High Dividend Yield
High Yield DJIA30
High Yield TSX60
Low P/E DJIA
Low P/E TSX60
Low P/B DJIA
Low P/B TSX60
Dividends at Risk
Dividend Risk DJIA30
Dividend Risk TSX60
Value Ratio DJIA30
Value Ratio TSX60
The Rothery Report
Graham Value Stocks
Tip Sheet Archive
High Dividend Yield
Investors seeking income often buy stocks that pay a healthy dividend. Canadian dividends are taxed more favorably than interest from GICs but dividends may be reduced and the initial investment is not guaranteed. Dividend investors employ a variety of popular approaches to pick stocks including dividend growth, relative dividend yield and the Dogs of the Dow. Here I focus on the wildly popular Dogs of the Dow approach which has a few flaws.
The basic Dogs formula is to buy an equal amount of the top ten yielding stocks in the Dow each year. The approach was first advocated by Michael O'Higgins in his book Beating the Dow. From 1973 to 1996 the Dogs of the Dow returned 20.3% which far surpassed the Dow's return of 15.8%. O'Higgins also suggested two variants which I'll call the Dow Five and the Dow Four. The Dow Five is formed by selecting the five lowest priced of the Dogs of the Dow. The Dow Four further refines the list and holds the four highest priced of the Dow Five (See Table 1).
The Motley Fool (www.fool.com) went a step further and created the Foolish Four. The Foolish Four tweaks the Dow Four by putting 40% of the portfolio in the lowest priced stock and 20% in each of the others.
Why bother with all of these variations? The answer is simple, each has a better performance record than the last (See Table 2). With strong past performance, many investors jumped on the Dogs bandwagon.
Popularity bred criticism and charges of data mining. Data mining is the process of testing many different stock-picking methods against the historical record to find a winning approach. For every 100 methods tested, a handful of apparent winners will be found by chance alone. These apparent winners are nothing more than statistical artifacts and are unlikely to be profitable in the future. Data mining is often compounded when the researcher tweaks a successful method to achieve additional gains. Continual tweaking will eventually result in a scheme that captures all of the best possible picks from the historical record. The result is high gains on paper, but such hyper-optimized methods regularly fail dramatically when put into practice.
The many Dogs of the Dow approaches seem to follow a data mining pattern. They start with the basic high-yield model and move into refinements. There are a variety of tests that can be used to detect data mining but perhaps the simplest is to keep an eye out for strange parameters. For instance, picking high-yield stocks may make sense but why select stocks based on price-per-share? A stock's price, on its own, doesn't represent anything fundamental about a company. After all, management could wake up one morning and split the stock two-for-one. After the split, there'd be twice as many shares each selling for half the price. Why should the Fool's method select a stock after an arbitrary split but not before?
Weird criteria can often be explained away, but if an approach is good then it should stand the test of time. For instnace, if you change the study period then the approach should still work. This is exactly what Grant McQueen and Steven Thorley did in their entertaining study "Mining Fool's Gold". Instead of looking at the 1973 to 1996 period they took a step back and examined the 1949 to 1972 period. From 1949 to 1972 the Dow gained an average of 14.11% a year whereas the Foolish Four gained 14.43% a year. A 0.32% performance boost is modest and quickly eroded by the corrosive impact of taxes and commissions. The Fool's themselves have pulled their approach, saying
If the complicated Foolish Four doesn't work then what about the basic high-yield approach? Here the evidence is mixed. David Dreman in Contrarian Investment Strategies: The Next Generation studied high-yield US stocks from January 1 1970 to December 31 1996 and found some evidence for outperformance (See Table 3). In his study, Dreman looked at the largest 1,500 stocks in the US and sorted them into groups of five ("quinitiles") based on their price-to-dividend ratio. The 20% lowest price-to-dividend (or highest dividend yield) stocks were put into quintile one and the highest 20% (lowest yielders) were put into quintile five. From 1970 to 1996, selecting the second lowest price-to-dividend group each year was most profitable (before taxes and commissions) and returned 17.5% vs. 14.9% for the market.
Before getting too happy about a high-yield approach it is useful to look at as much history as possible. Here I turn to What Works on Wall Street by James O'Shaughnessy and his broad study of high-yield US stocks from 1951 to 1996 (See Table 4). O'Shaughnessy split his universe of stocks into groups of ten ("deciles") with the top 10% of yields in decile one and the lowest 10% of yields in decile ten. A slight high-yield advantage remains but avoiding the highest yielding stocks continues to be a better strategy. The difference between the market's return of 13.2% and the 14.0% returned by high-yielding stocks is only 0.8%. Once again, taxes and commissions can easily consume more than 0.8% per year.
One possible reason for the lackluster performance of the highest yielding stocks is that there is an increased risk that they will cut their dividend. Dividend yield is usually based on last year's dividend divided by the current stock price. Should a company falter its stock price usually goes down rapidly which pushes up the apparent dividend yield. If the stock is in real trouble then its dividend is often reduced or cut altogether. Suddenly a juicy yield of 10% based on past dividends disappears. A very unhappy circumstance. Dividend investors should make sure that a company earns more than enough to cover its dividend and that it is reasonably likely to do so in the future.
Another possible reason for the poorer perfomance of the highest yielding stocks is that certain sectors are over represented. For instance, many utility companies have above average dividend yields. David Dreman took a look at this effect by sorting his list of stocks by yield within 44 industry groups. The average performance of selecting high-yield stocks in each industry group as shown in Table 5.
For high-yield investors, Dreman's results are good news. It appears that one can buy high yield stocks accross many industries and do quite well.
A simple high-yield strategy appears to provide a performance boost but the pickings may be slim. Backing away from the highest yielding stocks, or diversifying across many industries, looks like a better strategy. These stocks are more likely to have dividends that are well covered by earnings and may experience dividend growth. In the end, it is important to avoid being too greedy when it comes to high-yield stocks and fancy strategies.
Additional Articles on Dividends and High Yield Stocks:
|Disclaimers: Consult with a qualified investment adviser before trading. Past performance is a poor indicator of future performance. The information on this site, and in its related newsletters, is not intended to be, nor does it constitute, investment advice or recommendations. The information on this site is in no way guaranteed for completeness, accuracy or in any other way. More...|