- A small group of ETFs can produce a very respectable dividend growth.
- The key approach is to collect ETFs with some DG, determine relative weighting, then check overall metrics.
- ETFs as part of a retirement portfolio can reduce portfolio maintenance.
In April of 2013, I wrote the first article with this title, found here. In it was a discussion of the low yield-high dividend growth ETF segment of my retirement portfolio. Much of the how and whys are there and need not be repeated here.
Basically, this effort is to determine if a collection of ETFs, each with some moderate dividend growth, can contribute to a retirement dividend growth portfolio. Advantages are: ability to cover foreign stocks efficiently, provide a vehicle for replacing poor performing individual stocks so that researching stock replacements is not required, include a mechanism (external management) that culls poorly performing stocks so that dividend growth is retained over longer time periods, reduce portfolio maintenance levels so that less time/skill is required. Disadvantages are: reduced dividend income due to ETF expense fees, reduced dividend growth performance due to average/poor performing stocks in the ETF.
First, the result of one cut at the problem. It is a 6 ETF group with an 8yr non-linear dividend growth of 11.5% shown in the graph below.
Note that dividends have more than doubled in the 8 year time frame, which includes the financial crash. For details on how to calculate non-linear regressions, see my previous article, here.
In the graph shown, dividends summed from the 6 ETFs are amounts actually paid each year. However, due to ETF portfolio turnover, the portfolio of companies held in the ETF changes year to year. This means that the further you go back in time, more the portfolio changes and the harder it is to project future performance. However, as you can see from the curves, the 11.5% line is not a bad proxy for the more recent data. Doing a 5yr non-linear calculation (from 2009 on) yields a 14.1% growth, so that is not an issue here. Full disclosure: 2 of the ETFs were not operational in 2006 so their dividends were estimated.
Once a group of ETFs is selected, based primarily on yield and dividend growth, the next step is to determine how much of each should be owned. Traditionally, an equal dollar cost is selected or an equal number of shares. This either assumes equal performance or unknown performance, so they are treated equally. An alternative method is to select a performance metric and weigh allocation accordingly. Several options were selected and 8yr non-linear calculations performed to observe the effects on dividend growth. In each case, dividend growth rate is recorded along with the correlation coefficient 'r'. This parameter has a value of unity if the data fit the assumed curve exactly. Values much below about 0.9 indicate a wide data dispersion. Very low values, like 0.6 or 0.7, indicate the dividend growth rate may not be accurate. The option in the last column was selected for the graph above. This over-weights those ETFs having dividend increases that more closely match exponential growth.
8yr avg DG
Corr Coef 'r'
Non-l.r. DG Rate
Corr Coef 'r'
For calculations, EOY 2013 was used.
The table above shows results of the methods used to select the number of shares allocated to each ETF. In the Equal Cost case, a number was divided by the EOY Price. For the last 3 columns a number was divided by the EOY Price and multiplied by the parameter. One potential reason for the average DG coming in second is the fact that when there is a large negative number in the average, it takes a much higher subsequent number just to get back where you started. For example, from a 50% drop, it takes a 100% rise just to get even.
MorningStar's X-ray program was exercised for the mix of 6 ETFs shown in the graph above. The split globally is about 42% North America and the other half about equally divided between Developed (31%) and Emerging Markets (27%). Note the path taken here: first find ETFs that have dividend growth, determine how much of each to own, then check to see if the asset allocation and other performance metrics are acceptable.
Other metrics, a/o June 2014, gleaned from the X-ray are: Yield - 2.8%; Projected 5yr EPS growth - 10.8%; Price/Book - 1.7; ROA - 6.4; ROE - 17.8; Average expense ratio - 0.34%. About 76% of the underlying stocks are large cap, 16% mid cap and 8% small cap.
The 6 ETFs discussed here are: Vanguard Emerging Markets (NYSEARCA:VWO); Vanguard FTSE All-World ex-US (NYSEARCA:VEU); i Shares MSCI Canada (NYSEARCA:EWC); Vanguard Consumer Staples (NYSEARCA:VDC); iShares Latin America 40 (NYSEARCA:ILF); WisdomTree International Small Cap Div (NYSEARCA:DLS). Salient metrics are in the table below:
Corr Coef 'r'
Data from Morningstar.com (first 5 columns) - June 2014.
Portfolio weighing ($ cost): VWO (7.8%); VEU (20.7%); EWC (19.7%); VDC (21.0%); ILF (16.6%); DLS (14.2%).
It is informative to look at year-to-year dividend growth rates. For this set, they are: 20.6; 33.0; -20.5; 32.2; 18.5; -1.2; 15.0. Before the financial crisis (during a world economic expansion), values were in excess of 20%. After the crash, the world economy has not (yet) fully recovered and is spotty, reflected in the values shown. It would be expected that when the economies recover in each represented area, these numbers will increase…until the next downturn. Barring a hit from the side (ETF management going rogue), dividend performance from a group of ETFs (represented by this sample) should perform as well over long intervals as the economies they represent. Portfolio maintenance is reduced to just monitoring a few metrics once a year.
No statement is made, unequivocally, that this technique should be followed to the letter. More research is needed, both with other ETF candidates and in other time periods. It is offered as a fresh approach to questions of the desirability of adding ETFs to dividend growth retirement portfolios.