Ex-dividend day stock price behaviour of German stocks

Transcription

1 Ex-dividend day stock price behaviour of German stocks Final Research Report Presented to The Graduate School of Business Copyright University of Cape Town UCT In partial fulfilment for the requirements for the Masters of Business Administration Degree By Manfred Garben November 2009 Supervisor: Dr. Chipo Mlambo

2 ACKNOWLEDGEMENTS I wish to thank my supervisor, Dr. Chipo Mlambo, from the Graduate School of Business, University of Cape Town, for her candid and valuable guidance during the writing of this research report. And finally I wish to thank my partner, Andrea Volkmann, who has been an incredible support for me during this year. Last but not least, a large thanks to my family and friends who stood by me throughout this process. This research report is not confidential. It may be used freely by the Graduate School of Business. I certify that that except as noted above the report is my own work and all references used are accurately reported. Signed: Manfred Garben Cape Town, November 2009 II

3 ABSTRACT The stock markets have just seen a disastrous decrease in the past 18 months with a steady rally in the past 6 months. Is investing in equity still safe? Does it make sense to invest in equity? Where else should I invest my money? These are only a few questions investors and analysts asked in the past months. The purpose of this research report is to investigate the share price behaviour of German stocks on the ex-dividend day in order to make investment decisions more accurately. An investor needs to know whether it makes more sense to purchase a share before or after the dividend is paid. Buying at the wrong time could lead to a significant decrease in return. Only a one percent decrease in annual returns can have a significant effect on a portfolio over a period of 20 to 40 years. This research report deals with the problems of share price behaviour on the ex-dividend day compared to the day prior to the dividend payment. The study investigates the price behaviour of German stocks from 2004 to Within the study the author differentiates between different samples. First, he analyses whether the share prices of the overall sample show a significant movement difference compared to the dividend paid. Second, he analyses whether the share price drop for large capitalized companies vary significantly from that of small capitalized companies. Third, he analyses whether the share price for different levels of gearing have an impact on the drop ratio on the ex-dividend day. Lastly, the author will differentiate between thirteen different industry sectors and analyse whether there is a difference in the share price behaviour. This section will be the largest analytical section in this report, and it will show the most interesting results. The study reveals that it is not easy to make assumptions about share price behaviour on the ex-dividend day. Furthermore, it shows some new insights, as well as gives additional possible research areas to the reader. III

4 TABLE OF CONTENT ABSTRACT... III LIST OF TABLES... VI LIST OF FIGURES... VII LIST OF ABBREVIATIONS... VII LIST OF ACRONYMS... VIII 1 INTRODUCTION RESEARCH AREA AND PROBLEM RESEARCH QUESTION AND SCOPE RESEARCH ASSUMPTIONS RESEARCH ETHICS LITERATURE REVIEW THEORETICAL FRAMEWORK Influence of the tax rate for the price behaviour Influence of debt to the share price on the ex-dividend day RELATED STUDIES The first study on stock price behaviour on the ex-dividend day Stock behaviour on the ex-dividend day without side effects Further studies on the influence of the tax rate Microstructure Ex-dividend day behaviour in a unique setting THE GERMAN STOCK MARKET IN CRITICISM CONCLUSION RESEARCH METHODOLOGY RESEARCH APPROACH AND STRATEGY RESEARCH DESIGN AND RESEARCH INSTRUMENTS Drop ratio Liquidity SAMPLING Price decline in detail IV

5 3.3.2 Market values Capital structure Different industries The statistical significance RESEARCH CRITERIA DATA COLLECTION LIMITATIONS OF THE STUDY DATA DATA SAMPLE Market capitalisation Capital structure Different industries RESULTS ANALYSIS OF PRICE DROP Price decline in detail Price decline on the basis of the market capitalisation Price decline capital structure Price decline industries Analysis transportation sector Analysis other sector Analysis electro engineering sector Analysis mechanical engineering sector CONCLUSION FUTURE RESEARCH DIRECTIONS AND OUTLOOK BIBLIOGRAPHY ANNEX I - DEFINITIONS ANNEX II DATA V

6 LIST OF TABLES TABLE 1: TOTAL STOCKS AND DIVIDENDS PAID PER YEAR 2004 TO TABLE 2: STATISTICAL SUMMARY OF TOTAL TABLE 3: MARKET CAPITALIZATION OF COMPANIES / LARGE TO SMALL TABLE 4: COMPANIES WITH HIGH AND LOW DEBT TO EQUITY RATIOS TABLE 5: OBSERVATIONS PER INDUSTRY SECTOR IN PERCENTAGE TABLE 6: CHARACTERISTICS OF INDIVIDUAL INDUSTRY SECTORS TABLE 7: STATISTICAL SUMMARY OF DROP RATIOS / TOTAL TABLE 8: ADJUSTED STATISTICAL SUMMARY OF DROP RATIOS / TOTAL TABLE 9: DROP RATIOS CATEGORIZED AS TOTALS AND PERCENTAGES TABLE 10: STATISTICAL SUMMARY OF SMALL TO LARGE CAPITALIZED COMPANIES TABLE 11: STATISTICAL SUMMARY OF DEBT TO EQUITY RATIOS 2004 TO TABLE 12: SUMMARY OF DROP RATIOS FOR SMALL CAP AND LARGE CAP COMPANIES TABLE 13: T-VALUE / CRITICAL VALUE FOR CAPITAL STRUCTURE TABLE 14: STATISTICAL SUMMARY OF DROP RATIOS FOR DIFFERENT INDUSTRY SECTORS TABLE 15: LEVENE TEST STATISTIC / INDUSTRY SECTORS TABLE 16: MEDIAN ANALYSIS / INDUSTRY SECTORS TABLE 17: MEDIANS ABOVE/BELOW MEAN / INDUSTRY SECTORS TABLE 18: TWO WAY FREQUENCY CHI-SQUARE TEST STATISTIC / DIFFERENT INDUSTRIES TABLE 19: STATISTICAL SUMMARY DROP RATIOS FOR TRANSPORT SECTOR TABLE 20: STATISTICAL SUMMARY DROP RATIOS FOR OTHER SECTOR TABLE 21: STATISTICAL SUMMARY DROP RATIOS FOR ELECTRICAL ENGINEERING SECTOR VI

7 TABLE 22: STATISTICAL SUMMARY DROP RATIOS FOR MECHANICAL ENGINEERING SECTOR TABLE 23: T-VALUE / CRITICAL VALUE COMPARISON MECHANICAL ENGINEERING SECTOR TABLE 24: SUMMARY OF DROP RATIOS IN PERCENTAGE / MECHANICAL ENGINEERING SECTOR TABLE 25: SUMMMARY OF DROP RATIOS IN PERCENTAGE /OVERALL SAMPLE SECTOR LIST OF FIGURES FIGURE 1: COMPANIES ALLOCATED INTO DIFFERENT INDUSTRY SECTORS FIGURE 2: INDIVIDUAL DROP RATIOS CATEGORIZED IN STEPS OF ONE FIGURE 3: DISTRIBUTION INDIVIDUAL DROP RATIOS CATEGORIZED IN STEPS OF FIGURE 4: CATEGORY FROM ZERO TO ONE IN PERCENTAGES 2004 TO FIGURE 5: VARIATION OF MEAN DROP RATIOS IN DIFFERENT INDUSTRY SECTORS VII

8 LIST OF ABBREVIATIONS / ACRONYMS AG DAX MDAX SDAX TECDAX XETRA Aktiengesellschaft German stock index Mid-Cap-DAX Small-Cap-DAX Technology-DAX Exchange Electronic Trading VIII

9 1 INTRODUCTION 1.1 Research Area and Problem The subject of the decline in price has been a very interesting question for researchers for many years. In 1955 Campbell and Beranek studied the behaviour of shares traded at the New York Stock Exchange on the ex-dividend day. They showed that a price decline of only 90 percent of the dividend paid occurred, and they traced this phenomenon back to tax effects. In 1970 one of the most famous studies was published by Elton and Gruber. While Campbell and Beranek only assumed that the price decline is due to taxation reasons, Elton and Gruber analysed the influence of the tax effect. Elton and Gruber were the first researchers to find a plausible explanation for the ex-dividend day price behaviour. Like their successors, Elton and Gruber used the New York Stock Exchange for their study. In their sample size of 4,148 observations, the stock price dropped on average by 78 percent of the value of the dividend. The second part of their research concentrates on a model, created by the researchers, providing evidence that the tax effect explains the difference in the price drop from 78 percent to 100 percent on the ex-dividend day. In the years that followed, researchers have concentrated their studies on the tax effect on the ex-dividend day, and research confirms Elton and Gruber s results from Only in 1982, more than 10 years after Elton and Gruber s study, did the first critiques come along. Avner Kalay wrote an article in 1982 querying that tax issues alone should be responsible for the difference of dividend paid to share price decline. Even though Kalay admitted that a fraction of the decline could be accounted for by the tax effect, he argued that there must be further ascendancies, like arbitrageurs, involved in ex-dividend day price behaviour. Miller and Scholes in 1982 and Boyd and Jagannathan in 1994 further support Kalay s thesis. In 1998 Frank and Jagannathan studied the ex-dividend price behaviour at the Hong-Kong Stock Exchange, where neither dividends nor capital gains are taxed. Their study also shows that the share price on the ex-dividend day drops less than the dividends paid even without taxes being involved. In the case of the Hong-Kong Stock Exchange, the price-decline was only 43 percent of the dividend paid on the ex-dividend day. Frank and Jagannathan allocate this price decline to the microstructures of the stock market in Hong-Kong. 1-1

10 The authors French, Varson and Moon (2005) found a new explanation as they investigated the companies of the American stock market. In contrast to their predecessors, they did not condemn previous research but rather confirmed their influencing factors, like tax effects, microstructure of the market, or the presence of arbitrageurs. Furthermore, in their study they analyzed the ex-dividend price behaviour for companies with long term financial debt in comparison to the price behaviour of companies without long term debt. So what is the research problem and why is it important for the German market? The total gross monetary assets of the German citizens in the year 2004 were calculated to be 4,068 billion euro. This amount has doubled from the beginning of the nineties. Problematic is that the annual growth rate of only 5.5 percent over the past 14 years was considerably lower than the rate in previous years. In the nineties 80 percent of that growth came from savings accounts while only 20 percent of the growth came from capital gains. After taking the dot.com bubble into account, the percentage from capital gains drops to approximately 14 percent. Comparing these numbers to the United States of America, one will notice that the growth in gross monetary assets in America is mainly based on capital gains rather than on savings. Within the past same time period from the beginning of the nineties until 2004, the annual growth rate was 7 percent. About 66 percent of that growth came from capital gains, while only 34 percent came from savings. This is a return advancement of 1.5 percent annually. The main difference between Germany and America lies in the pure share portion in the individual portfolios. While the share portion in American portfolios average about 22 percent, the share portion in German portfolios only accounts for 6.5 percent. On the other hand, the pure savings account portion in America averages 20 percent, while it averages 40 percent in Germany. By understanding these numbers, one can easily estimate that a small change in German portfolio management can have a large impact on the overall wealth of the country. Many stock market and dividend behaviour studies were done on the American stock exchange. This study deals with the German stock exchange and how the German stock prices behave on the ex-dividend day in comparison to the dividend amount being paid. The fundamental aim of this study is to find further explanations of the behaviour other than just taxation issues. Since the German investors are very conservative, dividends could attract them. This study analyses, whether there is a significant difference in the share price movement to the dividend being paid. In order to achieve the goal, the 1-2

11 author will sample four major German stock exchanges over the period 2004 to Furthermore, he will analyze the sample more deeply, by splitting the sample into fragments. This will allow the author to make further conclusions and recommendations on what kind of companies or industries are preferable when it comes to buying their shares one day prior to the ex-dividend day and selling on the ex-dividend day. Just a small increase in return or a bit more knowledge about price behaviour of German stocks might draw more German citizens into the market which can lead, if done correctly, to a significant increase of wealth for the country. A one-time deposit of 100,000 euro at an annual return of 3.6 percent would give 170,000 euro after 15 years; the same amount at only 2.1 percent would lead to only 137,000 euro rounded to the nearest thousand. The chosen research is quantitative, while the purpose of this research paper is explanatory. Neither questionnaires nor interviews are needed. The only data collected will be observations further explained in the following chapters. 1.2 Research Question and Scope This extract of previous studies shows that there is no consensus on how to explain the exdividend day stock price behaviour. There are many different theories, and every theory has its supporters, but none of the studies is fully without doubt. All studies have one fact in common: they all find a price drop occurring, and this price drop of a share on the ex-dividend day is less than the dividend paid. The purpose of this paper is therefore not that of reappraising the previous studies, but rather to analyse the price decline on the German stock market more precisely. In order to do so, the author will analyse the price decline based on a number of different criteria. Firstly, the author will check if the price drop at the German stock market is significant. The author could not find a single study that has dealt with the German stock market. His results will then be compared to previous study results mentioned above. Secondly, previous research shows that the price decline in proportion to the dividend is between zero and one, but the distribution of the price ratios has not been adequately analyzed by any researcher. One aim of this paper is, along with the price drop analysis, to analyze the 1-3

12 distribution of the drop ratios and to show how many companies do not fall into the range of zero to one. In considering the distribution of the drop ratios, the author will look more closely at the arithmetic mean. The aim of this approach is to show, alongside the average decrease of the share price on the ex-dividend day, the allotment of shares that react differently from previous studies. The study by French, et al. (2005) is then taken and applied to the German market. This study was chosen because it is a fairly new study and has not been tested in many different variations and markets. With this research paper, the author wants to research this approach further to see if the German market reacts similarly. This will allow following researchers to take the French et al. study, together with this study on the German market, and apply it to other markets, like emerging economy markets in South Africa, Brazil, or China. The author tries to analyze more deeply into the area of stock behaviour on the ex-dividend day, because all previous explanations are unclear and not fully satisfactory. Some researchers have provided evidence that hypotheses are valid, but other researchers have shown that these hypotheses applied to a different market would be rejected. Lastly, the author will categorize the companies by their market value and test whether there are differences in the ex-dividend day behaviour of small market capitalized companies in comparison to large market capitalized companies. In order to measure the market capitalization, the author will multiply the number of shares by the share price before the ex-dividend day and then group all companies into three categories: large capitalized, medium capitalized and small capitalized companies. This approach, hand in hand with the approach by French, et al., will allow stock brokers to understand the price behaviour of shares on the ex-dividend day better. The risk profile of a company is always taken into consideration before investing, so therefore there should be a difference in ex-dividend day behaviour of stocks with different capital structures. The author intends to answer the following questions: 1. Is there a general difference in the stock price drop compared to the dividend paid? 2. Is this drop similar for small capitalized and large capitalized companies? 3. Is this drop similar for highly geared and less geared companies? 4. And lastly is this drop similar throughout all industries and sectors in the market? 1-4

13 1.3 Research Assumptions The author will look exclusively at the German stock market from 2004 until These years include years of economic growth as it happened from 2004 until 2007, and also years of economic downturn as occurring from 2008 until Even though the author looks at the years individually, it is assumed that there is no significant difference in the stock price behaviour on the German stock exchange due to different economic cycles. A second important factor to mention is the fact that Germany has passed a new law on dividend and capital gain taxation in 2008, which came into effect on the This law makes the taxation on dividends and capital gains very easy, by taking 25 percent taxation on every dividend payment received, and 25 percent taxation on any capital gain made. Formerly, dividends and capital gains were taxed with individual tax rates depending on your yearly income. Furthermore, capital gains on stocks that the shareholder was able to hold for more than one year were totally tax free. Even though Callaghan and Berry (2003) described how complex the American tax system is, the author will not try to compare the impact of the German tax system to the impact of the American tax system. Former researchers have shown that a taxation effect can be seen in comparing ex-dividend day stock price behaviours. The author of this paper therefore does not give this fact special consideration. Furthermore, the author assumes that by choosing the German DAX, the German MDAX, the German SDAX as well as the German TECDAX, the author has got the best possible sample out of the whole population of German stocks. The largest problem the author might have is the fact that different stock prices are announced for different stock exchanges in the German market. Even though these stock prices vary only marginally, the author assumes that using only the Xetra stock prices as explained in chapter 4.1, will lead to the best possible outcome of the study. It is also assumed that the stock prices from the other stock exchanges in the market will not give different overall results due to the margin difference in stock prices. 1.4 Research Ethics The researcher does not see any difficulties with ethical clearance. The ethical clearance form has been submitted. 1-5

14 2 LITERATURE REVIEW 2.1 Theoretical framework Influence of the tax rate for the price behaviour In 1970, nine years after the study by Miller and Modigliani (1961), the authors Elton and Gruber published an article in The Review of Economics and Statistics. Elton and Gruber used Miller and Modigliani s study from 1961 and built on their findings. The central theme of the research was not the price drop of a share on the ex-dividend day itself, but rather the reason why the share price did not drop for the total amount of dividend paid. Additionally, they looked at the influence of the marginal tax rate for the price drop. Elton and Gruber expected that an investor would be indifferent to buying and selling a particular share. Therefore they form the following condition of equilibrium: ( )= ( )+ +( ) (1) 1. the cost of a share 2. the price of a share the day before the stock goes ex-dividend 3. the price of the stock the day the stock goes ex-dividend 4. the capital gains tax rate 5. the tax rate on ordinary income The thought behind the first equation is the fact that a shareholder s income, if selling a share before the ex-dividend day, is similar to the price ( ) minus the capital gain tax for the period of holding the share ( ( ). If, on the other hand, the shares will be sold on the exdividend day, the capital gain will be calculated as the dividend minus income tax (D (1 ) plus the profit after taxation for the stock ( ( ). In order for an investor to be unaffected, the profit of the first selling option pre-dividend must be similar to the profit for the second selling option ex-dividend payment. After rearranging the formulas, the new equation looks as follows: ( ) / D = (1 ) / (1 ) (2) 2-6

15 If the new equation is in equilibrium, the left side would then represent the decrease in the stock price in proportion to the dividend paid. A stockholder with his/her individual income taxation rate would therefore be unaffected by selling the share on the pre dividend day or on the ex-dividend day. Elton and Gruber argue that the left part of the equation should be equal to one, as long as the taxation is not an issue in determining the stock price on the ex-dividend day. Otherwise one can also derive the marginal capital gains taxation as well as the income taxation from the equation. Elton and Gruber in 1970 were the first researchers to prove, on the basis of testing the price decline in proportion to the dividend paid, that the price drop on the ex-dividend day is smaller than one but still statistically significant. These findings were new insights for the topic, and for the first time were proven. In their research of all shares of the New York Stock Exchange from the 1 st of April 1966 until the 31 st of March 1967, they discovered an average price decline of the share price at the exdividend day of 78 percent of the dividend amount paid. The probability that the share price would drop at exactly the amount of the dividend paid was calculated to be 1.3 percent. In the second part of their study, Elton and Gruber concentrated their research on confirming Miller and Modigliani s study (1961) that different tax classes for capital gains and dividends would lead to different investor groups. These different investor groups were, according to Elton and Gruber, also responsible for the fact that the share price at the ex-dividend day did not fall for the same rate as the dividend paid on that day. In the years following, the majority of research concentrated on the impact of taxation on the price behaviour of a share on the ex-dividend day. Thereby, the findings by Elton and Gruber were mostly reconfirmed. The first critique of the theory came from Avner Kalay in 1982, more than ten years after Elton and Gruber s study in Kalay (1982) published the article The exdividend day behaviour of stock prices: Re-examination of the clientele effect in The Journal of Finance. Kalay extended the study by Elton and Gruber (1970) by adding transaction costs, arbitrageurs and long time investors. In his elaboration, Kalay explicitly attacks Elton s and Gruber s work. He implies that Elton and Gruber s results were skewed due to the fact that they left out various aspects in their study. Despite all the criticism, Kalay concurs with the findings of Elton and Gruber that one can observe a taxation effect at the ex-dividend day. In 2003 Elton and Gruber, in collaboration with Blake, researched the topic again. This time 2-7

16 the researchers looked into the field of tax exempt and normally taxed investment funds. As already proven in their previous study, the researchers came to the conclusion that the main driver of the share price drop in proportion to the dividend paid can be explained by taxation effects. They refute Kalay s (1982), Bali s and Hite s (1998), and Frank s and Jagannathan s (1998) studies that the microstructure of the market could be responsible for the price decline. In comparison to their previous study 32 years prior to 2003, they do not ascribe the total price drop on the ex-dividend day to tax effects, but rather close their study with the words: Thirty-two years after our original study, we find new and compelling evidence that taxes play an important part in affecting share price changes (page 18). Herewith they left enough space for different scientific explanations Influence of debt to the share price on the ex-dividend day A new and very interesting approach to explain the ex-dividend day stock price behaviour came from French, Varson and Moon (2005). In their work called Capital structure and the exdividend day return published in The Financial Review, the three authors acknowledge the influence of taxes, arbitrageurs and the microstructure of the market to be valid reasons why the decline in the stock price on the ex-dividend day is less than the value of the dividend paid. The authors assure us that these approaches can only explain a fraction of the price drop. French, et al believed that companies with different capital structures react differently on the ex-dividend day. They hypothesized that a company with debt in their capital structure will have a smaller price decrease on the ex-dividend day than a company where there is no debt in the capital structure. In the first theoretic part of their research, the authors calculate the stock price with the help of a modified model created by Black and Scholes to price options from Based on this valuation method, French et al came to the conclusion that the less debt a company has in its capital structure, the closer the slump in prices converges to the value of the dividend. The reasoning behind this approach is the fact that an investor owns a portfolio of risk attached to equity and debt in a company. On the day of the dividend payment, this portfolio also includes risk free capital in the form of the dividend just paid. The risk between the portfolio before the dividend payout day and after the dividend payout day needs to be similar, and therefore the dividend at the ex-dividend day increases the risk of the portfolio. If the increase of the risk can be absorbed by the equity in the portfolio, then the equity will decrease by exactly the same amount of the dividend. If, on the other 2-8

17 hand, debt is also existent in the company, then the dividend paid will be out of reach for the debt holders of a particular company, and therefore the debt risk increases, which leads to a price drop of the share. The part of equity proportional to the total risk has decreased, and therefore the value of equity decreased by less than the value of the dividend, due to the fact that a part of the price drop can be absorbed by the value of the debt. In the second part of the study by French et al, the authors analyzed the volatility of all stocks that traded on the American Stock Exchange as well as all stocks traded on the New York Stock Exchange between 1989 and 1996 that have paid a cash dividend. The price decline in proportion to the dividend (ex-dividend drop ratio) was calculated with the following formula by French et al (2005 pg. 370): DROP = ( ) / D (3) = Cum-dividend price, closing price of the stock on the prior day = Ex-dividend price, closing price of the stock on the ex-dividend day D = Dividend paid Based on the DROP formula, the ex-dividend day return for the share can be calculated by using the following formula: = ( + D) / (4) The equation to calculate the ex-dividend drop ratio, as well as the equation to calculate the ex-dividend day return, will be calculated for both companies with long term debt and companies without long term debt. The null hypothesis for the drop ratio is equal to: : DROP = 1.0 (5) The null hypothesis for the ex-dividend day return is equal to: : = 0 (6) The stock price of companies without long term debt drops by an average of times of the dividend payment. The authors found that the stock price decline is statistically not significantly different from 1.0 and therefore the hypothesis H : DROP = 1.0 cannot be rejected, but rather needs to be accepted. The average return for the same sample is percent, which is statistically not 2-9

18 significantly different from the hypothesis H : Rx = 0 and can therefore not be rejected either. The stock prices of companies with long term debt, on the other hand, drop by an average of times of the dividend payment. The authors tested whether the stock price decline was significant at the 0.01 level, and found that it was significantly different. The null hypothesis was rejected and the alternative accepted. The average return at the ex-dividend day was and the authors found this number to be significant at the 0.01 level. French et al conclude their study with the cognisance that the stock prices do not drop by the exact same amount of the value of the dividend paid. Companies that have debt in their financial structure will have a drop in their stock price that is lower than the share price drop of companies without debt in their financial structure. 2.2 Related studies The first study on stock price behaviour on the ex-dividend day One of the first studies ever to research the subject of ex-dividend day price behaviour was the study Stock price behaviour on ex-dividend days by Campbell and Beranek (1955). In two different studies of the New York Stock Exchange, they found that the stock price on the exdividend day drops on average by 90 percent of the value of the dividend paid. They also found that the price decline varies greatly. They concluded that a tax effect must influence the share prices, but do not investigate the subject further Stock behaviour on the ex-dividend day without side effects In the year 1961, more than six years after the study by Campbell and Beranek in 1955, the Journal of Business published an article Dividend, growth and the valuation of shares by Miller and Modigliani. Both authors proved that in a capital market environment without any side effects of financing, without taxation, flotation costs, information costs, transactional costs and bankruptcy costs as well as with a given investment policy, the dividend policy of a company is practically irrelevant. The authors spin the hypothesis further by arguing that the irrelevance of a dividend policy is still existent in an imperfect market environment. This is due to the reason that effects of dividend policy will be neutralized by different categories of investment classes. By looking at investors with different taxation rates for capital gains and for income, an 2-10

19 investor with a high marginal income tax would rather invest in a company paying out only a small fraction of their profits as dividend. This would give the investor the chance of realizing more capital gains via a rising stock price in the future. These capital gains would be taxed lower than dividends paid, and therefore preferred by investors. Investors with a comparable small marginal income taxation rate would, on the other hand, rather collect dividends. This phenomenon would lead to different investor classes, investing only in companies, with a dividend policy that fits their taxation preferences Further studies on the influence of the tax rate Brown and Clarke (1993) studied the ex-dividend behaviour of Australian shares before and after the dividend imputation. They were able to research this area because the Australian government had reformed the taxation payments in The first major change happened in 1985, when the government introduced a new law that allowed the taxman to tax all capital gains on all assets which were bought after the 19th of September of a particular year. The second and third change occurred on the basis of dividend imputation in 1987 and Brown and Clarke researched the area from July 1973 to June 1991, including the years 1985 until 1988, when there were major changes in the taxation laws in Australia. Their aim was to provide evidence that the Australian stock exchange on the ex-dividend day reacted statistically significantly differently in 1991 from what it did in Brown and Clarke find that the drop ratio was significantly less than one in Australia over the period from 1973 to 1984; while the drop ratio increased slightly after 1984, it was still significantly less than one. The paper ended with the following conclusion: In sum, the tax laws are not the whole of the explanation for the ex-dividend day trade-off between dividends and capital gains. The Australian share market, on average, still preferred returns in the form of capital gains over dividends, despite successive tax changes which increased the attractiveness to Australian shareholders of dividends relative to capital gains. The author Lasfer, in his study from 2007, analyzed the tax systems and tax reforms in four major European countries. These countries included the United Kingdom, France, Italy and Germany. Lasfer studied the impact of different tax systems and tax reforms on the ex-dividend day behaviour in each particular country. His most interesting hypothesis is the following: Ex-day returns will be highest in the country where tax differentials are the lowest. Since the differential 2-11

20 between taxes on dividends and taxes on capital gains in France is highest, followed by Italy, the United Kingdom and Germany, he expected to find a significant difference between France and Germany. He found that the return on the ex-dividend day is highest in France, with a value of 1.29 percent, and lowest in Germany, with a value of only 0.57 percent. Furthermore, Lasfer provided evidence that the drop ratio in the United Kingdom was compared to in Germany, in Italy and in France. These findings provided some evidence that the taxation effect is significant, especially in France, the United Kingdom and Italy; however, the author acknowledges that there must be further reasons other than just the tax differentials on capital gains and on dividends responsible for the price drop on the ex-dividend day. For me, this research was very interesting and I would have loved to further investigate, but the German tax laws were so complicated in the past that in my opinion there is not much sense in repeating a study about tax differentials. Today the tax differential between capital gains and dividends in Germany is zero, due to a new law which came into effect at the first of January Microstructure Two major articles were published that put microstructure arguments forward in order to explain the price drop of shares at the ex-dividend day. The first article was written by Bali and Hite in 1998 and the second was written in the same year by Frank and Jagannathan. Bali and Hite (1998) argue that there is a price drop less than the dividend paid, but they give the explanation that this phenomenon is due to the discreteness in prices rather than being tax related. The share price is discrete and this has wrongly been allotted to taxation effects. Frank and Jagannathan (1998) have researched this topic at the Hong-Kong Stock Exchange, where neither dividends nor capital gains were taxed. They accept Elton and Gruber s findings from 1970, but revert to the fact that it is difficult to prove their findings due to a very complex American taxation system. In order to avoid a complex tax system, Frank and Jaganathan chose the market of Hong-Kong. They hypothesize that the collection and reinvestment of dividends is bothersome for individual investors but not for market makers. In absence of taxation, which was central for previous studies, both researchers came to the same conclusion that the stock price drop at the ex-dividend day is less than the amount of dividend paid. They develop a model to illustrate their findings, in which the ask price as well as the bid 2-12

21 price play central roles. On the basis of this model they were able to show that a rational investor who wants to buy a share always prefers the ex-dividend day to purchase the share. In comparison to that point, a seller declines to sell his shares at the last day before the dividend is paid. As a consequence of these preferences and of the difference between the ask price and the bid price, the stock price should logically rise on the day the dividend is paid, autonomously of the value of the dividend paid. The reason for a price decline is the so called market maker, who intervenes in the market transactions. Market makers tend to buy a share the day before the dividend is paid and then sell the share on the ex-dividend day. Frank and Jaganathan argue that this so called bid/ask bounce is responsible for the drop in the share price at the ex-dividend day being less than the amount of dividend paid, whereas others misinterpret this phenomenon as a tax effect. Market makers are defined by the authors as intelligent traders who always work with limits on their trades, while the remaining traders always sell/buy for the current market value. The market makers are the only traders that one can call rational investors, because they buy shares one day before the dividend day and sell the shares on the dividend day, while it is not relevant for other investors when a transaction is fulfilled. These findings also support the views by Liano, Hardin and Huang (2003) who did not believe in the impact of taxation as the only major driver for the drop ratio on the ex-dividend day Ex-dividend day behaviour in a unique setting The study by Al-Yahyaee, Pham and Walter (2008) has major advantages compared to previously described studies. They study the drop ratio at the ex-dividend day at the Oman stock exchange. Firstly, it is important to know that there are no taxes on capital gains or on dividends paid in Oman. Secondly, it is important to know that dividends are paid annually, not quarterly like in the United States. Thirdly, stock prices in Oman are displayed decimalized instead of non decimalized. 1 Kadapakkam (2000) argued that the non decimalized stock prices in the United States were responsible for troubles in evaluating the magnitude of ex-dividend day drop ratios in comparison to the quarterly dividend. Both points could be avoided in Oman, due to decimalized prices and annual dividend payments. Lastly the researchers have data in order to analyze the so 1 The minimum tick in the United States used to be 1/8 th of a Dollar 2-13

22 called bid-ask bounce effects at the Oman stock exchange. This phenomenon was previously studied by Frank and Jagannathan (1998) at the Hong Kong stock exchange. In order to visualize the formulas used, the author will compare the Al-Yahyaee et al assumptions with previous equations by Elton and Gruber (1970): ( ) / D = (1 ) / (1 ) (2) Since taxes on capital gains as well as taxes on dividends are zero in Oman, the equation by Elton and Gruber (1970) can be rearranged to: ( ) / D = 1 (7) from The second equation introduced by Elton and Gruber (equation 1) can also be rearranged ( ) = ( ) + +( ) (1) to the following equation due to the reasons mentioned above; no capital gain taxation and no dividend taxation in Oman: ( + ) / = 0 / = 0 (8) Equation seven means that the authors expected the price drop of the stocks in Oman having exactly the same value as the dividend paid, while equation eight means that the authors further expected the ex-dividend day return in Oman to be zero. The advantages of this study mean that a drop ratio of less than one cannot be ascribed to taxation, price discreteness or transaction costs. Al-Yahyaee et al s findings were firstly that the stock price drops significantly less than the value of the dividend paid, and secondly that transactional costs are not significant. This study gives significant support to previous studies like Kalay (1982) and Frank and Jagannathan (1998), who question the significance of taxation effects as the key driver for the ex-dividend day price behaviour of stocks. Al-Yahyaee et al s findings on the bid/bounce effect reveal that there is no significant difference from zero. Lastly, the study ends with the words In sum, the results indicate that market microstructure influences the ex-dividend 2-14

23 day premium and ex-day return. Once market microstructure effects are taken into account, dividends and capital gains are valued equally at the margin. 2.3 The German stock market in 2009 The German Handelsblatt newspaper article Dividenden sprudeln trotz Finanzkrise in 2008 stated that the 30 companies at the German DAX will pay a total of 24 billion euro in dividends for the year 2009 to their shareholders, after having paid a record of 27.5 billion euro in dividends for the financial year Even though the world economy is in a recession, the amount of 24 billion euro is just about 10 percent less that the total dividend payments of the 30 DAX companies in the previous year, Unlike the United States, where dividends are paid quarterly, the dividends in Germany are paid annually. This fact results in higher dividend payouts with less transaction costs compared to countries where dividends are paid quarterly. The dividends in Germany are usually paid in late April until the end of May. Since the German DAX was introduced 20 years ago, almost 40 percent of its increase is due to dividend payments. Due to the financial crisis and the following stock market crash, dividend yields in Germany are at a record high for the German DAX, averaging 5 percent compared to a long-standing average of 2.7 percent. Lastly, it is important to know that stock prices in Germany are traded on decimals instead of non decimals, like the United States used to do. Bankhaus-Lampe, one of Germanys leading private banks, wrote in 2009 that the companies for the smaller MDAX paid 35 percent less dividends in 2009 than in the previous year, The total amounted to billion euro, significantly lower than the total dividend payments for the DAX in Small and medium sized companies prefer, in a crisis, rather to strengthen their capital basis before paying out unnecessarily high dividends. 2.4 Criticism The first study by Campbell and Beranek (1955) is more than 50 years old. This does not mean it is not valid, but it means that their results come from a time where the entire information network and the trading platforms were not fully sophisticated. Miller and Modigliani s study (1961) is just six years younger. They argue that in a capital market environment without side 2-15

24 effects, dividend policies are irrelevant. For the German market today, this is not true anymore. Every investor is taxed at 25 percent, so investors with high marginal income tax who rather invest in companies paying out small dividends do not exist anymore. Capital gains as well as dividends are both taxed at 25 percent since 1 st January The study by Brown and Clarke (1993) is a study that could be repeated in the German market in the year 2013, as mentioned in the future outlook section of this report in chapter 7. The approach by Lasfer (2007) was a very interesting one. He analysed the tax differentials in four major European countries. As mentioned in section 2.2.3, the German taxation laws are extremely complex, and measuring the influence of the tax legislation on the stock price behaviour in Germany seems to be very difficult. There used to be hundreds and thousands of exceptions to the rule on how to tax a person, so that the researcher cannot state a generalized assumption on the tax influence for the German market. Finally, the author would like to criticise the study by Frank and Jagannathan (1998). Both have researched the topic at the Hong-Kong stock exchange, where neither dividends nor capital gains were taxed. This fact is not true for foreign investors, and I doubt that the influence of foreign investors to the Hong-Kong stock exchange is marginal. I would even argue that with today s technologies every investor worldwide can invest in whichever market he wants to invest in, but he or she is still taxed with his personal income tax from his country of residence. 2.5 Conclusion This chapter gave an overview of the existing theories in the field of ex-dividend day price behaviour of stocks. The main findings were that the researchers have not found a conclusive way of explaining the causation for the price decline at the ex-dividend day. Every theory has its supporters, but on the other hand every theory has at least the same amount of researchers who do not agree with its findings. The study by French et al. (2005) is an exception. It is the first study that denies that the final outcomes were the only ones responsible for the price drop, but rather also allows other studies, results and explanations to be valid. Furthermore, the study by French et al. brings a totally new approach to the existing models and literature without just applying an existing model to a new stock market. The studies by Elton and Gruber (1970) and the study by Kalay (1982), as well as the study by Frank and Jagannathan (1998), have been reproduced many times. 2-16

25 The following study has the aim of not reproducing one of the frequently reproduced studies, but is rather focussed on the study by French et al, and applies to the German market in order to see if the effects are similar. 3-17

26 3 RESEARCH METHODOLOGY 3.1 Research Approach and Strategy In the following study the author analyzes whether the price drop at the ex-dividend day at the German stock exchange is statistically significant. Furthermore, the author analyzes whether the results by French et al. (2005) can be applied to the German stock exchange, and whether it can observe different stock price behaviours at the ex-dividend day between companies with debt and companies without debt. If the author can prove that the price decline at the ex-dividend day is unequal to the value of the dividend paid, then he will also analyze the companies on the following principles: the years, market capitalization of the individual companies, capital structure of the companies and different branches/industries the companies operate in. This separation will show whether there are differences between years, different sizes of companies, different capital structures of the companies and in the industries a company operates in compared to another industry. The author will not look at the influence of taxation on the price behaviour due to previous published studies that have all come to the same conclusion: that taxes are one relevant factor. The author fully appreciates these findings but does not want to analyze them again and therefore is not going to account for the taxes in this study any further. The author will start as a first step to calculate the proportion of the price decline to the dividend payment. This will only be possible for the companies that pay out a dividend, so the sample needs to be adjusted in order to only analyze companies that actually pay out dividends. In the second step the author will adjust the calculated drop ratios by removing all thinly traded companies. The new sample will then be compared to the old sample, and statistical tests will be applied. Following the descriptive statistic and the hypothesis tests, the behaviour of the prices will be analyzed in more detail. By means of the categorization explained, the drop ratios over the whole period of the years 2005 until 2009 will be compared and further analyzed. Additional to the analysis of the total sample, the author will also analyze each year separately in order to find differences within the research period. Intuitively, one can assume that companies with a high market capitalization will more likely behave in the same way as previous studies have researched. The price drop of companies with a high market capitalization should in the majority be less than the dividend paid, while companies with low market capitalization should rather show 3-18

27 abnormalities. As a basis for previous studies, the authors always used large stock exchanges, for example the New York Stock Exchange or the Hong-Kong Stock Exchange. At these stock markets low liquidity is not a big issue, due to the size of the markets. This paper deals with the four major stock exchanges of the German market, namely the German DAX, the SDAX, the MDAX and the TECDAX. Especially the SDAX is focused on small companies, where the phenomenon of thin trading can occur. The author chose all four German markets in order to see if there are any differences in the ex-dividend day behaviour of small companies compared to large companies. In order to see differences, the author will divide the companies into different categories based on their market capitalization. The third step is the comparison of the study by French et al. to the German market. The procedural method is the same one as in the primary study. Lastly, the author will analyze the drop ratio within different industries. Herein different industries will be compared according to the number of companies per industry and by the market capitalization per industry. Following this approach the author will, as explained, compare the results for each industry with the value for the whole sample. The larger industries will also be tested: whether the drop ratio is significantly different from the drop ratio of the total sample. 3.2 Research Design and Research Instruments Drop ratio In a first step, the price difference in proportion to the dividend needs to be calculated. This will be in accordance to the studies by Elton and Gruber, Kalay and French et al., by using the following drop ratio formula: DROP = ( ) / D (9) = Cum-dividend price, closing price of the stock on the prior day = Ex-dividend price, closing price of the stock on the ex-dividend day D = Dividend paid 3-19