Measuring Dividend Growth - Linear Case
To those of us who rank dividend growth high on the list of metrics for picking and choosing equities to stuff our portfolios, it is sad to see the lack of coverage this subject has in the literature. This article presents an improved method of characterizing dividend growth, rather than resort to 3- & 5-year averages. This is particularly true for stocks having linear DG (dividend growth) because yearly DG changes every year making averages less accurate.
First of all, there are two types of dividend growth: exponential growth or compounding (if you prefer), and linear growth. Exponential growth has a constant percentage (give or take) dividend growth each year. It is exemplified by Procter & Gamble (NYSE:PG), a company in the Consumer Staples industry, and typical of the low yield-high dividend growth group. Linear dividend growth has a constant amount (in dollars) of dividend growth each year, Enterprise Products Partners (NYSE:EPD), a case in point in the gas/oil pipeline business, is a MLP (Master Limited Partnership) in the high yield-low dividend growth field. REITs (Real Estate Investment Trusts) also have linear dividend growth. Calculations on data assuming there is exponential growth when, in fact, the data are linear are incorrect.
The figure below depicts these characteristics. Dividend data (in $/year) were taken from David Fish's 2011 end-of-year Dividend Champions/Challengers/Contenders lists for the years 1999-2011.
The dark blue and red curves are actual dividends plotted. Light blue and green curves are 'best fit' linear and exponential calculations which will be explained later. This format for graphs is used throughout the article.
If you think EPD is a better stock to have, consider what happens as more time is added to the picture. The chart below is 25 years worth, 13 real, 12 projected. Stocks like EPD are better short term for the yield, but stocks like PG pay off in the long run if you buy and hold.
The problem, IMO, is that basically the only metrics given for dividend growth are 3- and 5-year percent averages. These fail to tell a true story in many cases. Since, in linear dividend growth cases, percentage dividend growth decreases every year, these metrics are misleading at best. For example: the one-year DG rates for EPD from 2005-2011 are: 9.75, 8.13, 6.69, 6.79, 5.87, 5.54, 5.25, decreasing every year except one. The most recent 3-yr avg is 5.55. The most recent 5-yr avg is 6.03, and the 7-yr is 6.86. These calculations match the 1,3,5 yr DGR columns in Fish's CCC spreadsheet. The further back you go, the more the variation.
It is difficult to collect a deep history of dividends. It is also useful to compare different companies in the same time period. An additional factor is to have a long enough time period to show a reasonable sample of performance. For these reasons, an eight year collection of yearly dividends is selected, giving seven years of dividend growth. This is also about equal to the average business cycle, so a reasonable history should be obtained. Sometimes shorter time periods are necessary (as we will see), but 4 years is probably a minimum for linear dividend growth stocks.
A series of three articles will be written, this being the second, detailing a method for determining a dividend growth metric in the general sense. The first article, published as "Measuring Dividend Growth, Exponential Case," addressed exponential (compounding) dividend growth, which accounts for the majority of stocks. This (current) article will cover linear dividend growth, which cover MLPs, REITs and other higher yielding equities (including Utilities and Telecommunications). The third article will address ETFs, which distribute dividends from a collection of stocks (less fees). This area is unique because the population of equities contributing to distributions varies over time due to portfolio turnover.
The table below shows some analyses of stocks exhibiting linear DG in four categories: MLPs, REITs, Telecoms, and Utilities. LR (Linear Regression), utilizing the method described in the appendix, characterizes the dividend data, since it 'fits' a straight line to a straight (more or less) line of dividend data. This method calculates the coefficients for a linear dividend equation (dividend = 'a' times/year plus 'b') that attempts a 'best fit' to the stock dividend data. Linear regression is sometimes referred to as a 'least squares' calculation.
The 2012 DG LR column is the projected dividend growth for the year following the last data point. In the linear DG case, a single year's calculation is valid. The FOM (Figure of Merit) column uses the equation developed in my 'Dividend Growth as a Figure of Merit' article. For the linear DG case, it is: FOM = Y * (DG + 10), where Y is yield and DG is the single year LR calculation in the 2012 DG LR column. Both Y and DG are expressed in percent.
All calculations were made using dividend (2004-2011) and yield data from David Fish's CCC spreadsheets, EOY 2011. Results are shown in decreasing FOM.
Stocks listed are: Omega Healthcare Investors (NYSE:OHI); Genesis Energy (NYSE:GEL); Buckeye Partners (NYSE:BPL); Shaw Communications (NYSE:SJR); Kinder Morgan Energy (NYSE:KMP); National Health Investors (NYSE:NHI); Rogers Communications (NYSE:RCI); Plains All American Pipeline (NYSE:PAA); AT&T (NYSE:T); Enterprise Products Partners ; Avista (NYSE:AVA); Sunoco Logistics Partners (NYSE:SXL); Duke Energy (NYSE:DUK); Health Care REIT (NYSE:HCN); Verizon Communications (NYSE:VZ); ONEOK Partners (NYSE:OKS); NextEra Energy (NYSE:NEE); HCP (NYSE:HCP); Consolidated Edison (NYSE:ED); Federal Realty Investment Trust (NYSE:FRT).
Stock | Type | Yield % | a ; b From LR | 2012 dg LR % | FOM |
OHI | REIT | 8.27 | 0.1100 ; 0.6200 | 7.33 | 143 |
GEL | MLP | 6.10 | 0.1657 ; 0.3305 | 10.01 | 122 |
BPL | MLP | 6.41 | 0.1990 ; 2.4313 | 4.95 | 96 |
SJR | Telecom | 4.66 | 0.0982 ; 0.1502 | 10.49 | 95 |
KMP | MLP | 5.46 | 0.2625 ; 2.5050 | 5.70 | 86 |
NHI | REIT | 5.91 | 0.1092 ; 1.5616 | 4.49 | 86 |
RCI | Telecom | 3.62 | 0.1906 ;-0.1429 | 13.79 | 86 |
PAA | MLP | 5.42 | 0.2330 ; 2.1782 | 5.76 | 85 |
T | Telecom | 5.82 | 0.0756 ; 1.1511 | 4.31 | 83 |
EPD | MLP | 5.28 | 0.1263 ; 1.4040 | 5.23 | 80 |
AVA | Utility | 4.27 | 0.0855 ; 0.3432 | 8.32 | 78 |
SXL | MLP | 4.20 | 0.1225 ; 0.6307 | 7.60 | 74 |
DUK | Utility | 4.55 | 0.0544 ; 0.5941 | 5.28 | 70 |
HCN | REIT | 5.43 | 0.0656 ; 2.2871 | 2.33 | 67 |
VZ | Telecom | 4.99 | 0.0635 ; 1.4501 | 3.24 | 66 |
OKS | MLP | 4.12 | 0.1123 ; 1.4713 | 4.74 | 61 |
NEE | Utility | 3.61 | 0.1252 ; 1.1531 | 5.81 | 57 |
HCP | REIT | 4.63 | 0.0370 ; 1.6171 | 1.93 | 55 |
ED | Utility | 3.87 | 0.0200 ; 2.2400 | 0.83 | 42 |
FRT | REIT | 3.04 | 0.1069 ; 1.9080 | 3.87 | 42 |
First, a comment on the number of significant figures displayed in the table. The constants a & b were calculated from dividend data in David Fish's CCC spreadsheet where it is given in 1/100s of a penny. The values are accurate as shown, but that level of precision may not be needed in subsequent calculations. It should be noted that in this process two close numbers are subtracted and vital data are lost if rounding is done prematurely. As the analysis progresses, the implied precision is reduced to more practical amounts. It should be noted that these data are for one point in time (end of 2011) and will vary somewhat over later time periods. The biggest variations will be in yield. OK, guys?
The intent of this article is to present a methodology for dealing with linear dividend growth, not recommend or downgrade particular stocks. Stocks were all selected from Fish's CCC database. Of the 20, I am long on 12 (13, if my wife is included). The rest were selected, in part due to my interest and in part to round out the variety.
However, since I have rank-ordered them in a Figure of Merit sense, some comments are valid if they are pertinent to the analysis. Of the 20 analyzed, 10 had an essentially straight dividend curve, matched by the LR curve. These were: SXL, EPD, HCP, ED, HCN (1 mid data point low), VZ, BPL, KMP, NEE, NHI. Five stocks had a slightly higher slope in early years, like 2 straight lines with a bend in the middle. In all of these, differences in the dividend and the proxy LR curve in the final year were less than 4%, typically 2%. Stocks in this group were: PAA, OKS, FRT, DUK, T. These warrant a closer examination in future years to see if the lower growth rate experienced in the latter years is sustained, or if dividends pop back when the economy recovers. The worst case, DUK, is shown in the graph below:
The other stock shown is interesting. Not only is OHI a stellar performer (top of the list), it has been in overdrive for the last 2 years.
The two Canadian telecoms, SJR and RCI, have solid linear dividend growth in the past 4 years. Dividends in the prior 4 years were not representative of the latter data so were not used. Instead, the first 4 years were backfilled. Had I used early low dividend data, growth rates would have been unrealistically high. Dividends and LR are shown below (again, only the last 4 data points are real):
Stocks GEL and AVA are noteworthy because it looks like they started out as exponential growth candidates. More time is needed to see how they sort out, graphs are below.
General comments on the table: 1) All 4 stock types are well represented in the chart, although Utilities (as expected) cluster near the low end. 2) None of the stocks displayed any pronounced sign of the Financial Crisis in terms of lower dividends. This was a pleasant (and welcome) surprise. I would not, however, jump to the conclusion that all linear DG stocks were unaffected by this event. 3) It should be noted that income from linear DG stocks may not keep up with inflation (which grows exponentially). Data in the 2012 DG LR column indicate that 3 stocks are marginal with regard to current inflation levels (ED, HCP, HCN). Remember that 3&5 year DG averages overstate reality. Linear DG stocks have a place in retirement portfolios when handled properly.
One of the features in the FOM formula for linear dg, [FOM = Y * (dg + 10)], is that it works for stocks with no DG, including bonds. It is possible to compare a linear growth stock with a bond. For example: Investment: $10,000. Bond yield 5.5%, $550/year, FOM 55. HCP has a FOM of 55, the same as the bond. The question is - how many years will it take HCP to yield the same income as the bond? HCP stock price EOY 2011 was $41.43 buying 241 shares. Dividend (8th year) from the LR equation is: dividend = 0.037*8+1.617 = $1.91/share or $460 for 241 shares. At an increase of 0.037*241=$8.92/year, the LR equation is 8.92 * T + 460 = 550, or T = 10 years. Does this seem reasonable? In the early years you gain more income from the bond. This difference gradually decreases to where after 10 years, the stock pays more. The 10 year answer was no surprise, the FOM formula was derived based on equal FOMs, from any combination of Y and DG, producing the same dividend/income at the 10 year mark.
So, why is all this important, nitpicking over a percent or two in DG? I am looking for a stable DG characteristic for stocks that defines this important metric, one that represents the stock in all phases of a business cycle. I use this in FOM (Figure of Merit) calculations, as described in "Dividend Growth as a Figure of Merit". Also, DG metrics play an important part of my retirement portfolio distribution scheme as described in "Life Cycle of Retirement Portfolios, Part 1". In this scheme, higher DG in portfolio segments lead to a higher initial portfolio payout which may justify a smaller portfolio value needed at the beginning of retirement. (See note 1). I also use dividend growth trends and FOM comparisons in selecting stocks to buy. These metrics are also used in monitoring stocks in my portfolio. Looking at a graph comparing actual dividends with the LR proxy is vital in this process to catch subtle changes. For these reasons, I need a fast and accurate method for determining dividend growth. It is my intent not to re-calculate LR each year, unless it obviously no longer represents the stock.
Dividend growth is important in a retirement portfolio; there is need for both low yield - high DG (exponential) and high(er) yield - low DG (linear) stocks. Bonds can be lumped in with the latter, if you prefer. You get most of the income from the latter group in early retirement years and most of the income in later years from the first group. My portfolio has a 30% allocation to each.
Note 1: Longer term projections for DG are only important for portfolio segments. Some degradation over time is acceptable. The segment may require some turnover during retirement to maintain an adequate income stream.
Appendix
The method used in determining the linear Regression DG used in this article requires use of a spreadsheet. The following outlines the general procedure:
1) Input year number and yearly dividends in a spreadsheet. It is better to arrange these in columns, year (starting with one (1)), dividend.
2) Copy the 2 columns, year and dividend. Go to http://xuru.org/rt/lr.asp. Paste data in the box. Right click to paste.
3) Click on 'calculate'. Copy the numbers 'a' and 'b' from the equation y=ax+b.
4) In the spreadsheet, Copy 'a' and 'b' in the column the right of the dividend column, above the dividend data. The number 'a' is the slope of line described by the equation. It is also the yearly dividend increase in dollars. The second number 'b' is the 'y-intercept' or the dividend for year zero.
5) Calculate yearly dividends in this LR column by the formula LR = 'a' * T + 'b', where T is the year.
6) Calculate current DG = 'a' / final year's LR dividend.
Example:
Stock: OHI Dividends 2004-2011
Year | Dividend | Linear Regression |
a > | 0.11 | |
b > | 0.62 | |
1 | 0.72 | .73 |
2 | 0.85 | .84 |
3 | 0.96 | .95 |
4 | 1.08 | 1.06 |
5 | 1.19 | 1.17 |
6 | 1.2 | 1.28 |
7 | 1.37 | 1.39 |
8 | 1.55 | 1.50 |
2012 DG = a / year 8 LR = 0.11 / 1.50 = 0.073 = 7.3%