In Chapter Two the theoretical framework that surrounds the thesis is given. This will give the reader the results from previous studies and the knowledge needed to understand the method used and the results of the thesis.
Dividend Policy and the Demand for Dividends
The dividend policies of companies have been discussed in many academic papers during several decades. Modigliani and Miller (1961) suggested that a company’s dividend payout policy does not matter in a perfect market. Hence, it will not affect the value of the company’s shares. The important part here being that it applies under perfect market conditions. Textbooks in corporate finance still refer to the Modigliani and Miller (1961) dividend irrelevance theorem. However, since the early 1960’s the view of how the payout policy influences a company´s value has shifted. Still keeping some of the original assumptions of a perfect market, DeAngelo and DeAngelo (2006) states that the payout policy of a company will affect the company’s value. Either way, one can confirm by looking through the literature that there exist investors with a preference for dividend, both retail and institutional investors (Dong, Robinson & Veld 2005, Graham & Kumar 2006). In theory the possibility of recreating cash dividend by selling stock is often mentioned, Shefrin & Stateman (1984) argues the contrary, meaning that selling stocks should not be seen as a perfect substitute for receiving dividend. They also argue that, in some cases, there exists a willingness to pay a premium on the share price to be able to receive cash dividends. The reason for this, seen as quite odd behavior to some, is to minimize the risk of feeling regret if the sold share will increase in value soon after the investor has sold it.
Dogs of the Dow Studies
The literature surrounding dividend investing is extensive and there are several papers that are dedicated to the DoD strategy specifically. To get an overview of the existing literature a review of the articles was conducted. Articles were gathered by a thorough search of databases such as Primo with the key words; Dogs of the Dow and dividend investing. Some articles were found by being referred to in newer articles. The articles are divided into two sections, first are articles that have applied the DoD strategy but which methods are not explained thoroughly in the article itself. Thereafter, articles, with methods, that are more deeply explained and, that possibly are of use for this thesis are presented. The results from all articles that have applied the DoD strategy are summarized in Table 1 at the end of the section.
As briefly mentioned above, the DoD strategy seems to first appear in the Wall Street Journal in 1988. In a column written by John Slatter, at that time working as an analyst, evidence was presented that investing in the ten highest dividend yielding stocks of the DJIA would give a return of 18 percent compared to the market return of 11 percent. The return was the annual average during the sample period of 1972-1987 (Filbeck & Visscher, 1997). The strategy was further made popular when a couple of books were published confirming the, at that time, unquestionable success of the DoD strategy (see O’Higgins & Downes, 1991 and Knowles & Petty, 1992). In the books, the original DoD strategy was applied, just as before, on the DJIA. The periods investigated in both of the books, though they go back a bit further than previously, are similar to the period first tested by Slatter. This could be one obvious reason as to why all of the studies confirm the success of the strategy.
After that the DoD strategy itself was tested, researchers started to look for reasons for its success. In a study conducted by Domian, Louton and Mossman (1998) it was hypothesized that the performance of the DoD strategy can be linked to a ‘winner-loser effect’. In their study, Domian et al. (1998) formed two portfolios, one with the high yield stocks and one with low yield, and for their benchmark, they used the S&P 500 as the market portfolio. Then they tested the performance of the portfolios twelve months before and after the portfolios where set up. The results implied that stocks included in the high dividend portfolio had been underperforming before the portfolio was set up, and therefore had a high dividend yield. Hence, the high return can be traced back to a normalization of the stock price by the market. The opposite was true for the low dividend portfolio. The stock in that portfolio had over performed compared to the market prior to the portfolio construction.
Hirschey (2000) presents several reasons that he argues can explain the excess return found in previous studies. He points towards data mining and errors in the choice of test periods. In addition he states that the strategy’s easy to understand attribute and the way it has been marketed are reasons for its popularity. Hirschey (2000) argues that there does not exist any real evidence proving that the DoD strategy does create excess return compared to the DJIA. Da Silva (2001) conducted a study of the DoD strategy on the Latin American markets of Argentina, Brazil, Chile, Colombia, Mexico, Peru and Venezuela The study looked at 97 percent of the total Latin American and Caribbean stock markets, measured in market capitalization. Da Silva (2001) tested several versions of the DoD strategy, these were; top ten, top five, top one, and the second highest dividend yielding stock on its own. The strategy was also tested with different starting days, this to see if seasonal factors impact the performance. The portfolios were started on the first trading day of January, in accordance with the original strategy. An alternative starting day was set to the first trading day in July due to it being as far away from January as possible. To measure the risk adjusted return of the portfolios the Sharpe index was calculated. The portfolios were compered against a broad index for each of the markets. Da Silva (2001) does deduct both taxes and transaction costs from the return. The returns were calculated from the adjusted price of the stock and were calculated monthly. The risk-free rate used for the Sharpe index was a three month deposited rate from the individual country. He found that, despite evidence suggesting that the strategy may be successful in all countries except for Brazil the findings did not have statistical significance due to the short test period.
Up until the mid 2000s most of the academic literature on DoD often had its focus on the US market. However, starting in 2005, several papers applied the DoD investment strategy upon different markets around the world (e.g. Alles F Fin & Tze Sheng, 2008, Tai-Leung Chong & Keung Luk, 2010 and Rinne & Vähämaa, 2011). In Alles F Fin & Tze Sheng (2008) the DoD was evaluated on the Australian stock market between the years of 2000 and 2006. They set up a portfolio containing the ten highest dividend yielding stock as of December 31. The stocks were picked from the S&P/ASX 50 Index. The returns were continuously compounded and calculated annually. They found that the strategy did produce abnormal returns compared, in this case, with the expected returns calculated via the Capital Asset Pricing Model using data from the ASX All Ordinaries index. In their results they also argued that the abnormal return would hold for transaction costs. Another study that tried to explain the reasons behind the, now somewhat questioned, investment strategy was carried out by Wang, Larsen, Aninina, Akhbari & Nicolas (2011). They suggested that it is herding and the individual’s irrational behavior that are the reasons for why the DoD strategy works. They tested the investment strategy on the Chinese stock market where their findings, they argue, are in line with the behavioral finance theory of that any inefficiencies in the market can be traced back to the above mentioned human behavior.
As a contrast to the above articles that were all written from an academic point of view, Clemens (2011) investigates dividend investing from a practitioner’s view. He shows that a portfolio containing high dividend yield will have a higher return. He tests, between 1992-2012, both the strategy on the US market as well as the world market using world indexes. Clemens (2011) argues that the companies paying out dividend will face lower agency costs that are usually associated with stock buy backs, acquisitions of other companies or mergers, and that it could be a reason for the better performance of those companies’ stocks.
ap Gwilym, et al. (2005) sorted out the 350 biggest companies on the UK stock market measured by market capitalization and created several benchmarks out of them. These are the 100 biggest, the next 250 and then all of the 350 companies. They also used the FT 30 index, which they argue is the one that most closely resembles the DJIA. Portfolios consisting of the top-five and top-ten dividend yielding stock were created from all benchmarks. The portfolios were set up like the original DoD strategy on the first trading day of the year, they were equally weighted and held for one year. An equally weighted portfolio consisting of all the stocks in the FT 30 index was also set up. Like in the study by McQueen et al. (1997) the risk was adjusted by investing at the risk free rate and the transaction costs were assumed to be one percent. Their findings suggested that any return higher than the market disappeared when the strategy was adjusted for risk and transaction costs.
Chong & Luk (2011) tested two versions of the DoD strategy on the Hong Kong market. Both were created at the end of the year and were equally weighted. One consisted of the top-ten dividend yielding stocks listed on the Hong Kong Stock Exchange and the second consisted of the top five stocks of the Hang Seng Index, the sum of 1 000 000 USD was invested in each of the strategies. The returns were calculated by simple return (see eq. 3). They found that the results differed between the two indexes. The strategy applied on the Hong Kong Stock Exchange gave a negative return of almost 1.3 percent. Whilst the strategy applied on the Hang Seng Index provided a positive return of over 8 percent. However, for reasons left out in their paper, the different strategies were only applied on one of the benchmarks. They did not apply the top-ten strategy upon the Hang Seng index and vice versa.
These articles mentioned above are some of those written within the topic of dividend investing and concerning the DoD investment strategy. Articles that use methods of greater importance for this thesis will be presented below. Where the different methods used by researchers, both when it comes to constructing the portfolio and when evaluating the portfolio, will be explained more in depth to give the reader a good basis for Chapter Three.
Dog of the Dow Articles With Methods Used in This Study
McQueen et. al (1997) formed portfolios that consisted of both the top-ten highest dividend yielding stocks as well as all the 30 stocks of the DJIA. The portfolios were created in the beginning of the year and held intact for the remaining year, and both of them were equally weighted. The dividend was reinvested in the stock that paid it at the end of the month. McQueen et al. (1997) used the last quarterly dividends payment, which they annualized, and the price of the stock at the end of the year to calculate the dividend yields. First, they compared the return of the two portfolios with each other. Thereby they found that the top-ten portfolio outperformed the portfolio consisting of all the stocks with three percentage points, which they found to be statistical significant. However, to get a more fair comparison, hence, risk adjusting the returns, McQueen et al. (1997) invested a portion of the total wealth of the top ten-portfolio in Treasury bills so that the standard deviation of the two portfolios was the same. That meant that 13 percent of the total top-ten portfolio value was invested in Treasury bills, shrinking the difference of the returns from three percent to 1.5 percent. Drawing from this result, the conclusion is that most of the abnormal return can be associated to the higher risk in the top-ten portfolio. To fully answer their main question, further deductions were made for transaction costs. Since the top-ten portfolio on average changed almost three firms each year, to compare to only 0.35 firms per year for the comparison portfolio, they argued that with transaction costs of one percent, only 0.95 percentage points in difference between the two portfolios remain. Due to big differences in taxes depending on individual circumstances McQueen et al. (1997) were not able to make any deduction for tax, however they still stated that the strategy would probably not work.
Filbeck and Visscher (1997) applied the DoD strategy on the British market. They calculated the dividend yields of all of the 100 stocks on the Financial Times-Stock Exchange 100 Index on the first of March instead of on the first trading day of the year because the index was created in February 1984. They took the ten stocks with the highest yield and created a portfolio by investing 1 000 GBP in each of the stocks. The returns were calculated by, simply adding the price change with eventual dividend and then dividing the sum with the price in the beginning of the month. After a year they started all over again by investing a new 1 000 GBP in that years ten stocks with the highest dividend yield. The risk free rate was calculated every month and consisted of the twelfth root from the sum of one plus the 90-day Treasury bill return. They compered the top-ten portfolio with the Financial Times-Stock Exchange 100 Index, from which the portfolio was created, by calculating the difference between the top-ten and the market return. They also performed a paired difference test to get the t-statistic. For, as they say, comparison reasons, Filbeck and Visscher (1997) calculate the Sharpe index as well as the Treynor index. They calculated the Sharpe index by dividing the difference of the return (either the top ten portfolio or the market) and the risk-free rate with the standard deviation of that difference and then multiplying the quotient with the root out of 12. They argued that the Sharpe index is the most appropriate measurement in the case when the investor is exposed to company specific risk, i.e. when the portfolio is not fully diversified. The opposite is then true for the Treynor index, hence, it should be the measurement of choice when the portfolio is well diversified and the risk is systematic. The Treynor measure is the difference of the return (either the top ten portfolio or the market) and the risk-free rate divided by the respective beta. Their results indicate that the DoD investment strategy does not perform better than the market. During their test period of 1984-1994 the DoD portfolio beat the market in only four individual years. They lift the fact that they compared the portfolio against an index of a 100 stock as a reason for why it did not perform better when studies comparing against the DJIA, which only has 30 stocks, had shown that the strategy does work.
A portfolio consisting of the top-ten, highest dividend yielding stocks, was created from the Toronto 35 Index by Visscher and Filbeck (2003). In their study they hypothesized that the top-ten portfolio would beat the Toronto 35 Index as well as the broader TSE 300 index. They used the 31 of July as their starting and rebalancing date, one reason was because the Toronto 35 Index was created in March the same year as their sample period started and they wanted to give the index a few months to settle. The second reason was to avoid effects from possible tax trading on the last trading day, when they created their portfolio. The procedure followed closely the procedure of Filbeck and Visscher (1997). The returns were calculated monthly by adding the price change with the dividend and dividing the sum by the price of the stock in the beginning of the month. The portfolio value was determined by investing 10 000 CAD in each of the stocks and then multiplying it with the stock’s return and summarizing the values. After a year, they repeated the procedure by again investing 10 000 CAD in each stock included in the top-ten portfolio. Just like in previous studies, Visscher and Filbeck (2003) calculated both the Sharpe ratio and the Treynor index via the difference between the top-ten portfolio return and the index return. They also used the difference between the return i.e. the abnormal return, to calculate the t-statistic by a paired difference test.
1.2 Problem Discussion
1.3 The Swedish Market
1.4 Purpose and Research Questions
1.6 Methodology & Disposition
2 Theoretical Framework
2.1 Dividend Policy and the Demand for Dividends
2.2 Dogs of the Dow Studies
2.3 Dog of the Dow Articles With Methods Used in This Study
2.4 Log Returns vs. Simple Returns
3.3 Portfolio Construction
3.4 Evaluation of the DoD strategy
3.5 Statistical Significance
4 Results and Analysis
4.1 The companies
4.3 Abnormal Return and Sub-periods
4.5 Risk and Risk Adjusted Return
5.1 The Performance of the DoD Portfolio and the Benchmarks
5.2 Implications of the Findings
5.3 Limitations and Method Critic
GET THE COMPLETE PROJECT
Do Retail Investors Benefit From a High Dividend Yield? The Dogs of the Dow strategy applied on the Swedish stock market.