Many Seeking Alpha contributors who post articles on portfolio construction filter their searches using either or both of two mechanical stock selection methods. One method is considering only stocks which appear on lists of companies which have increased their dividends every year for a certain number of years. Another is considering only stocks which yield more than a certain threshold, usually ranging from 2% to 4%. While these common filters certainly shield dividend growth investors from many egregiously bad mistakes, and have the virtue of being very easy to apply (especially with a working knowledge of Excel and a ready-made spreadsheet such as David Fish's CCC list), I am going to argue that they leave many stones unturned, particularly for an investor in the "accumulation" phase. I have no problem with the idea that dividend increases and good yield are important fundamental criteria in stock selection for an income investor -- I just believe that somewhat broader criteria do a better job of identifying promising possibilities.
My critique of both filters comes from reflecting upon the concept of Compound Annual Growth Rate (of dividends), as exemplified in Robert Allan Schwartz's web site. Schwartz points to Investopedia's explanation of CAGR:
CAGR isn't the actual return in reality. It's an imaginary number that describes the rate at which an investment would have grown if it grew at a steady rate. You can think of CAGR as a way to smooth out the returns.
Thus, for instance, if a company's stock dividend grows 14.87% from the previous year in year one, 5.29% in year two, 7.26% in year three, and 16.15% in year four, the CAGR is 10.12%. The CAGR is exactly as if the company had smoothly increased the dividend 10.12% in each of years one, two, three and four. (In fact, these are the actual numbers from Air Products and Chemicals Corp. (APD) in the years 2008-2011, as detailed here.)
APD's yearly increases in dividend payout are not unlike those of many of the other dividend "aristocrats" or "champions." The effect of the Great Recession on APD's dividend in year two (2009) is obvious. But now, consider the record of Honeywell International (HON) for the same period. HON's dividend increased 10% in both year one and year two, was unchanged in year three, and increased 13.22% in year four, for a CAGR of 8.55%. Not as good as APD, but steadier. But HON is automatically excluded from consideration by investors who never consider purchasing a stock whose dividend has not increased every year, because of that one year without a dividend increase. Is this reasonable? In fact, there are many stocks on the lists of dividend aristocrats and champions which pay a dividend less than HON's current 2.2%, or can't match the dividend's CAGR.
Now, I am not trying to tout HON (it just happens to be a stock I am familiar with). There may be better examples. (See the list of "Frozen Angels" on the Notes tab of the CCC spreadsheet.) But the point is that here you have a company with a very good long-term record - good enough to have made it a member of the Dow 30 until recently - and a dividend with a better CAGR than many of the aristocrats and champions. Should it automatically be off a dividend growth investor's radar just because it failed to increase the dividend in a single year?
My answer is no, and I believe the remedy is to make CAGR the primary metric for those investors who are focusing on dividend growth. Not to be critical of David Fish, who has done such yeoman work putting together the CCC lists that order stocks by the number of consecutive years they have increased their dividends, but streaks seem to have an inordinate fascination for humans; e.g. (for baseball fans): Joe DiMaggio's 56-game hitting streak, or Don Drysdale and Orel Hershiser's streaks of 58 and 59 consecutive scoreless innings. But as an investor, what would you rather have: a portfolio of the stocks with the longest streaks of increasing dividends, or a portfolio of the stocks with the greatest long-term (say 20- or 25-year) CAGRs?
Well, of course, the two portfolios would probably largely overlap. DiMaggio, Drysdale, and Hershiser didn't only have great streaks, they also had tremendous career statistics. My point is, would you want to be selecting your All-Star team simply on the basis of who had the longest streaks? Wouldn't you also want to be evaluating the slightly under-the-radar players who had the overall best career record, even if they had one or two off-seasons, and never made the headlines for some kind of streak?
So again, I return to long-term CAGR as the overall best measure, and the one that picks up many stocks that fail the streak test. Unfortunately, to my knowledge, we currently have no widely available selection tool based purely on CAGR -- that is, a table which lists stocks in order of the 5, 10, or 20 year CAGR of their dividend, and without limiting the selection to "streakers" or other criteria. Some of the information is there in a matrix on Schwartz's site, but you have to dig it out stock by stock -- the kind of eyeballing for comparisons one can do in a spreadsheet is not possible. Plus, Schwartz takes his universe of stocks from the CCC list, plus the (also filtered) list from Dividends4Life -- meaning that it probably isn't a complete list.
In any case, there is more needed when evaluating dividend stocks by dividend CAGR. Just because CAGR smoothes out fluctuations, it is possible for a stock with a very high CAGR to also have very high volatility in its dividend growth rate - up one year by 25%, flat or even down the next. Conservative investors may prefer not to have a wild ride. Retired investors who are relying on this month's dividend check may be sorely disappointed if XYZ Corp fails to raise its dividend for a year, or even cuts it.
So if we recognize different kinds of dividend investors - some who can shrug off a fluctuation in dividends, and some who cannot - then what we also need is a measure of fluctuation in the CAGR. In statistics, this would be known as the standard deviation. To capture the concept intuitively, if the dividend growth rate of stock #1 fluctuated by only a couple percent over a period of five years - 9, 7, 8, 7, 9 - then its dividend growth would have a very low standard deviation of 1 (from the average of 8). If the dividend growth rates were (to take an extreme example) 15, 1, 8, 1, 15, then the average dividend growth rate would still be 8, but the standard deviation would be 5. This would obviously be unsatisfactory to some investors. But… to the best of my knowledge, not only do we have no unrestricted table of dividend CAGRs, but we have no measure of their standard deviations. Digging up the hidden gems is going to take some work.
So a stock screening filter that automatically excludes companies that have not raised their dividend every year for a decade or more is going to miss some very promising investments. Another filter that I also think is "overzealous," at least for investors in the accumulation phase, is a minimum-yield requirement of 2, 3, or 4%.
To see why, just open the All CCC tab of David Fish's spreadsheet and set a simple filter of 10-year dividend growth rate (column AO) greater than or equal to 15%. You will receive a list of some 92 stocks, many of them familiar dividend-growth favorites like AFLAC (AFL), McDonald's (MCD), and Walgreen (WAG). Now, start cutting the list down by applying various minimum-yield filters. For a minimum yield of 4%, you have only six choices; for a minimum yield of 3%, only 24. But try something different: set the filter to companies with a yield of less than 2%, and more than 1.5, and you have 24 new names to look at. Included on the list are companies such as CVS Caremark (CVS), Lowe's (LOW), Union Pacific (UNP), and not least, IBM (IBM).
Now, consider the following standard compounding table, and apply it to dividend yields. The numbers in each square represent the multiplier for an initial $1 of dividend income, for the given number of years and the given CAGR of the dividend.
Dividend Growth Rate
Once again, we have to appreciate Einstein's most powerful force in the universe. A company that grows the dividend 15% per year will more than double its payout in five years. So if an investor is willing to start with a 1.5% payout, he or she can get to a 3% yield on cost in just five years. Or viewing matters from a different perspective: suppose an investor who requires an initial yield of 3% buys a stock with a 10-year dividend CAGR of 7.5% per year. Looking at the 10-year row of the table, we see that this investor takes a decade to get to a yield on cost of a little over 6%, i.e. 3% times 2.061. But compare the investor who bought the 1.5% yielder with 15% dividend growth ten years ago: he or she is now also receiving a yield on cost of approximately 6%, i.e. 1.5 times 4.046. With a truly outstanding dividend growth rate, it simply does not take that long to catch up.
I've tried not to overdramatize the example. But 7.5% is not an unreasonable target dividend CAGR for a conservative investor. The average 10-year growth rate for the 111 dividend champions is 7.9% (although the number is much lower for many favorites such as AT&T or ConEd). But 15% is also not an unreasonably high target; the all-CCC list even shows 69 stocks with a DGR of 17.5% or above, and 47 with 20% or above. These stocks would be able to catch up with typical target yield requirements still more quickly. With a 17.5% or greater dividend CAGR, a stock that started out with a 1% yield could be producing a yield on cost of 5% or more in a decade.
And, there is an additional benefit to fishing in the waters of lower yield and higher dividend growth rate. After all, we have to ask ourselves, how can these companies afford to be raising their dividends so rapidly in the first place? In many cases, it is because their earnings are growing at rapidly as well; and in the long run, their stock price reflects their earnings growth. I am aware that that statement requires considerable qualification, but my point would simply be that other things being equal, the stock price of a rapidly growing company is very unlikely to sit still while the dividend increases dramatically.
Investors who fixate on dividend growth sometimes almost seem to be saying that they don't care about the share price, as long as they have the reliable dividend growth they want. But of course, price appreciation is nice too. There is no reason why a dividend growth investor can't take some profits from a dividend growth stock that has appreciated nicely, and then roll the money into another dividend growth stock which gives them further diversification, or takes advantage of changed market conditions. Granted that dividend growth is someone's primary objective, there is no reason why price appreciation can't be a secondary objective, and what I am suggesting is that some of these low-yield, high-dividend-growth companies may offer a better chance of achieving the secondary objective.
To be sure, there are potential pitfalls in the low-yield high-DGR strategy I am highlighting. For instance, very high dividend growth rates may not be sustainable, or may be the result of increasing free cash flow payout, neglect of R&D investment, etc. Of course: as if the fundamentals of all stocks don't need to be carefully monitored, both before and after selection. Rather, I am just arguing that the net needs to be cast more widely than many self-identified dividend growth investors here on SA seem willing to do. For instance, David Van Knapp's recent survey of dividend growth investors' favorite stocks reports that Target was removed from the list this year due to "overwhelming negative response." Yet Target has a 10-year DGR of 18.6, which is actually accelerating, while the share price increased more than 25% in the year leading up to Van Knapp's survey. Seems to me that some people are hard to please. I understand that retired investors who want immediate income may well not like a stock that yields only about 2%, but as I have been at pains to emphasize, I think dividend growth investors in the accumulation phase need to be looking at a broader universe.