Kurtis Hemmerling
Transcendent Stock & ETF Portfolios
Follow this author
Daily digest on all my authors
Send Message

x

Author Following Options:

Invest smart with exceptional portfolio models built for investors just like you. Learn more»

See their articles on your Seeking Alpha homepage and in your feed.

Real-time alerts on this author

Get their new articles emailed to you as they are published.

Get all your favorite authors’ new articles emailed to you in one daily digest.

Kurtis Hemmerling

Stop FollowingKurtis Hemmerling

View as an RSS Feed

## 5 Common Errors Of A Dividend Growth Forecast [View article]

But I am suggesting that your use of the geometric average is incorrect. I have never seen anyone use a geometric mean to tally portfolio yield. The yield does not need a geometric average applied to it. The way you apply it in this instance (to yield) would have this effect: I have 10 apples, 2 apples and 5 apples. 17 apples. The geometric mean is 4.64159. I quickly re-count my apples (3 piles) and suddenly I have 13.92477 apples?

Try out the CAGR calculator. http://bit.ly/ZJ8quk

If you double your money in 10 years, the compound annual growth rate is 7.18%.

But in the second example, if I have 3 different stocks (equal weight), one with a 9% yield, a 5% yield and a 1% yield, I can say that I have an average 5% yield. If I have $10,000 invested in each stock I will actually earn $1,500 in dividends or $500 on average or 5% average. This uses the simple arithmetic mean and not the geometric.

As for my use of the geometric mean or compounding growth ratios - look at the chart that starts with a 3.25% yield on cost and 15% growth rate. If I had expressed this as a arithmetic average I would have multiplied 24 years (first year isn't considered growth but is the baseline year) by 0.15 to get 3.6. This would then represent the multiplier to my original annual dividend stream (which I chose to express as yield on cost). If I used a simple arithmetic average of 15% growth, the final YOC would have shown 11.7%. It is obvious that I did not use the arithmetic average. If I didn't use the arithmetic mean in my forward growth numbers, why did you assume I used it in my trailing growth numbers? Perhaps the answer has more to do with a view that I am attacking dividend growth investing instead of actually wondering if I used a standard CAGR calculation.

If you find a money manager using one of those geometric averages for a stock index, run. http://bit.ly/1FaOp4i

## 5 Common Errors Of A Dividend Growth Forecast [View article]

The testing here is not limited to the CCC list. It may be that instead of 70% survivor-ship, one investor or another might feel inclined to include exceptions. Yes, this might change survivorship bias by a few percent depending on the investor leniency. But if you add more exceptions on one end (current list), you need to add it to the other (2001 historical list) with a similar result.

But to be fair, the one survivorship test did use the CCC list. I ran a point in time screen and tallied all div growth stocks with 5 years or more of div growth as of 2001. This P123 list was compared against the number of stocks on the current CCC list which have 19 years or more of consecutive div growth. It is my experience that the CCC list, as you say, includes certain exceptions which makes the list a tiny bit larger than the strict rules-based approach. Thus, in 2001 David Fish may have added a few more names than my 2001 list, and the survivorship rate would even be worse than the 77% attrition rate I calculated.

## 5 Common Errors Of A Dividend Growth Forecast [View article]

As for survivorship bias, I consider a stock as not surviving once it is removed from dividend growth universe. In my non-biased dividend example, a stock was sold as soon as it failed to increase the dividend. The 2.5% reduction in the compound annual growth rate shows the consequence of selling a stock when the dividend fails to rise and distributing the capital amongst the surviving names. This is showing the difference between computing the average return of the CCC list (as one example) and a real investment portfolio following the exact same guidelines as the CCC list.

I would suggest that the yield of the replacement stock is very important for yield on cost projections. If you invested $100K and have a 10% yield on cost after 5 years, and now you sell your shares and have $200K in capital to redeploy, you would need to invest in a stock with a minimum of a 5% current dividend yield to collect $10K in dividends and maintain your 10% yield on cost.

## 5 Common Errors Of A Dividend Growth Forecast [View article]

Perhaps you can add additional explanation why you use the geometric mean in this case. I disagree with your use of it.

As for my hypothetical example.... all growth rate calculations and figures used for the sake of this article do in fact use the compound annual growth rate or the geometric mean. The software I use, Portfolio123, uses CAGR for growth calculations. It is standard practice and I operate under the assumption that all readers would know this.

Back to the portfolio...if each variable already represents the 5 year geometric mean, you should not apply it a second time. If one portfolio has a compound annual growth rate of 1% and a second portfolio has a compound annual growth rate of 100%, the investor (holding both portfolios equal-weight) could say that his average return per year is 50.5% as this number already includes the compound annual growth rate. He would definitely not say that his average return is the geometric average of a geometric average which is 10%.

As for yield, I have no idea why you would apply a geometric average to it. That is a bit boggling if you ask me. It is indicated or expected dividend to price ratio to provide the first year of income so that you can apply the compound annual growth rate of dividend growth to it. If I use the geometric average of yields as you suggest the following bizairre situation would be true:

$30,000 invested. $10,000 per stock. Stock A gives $500 in dividends. Stock B gives $300 in dividends. Stock C gives $100 in dividends. I invested $30,000 and received $900 in dividends. My yield on cost - or dividends received for that year dividend by my initial investment is 3%. That is true. By using the geometric mean I am told that my portfolio yield on cost is 2.15% which means I made only $650 in dividends for the year. But this is not true.

You cannot apply a geometric average to every possible calculation. I use the compound annual growth rate where it makes sense to do so. But I do not agree in the manner in which you attempt to apply it. If you would like to explain your reason for doing so in the form of a hypothetical example I would be most interested.

## 5 Common Errors Of A Dividend Growth Forecast [View article]

http://bit.ly/RmM4R0

I had to cross-reference these tickers with the P123 database as sometimes it used a slightly different ticker or name. For instance, Household International (historical ticker HI) is linked to the name HSBC, which bought the firm up and not Hillenbrand which currently has the ticker HI.

I recreated and verified the list (as best I could) for every year since 2001 based on the link above.

## 5 Common Errors Of A Dividend Growth Forecast [View article]

My motivation is to gain a better understanding of the DGI life-cycle and which sub-types have more/less risk. Many principles of DGI can be applied to other areas of the market. For example, stocks with 10 years of DGI and 10 years of consecutive earnings growth have performed very well even if the number of tickers are low. Some might choose to overweight these names.

## 5 Common Errors Of A Dividend Growth Forecast [View article]

## 5 Common Errors Of A Dividend Growth Forecast [View article]

The second test was to record all stocks (as of 2001) which had a minimum of 5 years of div growth. I then zipped forward to current and counted which of these tickers had 19 years or more of dividend growth (the 5 years at inception plus the 14 year test).

The survivorship rate in both studies were similar.

## The Hidden Risk Of High Dividend Growth Stocks [View article]

Thank you for your input. Can you give us a little more information?

Do you mean that you invest in high trailing dividend growth stocks but the income appears to compound at a similar rate going forward and that over time your total yield on cost is greater than a high yield selection (while still investing in div growth stocks)? I will dig more into this subject next.

It is also possible that there are sector, size, minimum yield, starting date, holding period and other factors that need to be accounted for when using a high div growth to project forward income streams but the initial round of testing was to use as few pieces and processes as possible to see if factor A on Universe A leads to income alpha or at least an income stream that is somewhat near our projections.

## The Hidden Risk Of High Dividend Growth Stocks [View article]

Perhaps I will narrow the test and run the top 10% or 25% minus the bottom 10% or 25% to see the net difference.

Good ideas. Thank you.

## The Hidden Risk Of High Dividend Growth Stocks [View article]

I plan to test a few methods before the next article. What you suggest is good, stagger the starting date. I am also interested in graduated lengths of holding periods from 1 to 10 years to see if dividend growth persists over a certain horizon and then decays. I also want to couple dividend growth with earnings growth (earnings increase annually for x amount of years) to see if that adds any stability. And a couple value factors that might suggest which companies have long-term ability to raise the dividend. Using forward earnings growth estimates or payout ratio is tricky at best.

Lots to dig into, so little time.

Thanks again.

## The Hidden Risk Of High Dividend Growth Stocks [View article]

As User123 accurately pointed out, the 2013 Fiscal Cliff scare prompted many companies to shift the first dividend of 2013 into the end of the 2012 calendar year. This gave the appearance of additional income in 2012 and a drop in 2013 while not affecting the 2 year totals. Some examples of this are as follows: ACU, AFSI, BANF, BEN, ELS, EV, HAS, JBHT, LLTC, MDT, MNRO, MSM, NEU, ROP, SWY, THFF and TR. This put a portion of 2013 income into 2012 which doubled the appearance of income volatility as it was added to one side and subtracted from the other.

In addition to this, many companies offered special dividends. Some did this in response to the 2013 Fiscal Cliff scare. A partial list of the dividend growth firms which offered special dividends in 2012 are: RBCAA, COST, WLK, NEU, NATL, WSO, FAST, and GK. When the special dividends of 2012 are totaled up, it is equal to more than 25% of the portfolio income generated in 2013. Special dividends have a 1x multiplier effect whereas the shifting of dividends from 2013 to 2012 have a 2x multiplier effect as outlined in the previous paragraph..

Finally, there is a smaller list of names that needed to be replaced because the dividend did not increase (e.g. DCM), or were bought out (MOLX). Turnover can also reduce income in certain instances. Still others such as Devry shifted the payment slightly so 2013’s dividend landed in 2012, but then they kept this new schedule so it only had a one-time effect of boosting 2012’s income.

This explains the rise in 2012 income and the subsequent drop in 2013. The trailing dividend growth factor used to select stocks only used regular dividends while the excel chart of income shows what actually lands in the account (all forms of income including special dividends).

As a rule of thumb, I don’t give out raw data as this is in violation of the Compustat license. But if any respectful commenter has a legitimate question, I will investigate and come up with the answer if at all possible. I do, however, reserve the right to decline engagement from those who troll Seeking Alpha articles for ulterior motives – those of which I am not interested enough to investigate.

I would like to thank those who read this article, posted insightful comments and remained patient while I investigated what led to the income drop in 2013.

## The Hidden Risk Of High Dividend Growth Stocks [View article]

1)The chart based on calendar years was something I did in excel by adding up the thousands of dividend transactions based on when the cash was deposited in the account in the simulator. This is to simulate actual cash per year that is available to the investor for spending or re-investment.

2)The trailing dividend growth factor used to screen dividend growth stocks (to the best of my knowledge) uses dividends paid as per the annual report. Thus, it is not affected by late payments or pay dates that cross over to the next calendar year.

3)The excel chart showing income does include special dividends as this shows what would actually happen to your account. The trailing dividend growth factor to screen and determine what a dividend growth stock is or isn't, and to calculate the % of div growth, does not use special dividends. It only used regular dividends to calculate the trailing compound annual growth rate.

The trouble is, there are differences between reporting periods, fiscal years and when or how this all trickles down to your account. I felt this was the best method as it mimics an actual process. I adjust my screening process based on how the company reports its data but I report my portfolio or income value based on what I see in my trading account.

Maybe I will add some extra documentation to clarify future charts.

Thank you for the suggestions and helpful comments.

## The Hidden Risk Of High Dividend Growth Stocks [View article]

The simulation needs to use pay dates as it tracks inflows and outflows of cash, determines when enough cash available to purchase a new holding and so forth. You cannot buy shares until the cash lands in the account.

The next time I report income I think I will use a different chart so as not to cause confusion. Maybe more data points or a different format using rolling periods...I will keep thinking about it.

## The Hidden Risk Of High Dividend Growth Stocks [View article]

Look at ELS as an example.

http://bit.ly/1AlHaW0

The chart in my article tracks when dividends based on pay date - or when the money shows up. While 2012 had 5 actual dividend payments to the account, 2013 had only 3. Depending on when the money shows up in your account it will be subtracted from one year and added to the next. This doubles the effect.

If a stock paid a quarterly dividend of $10 per share, and one year had $50 in dividends based on pay dates and the other had $30 based on pay dates - it will appear that income declined by 40% when just looking at income received by calendar year. The more frequently you sample the portfolio data, the less lumpy the returns appear. But it doesn't alter the total income, just which year it is reported in.

The question I was pursuing before this red herring is this: how come high trailing dividend growth doesn't compound portfolio size as some think it should?

Thank you to those who were willing to patiently wait the 24 hours so I could investigate the case of the lumpy dividend chart.