Master Thesis Financial Management. The Dividend Price Shock and Taxes in the Netherlands

Transcription

1 Master Thesis Financial Management The Dividend Price Shock and Taxes in the Netherlands Name: Quinten Blok BSc. ANR: S Supervisor: Dr. V. P. Ioannidou Chair of committee: Prof. dr. S.R.G. Ongena Faculty: Economics and Business Administration Date: July

2 Abstract In a perfect capital market, share prices should fall exactly by the amount of dividend on the ex-dividend day. Empirical evidence however shows that stock prices on the ex-dividend day drop less than the dividend distributed. This paper examines the ex-dividend stock prices behavior in the Netherlands between 2002 until 2009, a period in which dividend taxation decreased from 25 percent to 15 percent. A study has been conducted based on a sample of 437 ex-dividend day events for 43 different large and middle size Dutch companies listed at the Amsterdam Stock Exchange. An average price drop to dividend ratio is found of 0.81 before the tax change in 2007 and 0.75 after the dividend tax change. The empirical results of this paper show decreasing ex-dividend day price drops when dividend tax is being reduced. Microstructure arguments explaining ex-dividend day behavior of stocks are generally absent in the Netherlands. Clientele effects predict that stocks with higher dividend yield should produce dividend price drop ratios closer to one. Some estimated Dividend Price Drop Ratios found are positively related to the annual dividend yield in the Netherlands. These results provide moderate support for the hypothesis that clientele effects in the Dutch stock market exist. 2

3 Table of Contents Abstract... 2 Table of Contents Introduction Literature overview Differential taxation of dividends and capital gains Changes in tax policies The interaction of taxes and transaction costs to trading stock: the effects of traders Market microstructure: Price discreteness and bid-ask bounce Tax Treatment of Dividends and Capital Gains in the Netherlands Taxation of Dividends Taxation of Capital Gains of Residents Taxation of Capital Gains and Dividends of Corporations Data and Methodology Section Research Hypotheses Sample Selection and Data Sources Descriptive Statistics Methodology Ex-dividend day identification Ex-Dividend Day Price Drop Using Opening/ Closing Prices Benchmark Model The Market Model Hypotheses with Market Adjusted Data Microstructure Effects Hypotheses to test Microstructure Effects Microstructure Effects and Adjustment for Daily Stock Returns The 2007 Dividend Tax Reform Clientele Effects

4 4. Empirical Results Conclusion Dividend Tax Change Dividend Clienteles Conclusion Recommendations References Appendix A1: Beta s Beta of AEX Stocks (Compared to AAX Index) Beta of AMX Stocks (Compared to AAX Index) Appendix A2 Descriptive statistics Descriptive statistics of dividend payments Descriptive statistics of 2002 dividend payments Descriptive statistics of 2003 dividend payments Descriptive statistics of 2004 dividend payments Descriptive statistics of 2005 dividend payments Descriptive statistics of 2006 dividend payments Descriptive statistics of 2007 dividend payments Descriptive statistics of 2008 dividend payments Descriptive statistics of 2009 dividend payments Whole sample Descriptive statistics of DPDR and Market Adjusted DPDR (DPDRMA) Pre 2007 Descriptive statistics of DPDR and Market Adjusted DPDR Post 2007 Descriptive statistics of DPDR and Market Adjusted DPDR First Quintile (20 percent smallest D/Pc) Second Quintile (20/40 percent D/Pc) Third Quintile (40/60 percent D/Pc) Fourth Quintile (60/80 percent D/Pc) Fifth Quintile (80/100 percent D/Pc) Appendix A3: Regressions Regression Estimates of the DPDR (Whole sample with intercept)

5 Regression Estimates of the Market Adjusted DPDR (Whole sample with intercept) Regression Estimates of the DPDR (Whole sample without intercept) Regression Estimates of the market Adjusted DPDR (Whole sample without intercept) Regression Estimates of the DPDR with Opening prices (Whole sample with intercept) Regression Estimates of the DPDR with Opening prices (Whole sample without intercept) Regression Estimates of the DPDR ( period with intercept) Regression Estimates of the market Adjusted DPDR ( period with intercept) Regression Estimates of the DPDR ( period without intercept) Regression Estimates of the DPDR (Post 2007 period with intercept) Regression Estimates of the Market Adjusted DPDR (Post 2007 period with intercept) Regression Estimates of the DPDR (Post 2007 period without intercept) Regression Estimates of the Market Adjusted DPDR (Post 2007 period without intercept) Dividend Tax Change Regression Estimates (with intercept) Dividend Tax Change Regression Estimates (without intercept) Dividend Tax Change Market Adjusted Regression Estimates (with intercept) Dividend Tax Change Market Adjusted Regression Estimates (without intercept) Dividend Tax Change Estimates with Opening Prices (with intercept) Dividend Tax Change Estimates with Opening Prices (without intercept) Appendix A4: Clientele Regressions Clientele Effects Unadjusted (with intercept): 1 st Quintile Clientele Effects Unadjusted (with intercept): 2 nd Quintile Clientele Effects Unadjusted (with intercept): 3 rd Quintile Clientele Effects Unadjusted (with intercept): 4 th Quintile Clientele Effects Unadjusted (with intercept): 5 th Quintile Clientele Effects Unadjusted (without intercept): 1 st Quintile Clientele Effects Unadjusted (without intercept): 2 nd Quintile Clientele Effects Unadjusted (without intercept): 3 rd Quintile Clientele Effects Unadjusted (without intercept): 4 th Quintile Clientele Effects Unadjusted (without intercept): 5 th Quintile Clientele Effects Market Adjusted (with intercept): 1 st Quintile

6 Clientele Effects Market Adjusted (with intercept): 2 nd Quintile Clientele Effects Market Adjusted (with intercept): 3 rd Quintile Clientele Effects Market Adjusted (with intercept): 4 th Quintile Clientele Effects Market Adjusted (with intercept): 5 th Quintile Clientele Effects Market Adjusted (without intercept): 1 st Quintile Clientele Effects Market Adjusted (without intercept): 2 nd Quintile Clientele Effects Market Adjusted (without intercept): 3 rd Quintile Clientele Effects Market Adjusted (without intercept): 4 th Quintile Clientele Effects Market Adjusted (without intercept): 5 th Quintile Clientele Effects Opening Prices (with intercept): 1 st Quintile Clientele Effects Opening Prices (with intercept): 2 nd Quintile Clientele Effects Opening Prices (with intercept): 3 rd Quintile Clientele Effects Opening Prices (with intercept): 4 th Quintile Clientele Effects Opening Prices (with intercept): 5 th Quintile Clientele Effects Opening Prices (without intercept): 1 st Quintile Clientele Effects Opening Prices (without intercept): 2 nd Quintile Clientele Effects Opening Prices (without intercept): 3 rd Quintile Clientele Effects Opening Prices (without intercept): 4 th Quintile Clientele Effects Opening Prices (without intercept): 5 th Quintile

7 1. Introduction The influence of taxes on stock prices around dividend payments is especially of interest in the field of finance. It indicates the relative valuation of dividend payments versus capital gains on stocks. In a perfect market, the share price of a stock should fall exactly with the dividend amount distributed. However, previous research has indicated that on average share prices fall less than the dividend amount paid out. Elton and Gruber (1970) provide one of the first and most popular explanations for this phenomenon, and being examined even further by a large amount of other papers. They suggest that shareholders only care about after tax returns and would form dividend clienteles based on their relative taxation. Investors in low marginal tax brackets prefer stocks with high dividend (and a lower expected return in the form of capital gains) yields due to their relatively low taxation, while those in high marginal tax brackets concentrate their holdings in low dividend yield stocks, which are expected to have a larger capital gain yield. Others reject the tax clientele explanation of Elton and Gruber and indicate that arbitrage by short term traders and other microstructure effects affect the ex-dividend day price behavior. No consensus is reached on whether taxes of microstructure effects provide justification for the ex-dividend day stock price behavior. This paper examines the ex-dividend day behavior of stocks in the Netherlands using data from dividend events of listed stocks between 2002 and The ex-dividend behavior of the Dutch stock market examined in this paper is particularly interesting because of some unique institutional and specific features in the Dutch Tax System, including a dividend tax cut in 2007 as well as domestic investors for who withheld dividend tax functions as an advance levy. Frictions observed in other markets which prevent stock prices to drop with the distributed dividend, such as the economic double taxation of dividends for domestic shareholders, are nonexistent in the Netherlands. 7

8 The objective of this paper is to analyze the relation between the behavior of stock prices around the exdividend day, the influence of taxes and market efficiency in the Netherlands. This paper examines the influence of dividend taxes and existing microstructure effects on the ex-dividend behavior of Dutch stocks. Secondly, the impact of a major change in dividend taxation introduced in 2007 is being investigated. Finally, the relationship between dividend yield and dividend price drop ratio is being studied. After the theoretical chapter of this paper in chapter two, providing an overview on the extensive literature on the ex-dividend day behavior and explaining the relevant Dutch tax laws, the remainder of this paper is organized as follows. Chapter three describes the data and methods used to test and chapter four presents the empirical results which were found conducting this study. Chapter five contains the conclusions and chapter six provides some recommendations for future research. 8

9 2. Literature overview The behavior of stock prices around the ex-dividend days has been the subject of extensive theoretical and empirical research. Since the revolutionary paper of Elton and Gruber (1970) on the ex-dividend stock price behavior and the influence of taxes, more that hundred papers have been published giving support to, refining or questioning the findings of Elton & Gruber. These findings can generally be categorized into four categories. 1 The first group of papers provides an explanation of the theory of Elton & Gruber and replicates their tests on other markets or in other time periods. A second group reexamines the theory of Elton & Gruber in specific periods with changes in tax policy. The third group of papers investigates the influence of arbitrage by short term traders on the ex-dividend price drop. The last group of papers questions the tax explanation of the ex-dividend day stock price behavior of Elton & Gruber by introducing a market microstructure explanation. This chapter will present an overview of all literature on the ex-dividend stock price behavior and the influence of taxes. 2.1 Differential taxation of dividends and capital gains The closing price of a stock on the last cum dividend day consists of two components: the current dividend and the present value of all the future dividends. According to the capital structure theory of Modigliani and Miller (1961), in a perfect market with no taxes or transactions costs stock prices would exactly fall on the ex-dividend day with the value of each dividend that is paid. Campbell and Beranek (1955) investigated as one of the first the ex-dividend day price drop and found an ex-dividend day price drop of 90 percent of the paid out dividend. The revolutionary paper Elton and Gruber (1970) studied the behavior of stocks around ex-dividend days by testing for tax effects in the pricing of stocks. Corrected for the one day expected rate of return and in the absence of taxes and transaction costs, the price drop and the dividend should be equal. Elton and Gruber however find that during their research period (April, 1966 to March, 1967) the average ex-day share price drop to average dividend ratio, P/D, of all stocks at the New York Stock Exchange was Elton and Gruber interpreted their results as evidence of price influences by taxes around the ex-day since dividends and capital gains are taxable at different rates in the United States. 1 Elton, E., Gruber, M. and Blake, C. (2004). Marginal stockholder tax effects and ex-dividend day behavior- Thirty-two years later, page 1. 9

10 The drop in stock price on the ex-dividend day should indicate the value of the dividend compared to the capital gain of the marginal shareholder. Elton and Gruber also suggest that investors might form tax clienteles, since estimated marginal tax rates of investors decline with dividend yields of stocks. Investors in low marginal tax brackets prefer stocks with high dividend (and a lower expected return in the form of capital gains) yields due to their relatively low taxation, while those in high marginal tax brackets concentrate their holdings in low dividend yield stocks, which are expected to have a larger capital gain yield. In their study on an after tax version of the Capital Assets Pricing Model, Litzenberger and Ramaswam (1979) find evidence for a dividend clientele effect. In line with the empirical results of Elton and Gruber (1970), Litzenberger and Ramaswam discover a lower aversion for dividends relative to capital gains for high yield stocks and a higher aversion for low yield stocks although they indicate comprehensive research on institutional restrictions on short sales and personal taxes is needed. Eades, Hess and Kim (1984) reexamine the findings of Elton & Gruber with different samples. When using a sample of dividend payments of common stock, the results of Eades, Hess and Kim confirm the findings of Elton & Gruber (1970). However, according to the results of Eades, Hess and Kim the excess returns on ex-dividend days for taxable preferred stock dividends, in contradiction to the results for normal stocks, cannot be completely explained by the tax hypothesis of Elton & Gruber. They also question, based on their findings, whether differential tax rates between capital gains and dividend payments can be inferred from the average ex-day share price drop to average dividend ratio. Eades, Hess and Kim check for possible explanations for their ex-dividend period results for taxable preferred stock dividends, like the influence of the day in the week effect, dividend announcement effects or infrequent trading. According to Eades, Hess and Kim, none of these explanations could clarify their results. Eades, Hess and Kim find, as Boyd and Jagannathan (1994) describe it characteristically, a run up in prices before the ex-dividend day and a run down after the ex-dividend day, which could possibly be explained by the influence of traders. McDonald (2001) investigates the ex-dividend behavior in Germany. German dividends usually carry a tax credit (of percent from 1994 to 1998, but eliminated in October 2000) to compensate domestic shareholders, at least to some extent, for double taxation and creating a wedge between the returns of domestic and foreign shareholders. This tax credit attached to dividend payments makes 10

11 dividends more valuable to domestic than foreign shareholders and creates possibilities for cross border tax arbitrage. McDonald finds an ex-dividend day premium of 1.26, although McDonald s calculated tax consideration should imply a value of the ex-dividend day premium of He concludes only sixty percent of the dividend tax credit is embedded in German stock prices and option and futures prices impound about fifty percent of the tax credit. This evidence is according to McDonald in line with tax induced trading and with German tax arbitrageurs facing tax risk with dividend capture transactions. McDonald discovers abnormal trading volume from six days up to the ex-dividend day, consistent with tax induced trading before the ex-dividend day by foreign shareholders who are selling their German stocks. 2.2 Changes in tax policies Michaely (1991) studies the effects of the 1986 Tax Reform Act in the United States on the behavior of stock prices around the ex-dividend day. The 1986 Tax Reform Act reduced in 1987 and eliminated in 1988 the difference in tax treatment of realized long-term capital gains and dividend revenue and fiscally treated those forms of income similarly. The 1986 Tax Reform Act did not have any impact on the fiscal position of short term traders and had only a small impact on the fiscal position of corporate traders. Michaely finds similar average ex-day share price drop to average dividend ratios before and after the 1986 Tax Reform Act. He suggests that the change in tax treatment of realized long-term capital gains and dividend revenue has no significant influence on the behavior of stock prices during the ex-dividend day. Michaely also finds ex-dividend day ratios above one for high yield stocks, attributing them to the role of corporate traders who favor dividend income over capital gain income. When comparing the average ex-dividend day ratios between different time periods, Michaely signals significantly lower (and below one) ratios in the sixties compared to 1986/1986. Michaely attributes these differences to other effects than the 1986 Tax Reform Act, like the change in transaction costs, the emergence of liquid futures and options markets and changes in investors' knowledge or attitudes toward risk. Poterba and Summers (1984) analyze data on stocks listed in the United Kingdom (U.K.) to study the relative valuation of dividend income and capital gains in the United Kingdom. Using British data, offers Poterba and Summers the opportunity to illuminate the dividend valuation question because of two 11

12 far-reaching transformations and different smaller scale changes in British corporate tax policy, affecting both individual investors and professional stock traders. The first tax reform by the newly elected Labour Government occurred in 1965 and a capital gains tax of 30 percent was introduced. This tax reform should have led to an increased preference of investors to dividend income. In 1973, a Conservative government instituted an integrated corporate income tax, effectively reducing the personal and corporation s dividends tax and refunding the dividend tax to tax exempt investors (please see also Bell and Jenkinson (2002)). Poterba and Summers find a substantial effect of changes in dividend taxation on the premium required by investors to induce them to receive income in the form of dividends over capital gains. Their results show that investors are aware of dividend taxes and that dividend taxation affects the premiums required by investors to induce them to receive returns in the form of dividends. Since Poterba and Summers notice a changing valuation of dividends across different tax regimes, they state that taxes account for part of the positive relationship between yields and stock market returns. Their results support the theory of Elton en Gruber, but Poterba and Summers point out that their results of changes in the specific British tax code cannot be generalized to the United States. Bell and Jenkinson (2002) examine the impact of a major change in the United Kingdom in the system of dividend taxation in July The reform focused on pension funds, the largest single shareholders group in the U.K.. In contrast to the U.S., where dividends are taxed more heavily compared to profits retained within the firm, dividend payments were tax preferred by certain classes of invertors in the U.K. until July The main aim of the tax reform was to make tax-exempt investors indifferent, at least in tax terms, between dividends and retained earnings. Before the tax reform, tax exempt investors, like pension funds, paid Advance Corporate Tax (ATC) of twenty percent on dividend distributions but obtained a full refund of the paid ATC. This system led to a strong preference of tax exempt investors for dividends over retaining profits in a company. The tax reform withdrew the possibility of tax-exempt investors to reclaim the ACT on dividend distributions. What makes the U.K. particularly interesting is the relevance of short term trading. Since short term trading is seriously constrained by tax restrictions according to Lasfer (1995), ex-dividend day returns are not affected in the U.K.. The results of Bell and Jenkinson are in line with the tax clientele hypothesis of Elton & Gruber (1970). Bell and Jenkinson find strong dividend clientele effects with positively related average ex-day share 12

13 price drop to average dividend ratios, P/D, to dividend yield. They also discover significant changes in P/D for high dividend yield companies before and after the tax reform. Since pension funds had a strong preference for investing in high yield stocks, this provides strong evidence that, according to Bell and Jenkinson, pension funds were the effective marginal investors in high-yielding stocks and that taxation influences the valuation of companies. In 2000, Kalay and Michaely study the effects of a different fiscal treatment of capital gains and dividend. Kalay and Michaely reexamine findings of Litzenberger and Ramaswam (1979) whether their results are the effect of cross sectional differences in dividend yield (risk-adjusted pretax returns should be positively correlated to dividend yields) or the results of Litzenberger and Ramaswam could be explained by time-series variations between dividend and non-dividend paying months. They discover that the time horizon to identify and calculate the dividend period is a crucial issue to interpret research results. In line with the results of Michaely (1991), the findings of Kalay and Michaely indicate that the long term risk adjusted returns are not correlated with dividend yields. They suggest that this effect might be explained by a complex theory of tax effects. 2.3 The interaction of taxes and transaction costs to trading stock: the effects of traders Lakonishok and Vermaelen (1986) study trading volume before and after ex-dividend days on the NYSE between January, 1970 and December, They find significant higher trading volumes, especially for high yield and actively traded stock and following the start of negotiable brokerage commissions. Lakonishok and Vermaelen argue that their results are in line with the potential and dynamic tax related trading on the ex-dividend day of a stock. Since they observe most of the volume increase not immediately after the ex-dividend day but some days later, they suggest that corporate traders, subject to a sixteen-day holding rule, could be held responsible for (a large part of) the abnormal volume. Lakonishok and Vermaelen suggest that stock returns could indicate that the degree of trading activity in ex-dividend months is influenced by transaction costs, the tax status of traders or possibilities to hedge against price risk. During their data sample between 1970 and 1981 and during tax law situation at that point in time, Lakonishok and Vermaelen show that incorporated investors prefer high dividends since cum-dividend/ex-dividend trading allows them to acquire significant tax benefits. In the three days 13

14 before the ex-dividend day, high yield stocks even increase abnormally by 1 percent as a result of competition for the tax benefits. Consistent with the explanation of Lakonishok and Vermaelen, Karpoff and Walkling (1988) discover that larger transaction costs lead to higher abnormal ex-dividend day returns for high-yielding stocks in the United States. Karpoff and Walkling investigate whether short term traders are the marginal investors in dividend paying stocks on the ex-dividend days. They state that in case of short term trading, the exdividend day stock returns should be influenced by transaction costs. Karpoff and Walkling investigate a large sample bid-ask data of NASDAQ stocks between 1973 and 1985 and find a strong correlation between ex-dividend day returns and the bid-ask-spreads. Karpoff and Walkling indicate that dividendcapture trading is an important part of the dividend puzzle and that the interpretation of ex-dividendday returns is more complicated than implied by the tax-penalty hypothesis associated with Elton and Gruber (1970). According to Karpoff and Walking, the tax based and the short term trading explanations of ex-dividend day behavior of stocks are not competing but complementary. Vila and Michaely (1995) also study the relationship between tax heterogeneity and the behavior of stock prices and the trading volume of stocks on the ex-day by using an equilibrium model. They find evidence of abnormal trading volume and show that the trading volume around the ex-dividend day holds information about the tax preference of investors. They discover that trading volume around the ex-dividend day is positively related to the tax rate heterogeneity across investors and the dividend size, but negatively related to the variance of the dividend paying security. Vila and Michaely also state that tax induced trading of stock traders, who have a required holding period to retain a specific dividend exclusion, around the ex-dividend day could be subject to considerable fundamental risk. Traders are exposed to overnight risks between the cum-dividend and the ex-dividend day of the stock. Information about the economy, like unemployment or money supply, or information on a specific company is often published before the opening of the exchange and effect prices during non-trading hours. Vila and Michaely also suggest that the ex-dividend day risk premium in stock prices, because of considerable fundamental risk, could possibly account for the ex-dividend day excess return. Still the question remains whether or not the differential tax treatment of dividends and capital gains can explain the ex-day share price drop to average dividend ratio results of Elton and Gruber. Kalay (1978) states that a difference between the dividend and the ex-day price drop if based on tax 14

15 treatment would be arbitraged away by tax exempt institutions, short term traders or corporations facing relatively lower taxes on dividends than on capital gains. Kalay (1982) re-examines the ex-dividend day behavior of stock prices and the clientele effect. He indicates that in previous studies empirical evidence of the positive correlation of the ex-dividend price drop of the stock and the dividend yield could be a result of an incomplete adjustment for normal stock price movements and the use of closing prices on the ex-dividend day. Kalay shows, after adjusting for potential biases, that in contrast to previous studies the ex-dividend day behavior of stock prices is not necessarily the result of tax induced clientele effects since marginal tax rates cannot be inferred from the ex-dividend price drop. However, he finds a positive correlation between the relative ex-dividend price drop and the dividend yield, consistent with the tax effect and tax induced clientele effect. He also finds that transaction costs may be important since they make certain tax arbitrage strategies unprofitable. Boyd and Jagannathan (1994) acknowledge that not every investor in the United States has the same tax-induced preference for capital gains over dividend payments. Several groups of investors, like tax exempt pension funds or market makers, face the same taxation of capital gains and dividend income. Boyd and Jagannathan also include in this group corporate investors because of the corporate tax treatment of their dividends. Corporate investors prefer to capture dividends since a large fraction of the dividend income is excluded from corporate taxation. Boyd and Jagannathan empirically investigate the ex-dividend behavior of stock prices and develop a model with different classes of traders facing different transaction costs. Boyd and Jagannathan find, in line with their developed theory, a non-linear relationship between ex-dividend day price movements and the dividend yield. 2.4 Market microstructure: Price discreteness and bid-ask bounce Bali and Hite (1998) argue that the price-drop-to-dividend-ratio can be influenced by the tick size of a stock. Bali and Hite investigated cash dividends paid out by firms listed on the New York Stock Exchange between July, 1962 and December, While dividends generally are continuous, stock prices have a minimal price increment or tick size. Due to the constraint of the tick size, the price drop on the exdividend day cannot always equal the dividend amount. Bali and Hite argue that because dividends are usually small, discreteness plays an important role in explaining stock price movements on the ex- 15

16 dividend day. When the dividend increases, the ex-dividend day share price drop ratio approaches one, but this ratio declines between ticks showing a saw tooth pattern, influenced by the tick size. Bali and Hite find that the discreteness causes abnormal returns on the ex-dividend day to be even smaller than the rounding error. The abnormal returns are too small to generate arbitrage profits around the exdividend day, since the tick size multiples restrict continuous trading prices. Bale and Hite also question the findings of Elton and Gruber (1970) on tax-induced dividend clienteles as incontrovertible evidence, since discrete trading prices could also account for the ex-dividend day share price drop ratio increasing with dividends. Based on their research results Bali and Hite however cannot rule out the evidence of these tax induced clienteles. Frank and Jagannathan (1998) present a microstructure explanation for the P/D-ratio based on their findings on the Hong Kong stock market. In Hong Kong similar ex-dividend day price effects as in the United States are being observed, although neither dividends nor capital are taxed in Hong Kong. Due to the fact that neither dividends nor capital gains are taxed, the tax theory predicts that the P/D should equal to zero. Frank and Jagannathan hypothesize that the collection and reinvestment of dividend will be done by market makers, since this to bothersome for individual investors, and state that market makers have a comparative cost advantage and buy stocks cum dividend at bid price, receive the upcoming dividend and sell these stocks at the ask price after receiving the dividend. The bid-ask price movement results in a positive price movement. When this microstructure effect is taken into account, Frank and Jagannathan find that dividends and capital gains are equally valued. Kadapakkam (2000) however investigates the abnormal ex-day returns at the Hong Kong stock market when it improved conditions for short term trading after it switched from a cumbersome physical settlement procedure to an electronic procedure. After switching to an electronic procedure, the system enabled short-term arbitrage trades around the ex-dividend day by removing physical delivery requirements. Kadapakkam finds that the abnormal ex-dividend day returns almost drop an insignificant 0.17 percent. He points out that this drop is more pronounced for high-yield stocks, being more attractive for quick turnaround arbitrage trades around ex-dividend days. Graham, Michaely and Roberts (2003) question the price discreteness claimed by Bali and Hite. Price discreteness was greatly reduced by a decreased tick size from 1/8 th to 1/16 th in 1997 and to decimals in 2001 on the New York Stock Exchange (NYSE). Graham, Michaely and Roberts examine the effect of the 16

17 refinement of the stock pricing grid on the average ex-day share price drop to average dividend ratio and find a decreasing P/D-ratio. Their findings, a decrease of the ex-dividend day share price drop ratio, are inconsistent with the microstructure explanations of price movements on the ex-day of Bali and Hite, since a decline, a minimum tick size and the associated reduction in bid-ask spread should increase the P/D-ratio, especially for those stocks experiencing the greatest reduction in spreads. In case price discreteness (Bali and Hite (1998)) and the bid-ask bounce (Frank and Jagannathan (1998)) influences ex-dividend day share price movements, a reduction of the tick size could lead to a reduction of the bid-ask spread and an increase in dividend arbitrage around the ex-dividend day, the P/D-ratio becoming closer to one. Graham, Michaely and Roberts cast doubt whether the tick size and the bid-ask spread could be the dominant cause of the positive association between yields and premiums, since they do not discover a change in the relation of dividend yield and dividend premiums when bid-ask spreads and price discreteness drop. Although the results of Graham, Michaely and Roberts are in line with the tax explanation, they suggest that ex-dividend day pricing patterns might be caused by a phenomenon that has not been identified in the financial economics literature. Additionally, Graham, Michaely and Roberts (2003) investigate the reduction in the capital gains tax rate from 28% to 20% in the United States on the 7 th of May, A relative increase of dividend taxation to capital gains should, based on the tax argument of Elton and Gruber, reduce P/D-ratio (the ex-dividend premium). Graham, Michaely and Roberts find in line with the tax explanation of Elton and Gruber a falling dividend premium in 1997 after the capital gains tax was reduced, since dividends are taxed more disadvantaged compared to capital gains. Thirty-two years after their revolutionary paper, Elton, Gruber and Blake (2002) reexamine the effects of taxes on the ex-dividend day behavior of stocks. Elton, Gruber and Blake acknowledge that the microstructure effect has become the most serious alternative to the tax argument after publishing their original paper in They state that microstructure arguments can only explain a price drop less than the dividend in the absence of taxes, since market makers will buy stocks at the ask price of the cum dividend day, receive the dividend and sell the stock ex-dividend. Elton, Gruber and Blake study whether the microstructure explanation dominates ex-dividend-day behavior of stocks by examining the tax-free distributions of municipal bond funds. They find new evidence for the traditional ex-dividend tax effects. For the tax-free distributions of municipal bond funds Elton, Gruber and Blake find a price drop larger 17

18 than the tax exempt dividend, explained by a tradeoff between the tax free dividend and the taxable capital gains. In the case of non tax exempt funds Elton, Gruber and Blake also find supporting evidence for the ex-dividend tax effects since the fall of the stock price during the ex-dividend period is less than the dividend in periods when capital gain taxes are lower than dividend taxes. 2.5 Tax Treatment of Dividends and Capital Gains in the Netherlands In the United States, corporate income is double taxed. First, corporate income is taxed by corporate taxes and secondly, accumulated profits are taxed with dividend taxes when distributed to shareholders. In the Netherlands, the tax system is specifically designed to reduce the multiple taxation of income by providing compensation to shareholders for dividend taxes. Therefore, the Dutch tax system can be classified as an integrated or imputation tax system. Since special tax rules 2 apply in the Netherlands for large percentage holdings in companies, this paper only takes small ownership (less than 5 percent of the outstanding shares) into account. Since a overall explanation of dividend taxation in the Netherlands is beyond the range of this study, this paper presents the key issues Taxation of Dividends In the Netherlands, earnings distributed as dividends by Dutch tax-resident companies are taxed with a 15 percent statutory Dividend Withholding Tax. Based on article 5 of the Dutch Dividend Tax Act (Wet op de dividendbelasting 1965), the distributing company (based on efficiency, tax fraud prevention and pay-as-you-go arguments 3 ) has to withhold 15 percent Dividend Withholding Tax from the dividend distributed to a shareholder, independent of the origin or home country of the shareholder. Only domestic shareholders, i.e. resident in the Netherlands, are generally entitled to a full tax credit of the Dividend Withholding Tax paid, which can be offset against income tax liabilities, or can be received as a refund of the Dividend Withholding Tax. The paid Dividend Withholding Tax functions as a advance levy. The purpose of this advance levy for domestic shareholders is to neutralize the effects of double taxation. Foreign holders of stock are not qualified for the Dutch Dividend Withholding Tax credit, since the Dutch government cannot fully levy taxes on non-resident shareholders. Foreign countries in general do not recognize the Dutch Dividend Withholding Tax (credit). 2 For participations ( Deelnemingen ) of 5 percent of the outstanding stock, special tax regimes apply for individuals in the Dutch Income Tax Act 2001 (Wet inkomstenbelasting 2001) and for corporations in the Corporations Tax Act 1969 (Wet op de vennootschapsbelasting 1969). 3 Marres, O.C.R. and Wattel, P.J. (2006). Dividendbelasting, Fed Fiscale Studieserie, page 4. 18

19 For foreign shareholders the Netherlands has a wide range of tax treaties (often based on the standardized OECD Model Convention) with of different nations trying to minimize double taxation on income and dividends. The dividend declared in a Dutch company by a foreign shareholder is subject to 15 percent statutory Dividend Withholding Tax, irrespective of whether the shareholder is entitled to benefits of a tax treaty concluded with the Netherlands. In case a tax treaty of The Netherlands with the shareholder s home country would offer a Dividend Withholding Tax rate lower than 15 percent, the foreign shareholder can request a refund of the paid Dividend Withholding Tax surplus. On January 1, 2007 the tariff of the statutory Dividend Withholding Tax was lowered from 25 percent to 15 percent to make the Netherlands a more attractive country to invest in. As a result of the reduction of the statutory Dividend Withholding Tax rate, the amount of additional refunds of Dividend Withholding Taxes to residents outside The Netherlands under the double tax treaties concluded with the Netherlands will be very limited. Since foreign shareholders face an effective Dividend Withholding tax rate of 15 percent under the current Dutch law, a 1 gross dividend would hold a net worth 0.85 for a foreign investor. For a taxable shareholder residing in the Netherlands, a gross dividend of 1 results in a net value of 0.85 after Dividend Withholding tax. In contrast to foreign shareholders, the 0.15 Dividend Withholding tax of taxable domestic shareholders can be offset against income tax liabilities or can be refunded. As a result, shareholders, based in the Netherlands, face an effective tax rate of zero percent. The differential tax treatment of dividends distributed to Dutch and foreign shareholders, creates an incentive for foreign investors to transfer stocks to Dutch shareholders during the dividend payout period. Several rules apply to prevent Dutch shareholders to arbitrage their tax advantage. The Dutch tax authorities reserve the right to invalidate transactions having no other purposes than reclaiming the Dividend Withholding Tax paid as an advance levy. Only when the economic owner of the shares is also the one entitled to the (taxed) dividend, the paid Dividend Withholding Tax can be qualified as an advance levy, according to the Dutch Income Tax Act (Wet Inkomstenbelasting 2001) and Corporations Tax Act (Wet op de Vennootschapsbelasting 1969) 5. This prevents custodial holders of stocks to reclaim the Dividend Withholding Tax paid by customers as an advance levy. 4 Article 9.2, paragraph 2 and paragraph 3, Wet Inkomstenbelasting Article 25, paragraph 2 and paragraph 3, Wet op de vennootschapsbelasting

20 2.5.3 Taxation of Capital Gains of Residents The taxation of capital gains of individuals and corporations is more complicated. In contrast to the United States, realized capital gains of residents are not subject to direct taxation in The Netherlands. The Dutch Income Tax Act 2001 distinguishes three types of taxable income of residents, classified into three boxes. The third box contains the taxable income from savings and investments, like stocks. A notional yield of 4 percent on the average net capital (assets minus liabilities, although a threshold applies in 2010 for the first euro of assets 6 and 2900 euro of liabilities 7 ) during the calendar year is taxable at a flat rate of 30 percent, resulting in a 1.2 percent duty (4 notional yield * 30 percent flat tax rate) over the average net capital. The actual level of return on the average net capital is not relevant Taxation of Capital Gains and Dividends of Corporations The Netherlands levies Corporate Tax based on the Corporations Tax Act on the worldwide income of practically every corporate entity established and residing in the Netherlands. Corporate Tax is levied a rate of 25,5 percent in the Netherlands. A lower rate of 20 percent corporate tax rate applies for the first of the taxable profits. Capital gains and losses on stock holdings and dividends are treated as ordinary corporate income and are, as such, included in the taxable profits. Based on the general rule of sound business practice (Goed Koopmansgebruik), capital gains are generally not taxed until realized. Capital losses can be deducted on an accrual basis. Important to mention, this paper only takes small ownership (less than 5 percent of the outstanding shares) into account, since capital gains on holdings of corporations of less than five percent are taxed as ordinary income in the Netherlands. Different rules apply for holding five percent or more of the outstanding stock. This chapter has provided an overview of the extensive literature on the ex-dividend day behavior and explained the relevant Dutch tax laws involved in ex-dividend day price behavior. In the next chapter, the data and methodology part of this study is presented. 6 Article 5.5, paragraph 1, Wet Inkomstenbelasting Article 5.3, paragraph 3, e, first phrase Wet Inkomstenbelasting

21 3. Data and Methodology Section This part of the paper provides a detailed description about the data used to investigate whether stock prices in the Netherlands are affected around the ex-dividend day by dividend taxes in The Netherlands and whether dividend clienteles based on the Dutch tax-code exist. By examining the ex-dividend day behavior of share prices, the valuation of dividend payments is studied. 3.1 Research Hypotheses In a perfect, frictionless market, as a starting point the price of a stock should drop exactly with the dividend payment from the last day the stock trades with the dividend, the cum dividend day, to the next day when the stock goes ex-dividend, the ex-dividend day. This economy can be explained, based on Kalay (1982), by the following assumptions: Assumption 1: No transaction costs exist; Assumption 2: Capital gains taxation and dividend income taxation are known; Assumption 3: Unrestricted short sale possibilities exist; Assumption 4: Investors are risk neutral; Assumption 5: Stocks are traded at continuous trading prices; Assumption 6: All investors are subject to identical tax rates. The price movement of the stock in relation to the dividend payment can be mathematically described as ( )= (1) where Div is the amount of dividend paid out, P c the last closing price of the stock before going exdividend and P ex the closing price after the stock went ex-dividend. The Dividend Price-Drop-Ratio (DPDR), or referred to as the ex-dividend day premium, is calculated as follows: = (2) 21

22 According to Elton and Gruber (1970), market prices in the equilibrium will be formed in such a way that investors will be indifferent, ignoring discounting, between receiving the dividend and selling the share on the ex-dividend day and selling the share on the cum dividend day. Elton and Gruber expect, ignoring risk and transaction costs, the price-drop-to-dividend ratio in a rational market should reflect the value of dividend versus capital gains to the marginal stockholder. Since dividends and capital gains are taxed differently, the price-drop-to-dividend ratio should be affected by the relative taxation of dividends and capital gains. This market equilibrium tax hypothesis can, according to the theory of Elton & Gruber, be expressed as: P c T cap (P c P o ) = P ex T cap (P ex P o ) + Div (1- T div ) (3) where T div is the applicable Dutch tax rate on dividend income, T cap the applicable tax rate on capital and P o the original purchase price of the share. In this equilibrium, investors would possess the same amount of wealth whether selling before or after the stock trades ex-dividend. The tax hypothesis implies that the dividend price-drop-ratio equals one minus the taxes on dividend divided by one minus the taxes on capital. The market equilibrium tax hypothesis implies that shareholders are indifferent to continue holding the stock, receiving the dividend -possibly pay dividend taxes on the received dividend- and the price drop or selling the cum-dividend stock before it goes exdividend, possibly being taxed and not receiving the dividend. Rearranging (3) leads to: = = (4) In case the preferences of investors do not differ before and after the dividend payment date, the implied Dividend Price-Drop-Ratio should reflect the marginal tax rate of investors on dividend payments relative to capital gains. In the United States and many other countries profits distributed as dividends are tax wise disadvantaged compared to capital gains. The Dutch tax system is very different compared to the United States. The Dutch Income Act introduced in 2001 was designed to reduce the multiple taxation of income by providing compensation to domestic shareholders for dividend taxes. A 15 (pre 2007: 25 percent) percent Dividend Withholding Tax is withheld from the dividend distributed to a shareholder, independent of the origin or home country of the shareholder. For domestic 22

23 shareholders, a notional yield of 4 percent on the average net capital during the calendar year is taxable at a flat rate of 30 percent, resulting in a 1.2 percent duty over the average net capital. A corporate tax is levied on capital gains and losses on stock holdings and dividends of less that 5% of the outstanding shares of every corporate entity established and residing in the Netherlands at a rate of 25,5 percent. Domestic 8 shareholders are generally entitled to a full tax credit of the Dividend Withholding Tax paid, which can be offset against income tax liabilities, or can be received as a refund of the Dividend Withholding Tax. Foreign shareholders on the other hand, do not qualify for such a credit or refund although some tax treaties of the Netherlands with their domestic countries only partially provide relief by providing them a fractional refund in the Netherlands for the paid Dutch Dividend Withholding Taxes. Providing several classes of investors a partial credit for the withheld dividend tax but not to other classes of (foreign) investors, influences the value of the dividend for these different classes of investors. Since the taxation of effective capital gains rate in the Netherlands is equal to zero, the Dividend Price- Drop-Ratio can be determined as: h = =(1 ) (5) As described earlier, dividend taxes in the Netherlands are designed to immediately and significantly impact only non-domestic investors since domestic shareholders are able to reclaim the withheld dividend tax. Combining the theoretical arguments obtained earlier with the acquired knowledge of the Dutch tax system and assuming the inexistence of other factors influencing the price of dividend distributing stocks around the ex-dividend day, such as microstructure effects or transactions costs (Kalay 1982), one would expect that the estimated DPDR in the Netherlands will be smaller than one. Combining these arguments, leads to the first research hypothesis. Proposition 1: Foreign investors are the marginal investors in Dutch equity and price drop ratios are directly related to the quotient (1 ). As a result, the exdividend behavior of stock prices in The Netherlands, the Dividend Price Drop Ratio, is not significantly different from 0.75 before 2007 and 0.85 from Residents of the Netherlands are qualified as domestic by article 4 the Dutch Tax Code (Algemene Wet Rijksbelastingen (AWR)) based on circumstances. 23

24 Apart from the first proposition on the influence of taxes on the ex-dividend price behavior of equity, this paper also tests whether Dividend Price-Drop-Ratio might not even exist in the Netherlands. Proposition 2: Price drop ratios are not related to taxes. As a result, the ex-dividend behavior of stock prices in The Netherlands, the Dividend Price Drop Ratio, equals 1.00 before and after Thirdly, the clientele effect in the Netherlands is analyzed in this study. If taxes, as stated by Modigliani and Miller (1961) and Elton and Gruber (1970), would influence the value of dividends, this would imply that more heavily taxed (foreign) investors would prefer lower dividend yields stocks. Following the Elton and Gruber argument, stocks with higher dividends and dividend yields would have price to dividend ratios closer to one. This leads to the next proposition where the tax-clientele effects Elton and Gruber are tested. Proposition 3: As a result of the clientele effect, the Dividend Price Drop Ratio should be positively related to dividend yield, since more heavily-taxed investors show a preference for lower yielding stocks. Non tax-factors arising from transaction costs and market microstructure effects as studied by Bali and Hite (1998) and Frank and Jagannathan (1998) potentially influence the ex-dividend day behavior of prices. To overcome these issues, this study also analyses, based on Bell and Jenkinson (2002), the change in ex-dividend day behavior in response to the dividend tax reduction, being part of the Wet Werken aan Winst (Working of Profit Act), on 1 January The Act brought some major (budget neutral) changes in the Dutch Corporate Tax Act 1969 (Wet op de Vennootschapsbelasting 1969). The corporate tax rate was lowered from 29.6 percent to 25.5 percent but more strict rules on the depreciation were introduced. The Act also included a direct consequence for shareholders. Before 2007, dividend distributions were taxed at 25 percent. As from 1 January 2007, dividend tax was reduced to 15 percent. This change leads to the fourth research proposition. Proposition 4: In 2007, due to a dividend tax rate decrease from 25 percent to 15 percent, it is expected that the average DPDR will be significantly affected compared to the observed period. 24