations: firm specific or industry-wide?”, Financial Management, Vol. 27, pp. 17-26.
Jensen, M. (1986), "Agency Costs of Free Cash Flow, Corporate Finance, and the Market for Takeovers",
American Economic Review, Vol, 76, pp. 323-329.
Jin, Z. (2000), “On the differential market reaction to dividend initiations”, The Quarterly Review of
Economics and Finance, Vol. 44, pp. 263-277.
John, K. and Williams, J. (1985), “Dividends, dilution and taxes: a signaling equilibrium”, Journal
Lang, L., and Litzenberger, R. (1989), "Dividend Announcements: Cash Flow Signaling vs. Free Cash Flow Hypothesis?," Journal of Financial Economics, Vol. 24, pp. 181-191.
Michaely, R., Thaler, R. and Womack, K. (1995), “Price reactions to dividend initiations and omissions: overreaction or drift?”, Journal of Finance, Vol. 50, pp. 573-608.
Michaely, R. and Villa, J., (1995), “Trading volume with private valuations: Evidence from the ex-dividend day”, The Review of Financial Studies, Vol. 9, pp. 471-509.
Lee,Y., Liu, Y., Roll, R. and Subrahmanyam, A. (2006), “Taxes and dividend clientele: Evidence from trading and ownership structure”, Journal of Banking & Finance, Vol. 30, pp. 229-246.
Lintner, J. (1956), “Distribution of incomes of corporations among dividends, retained earnings, and taxes”,
American Economic Review, Vol. 45, 97-113.
Lonie, A., Abeyratna, G., Power, D. and Sinclair, C. (1996), “The stock market reaction to dividend announcements. A UK study of complex market signals”, Journal of Economic Studies, Vol. 23, pp. 32-52.
Mackinlay, A., (1997), “Event Studies in Economics and Finance”, Journal of Economic Literature, Vol. 35, pp. 13-39.
Meusen, G. (2000), “Income Tax Act 2001”, European Taxation, vol. 40, pp. 490-498.
Michaely, R., Thaler, R. and Womack, K. (1995), “Price reactions to dividend initiations and omissions: Overreaction or drift?”, Journal of Finance, Vol. 50, 573-608.
Miller, M.H. and Modigliani, F. (1961), “Dividend policy, growth, and the valuation of shares”,
Journal of Business, Vol. 34, pp. 411-433.
Miller, M.H. and Rock, K. (1985), “Dividend policy under asymmetric information”, Journal of
Finance, Vol. 40, pp. 1031-1052.
Myers, S. and Maljuf, N. (1984), “Corporate financing and investment decisions when firms have information investors do not have” Journal of Financial Economics, Vol. 13, 187-221.
Richardson, G., Sefcick, S. and Thompson, R. (1986), “A test of dividend irrelevance using volume reactions to a change in dividend policy”, Journal of Financial Economics, Vol. 17, pp. 313-333.
Rozeff, M. S. (1982), “Growth, Beta, and Agency Costs as Determinants of Dividend Payout Ratios”
Journal of Financial Research, Vol. 5, pp.249-259.
Shefrin, H. and Statman, M. (1984), “Explaining investor preference for cash dividends”, Journal of
Financial Economics, Vol. 13, pp. 253-282.
Venkatesh, P. (1989), “The impact of dividend initiation on the information content of earnings announcements and returns volatility”, Journal of Business, Vol. 62, April, pp. 175-197.
Yoon, P. and Starks, L. (1995), “Signaling, Investment Opportunities, and Dividend Announcements”, The
8.2 Appendix 2: Regression Robustness Tests
Appendix 2.1: Correlation Table of Independent Variables
INST LN_SIZE YIELD ANT EA INST 1 LN_SIZE 0.07 1 YIELD 0.06 -0.26** 1 ANT -0.03 0.13 -0.01 1 DTAX 0.29 0.35 -0.16 -0.15 1 EA -0.01** -0.18 0.11 0.05 -0.33*** * Statistically significant at the 0.1 level
** Statistically significant at the 0.05 level *** Statistically significant at the 0.01 level
Appendix 2.2: Heteroscedasticity Tests
White’s Test
