Dividend-Based Design Example
I use the term design when I break the components of a retirement strategy into discrete elements and then tie them together.
This was my first attempt at dividend-based design. It was a success. It includes a breakthrough.
Goals
These were my goals:
1) a safe withdrawal rate of 4% of my initial balance (plus inflation),
2) an indefinitely long withdrawal period and
3) long-term growth.
I based my strategy on dividend payouts. I used TIPS at a 2% real interest rate to fill in any gaps.
S&P500 Index Dividend Payments
I focused on the S&P500 index because I have the relevant data. I extracted the annual January values from Professor Robert Shiller’s S&P500 data.
Professor Shiller’s Web Site
I determined the payout ratio in terms of smoothed earnings E10.
I calculated E10, which is the average of the previous ten years of (real) earnings, by dividing the current value of the Real Price by P/E10. Then I divided the (real) dividend amount by the smoothed (real) earnings E10.
I plotted the Payout Ratio 100D/E10 (i.e., 100% times the dividend amount divided by the average earnings throughout the previous decade) versus the percentage earnings yield 100E10/P (which is 100/[P/E10]). I used Excel’s plotting capability to determine regression equations (i.e., linear curve fits).
Using 1921-1980 data, the equation is: y = -2.547x+83.55, where y is the payout ratio in percent and x is 100E10/P. R-squared is 0.249. Using 1981-2004 data, the equation is: y = -0.9489x+55.329 and R-squared is 0.2159. The payout ratio was 36.4% in 1921. Otherwise, it has remained above 40% ever since. The payout ratio was 83.4% in 1951. It was above 80% nine times from 1928 through 1941. The total was ten times since 1921. The payout ratio has remained below 80% since 1941.
The spread on the data is greatest at the lower values of earnings yield.
The most useful information that I can extract is that recent dividend payout ratios have ranged between 40% and 60%. Historically, with several exceptions, payout ratios have been between 40% and 80%.
Future S&P500 Dividend Yields
I estimate future S&P500 dividend yields to be between 40% and 60% of the percentage earnings yield 100E10/P.
Scaling from P/E10
The payout ratio equals the (percentage) dividend yield D divided by the percentage earnings yield 100E10/P. This equals the (percentage) dividend yield D times P/E10 divided by 100. The payout ratio, when expressed as a percentage, equals the (percentage) dividend yield D times P/E10.
To calculate the percentage dividend yield D, we divide an estimate of the payout ratio (between 40% and 60%) by our estimate of P/E10.
If P/E10 equals 10 and if the payout ratios range between 40% and 60%, then dividend yields will range between 4.0% and 6.0%.
If P/E10 equals 15 and if the payout ratios range between 40% and 60%, then dividend yields will range between 2.7% and 4.0%.
If P/E10 equals 20 and if the payout ratios range between 40% and 60%, then dividend yields will range between 2.0% and 3.0%.
Scaling from the S&P500
I use DVY, a high dividend exchange traded fund, to calculated a scale factor to calculate the yield that a dividend investor is likely to receive when compared to the S&P500 index.
DVY yield was 2.95% when I checked today. The S&P500 index yield was 1.71%. The scale factor for dividend investors is (the ratio, 2.95/1.71 or) 1.725.
Scaled Dividend Yields
We estimate future values of P/E10 (of the S&P500) based on Professor Shiller’s S&P500 data. We use these to estimate future S&P500 dividend yields. We multiply these by our 1.725 scale factor to estimate what a dividend investor is likely to receive from stable companies.
If P/E10 equals 10 and if the payout ratios range between 40% and 60%, then S&P500 dividend yields will range between 4.0% and 6.0%. Dividend investors are likely to get 6.9% to 10.4%.
If P/E10 equals 15 and if the payout ratios range between 40% and 60%, then S&P500 dividend yields will range between 2.7% and 4.0%. Dividend investors are likely to get 4.7% to 6.9%.
If P/E10 equals 20 and if the payout ratios range between 40% and 60%, then S&P500 dividend yields will range between 3.5% and 5.2%. Dividend investors are likely to get 6.9% to 10.4%.
Projecting Dividend Yields
In January 1985, the S&P500 dividend yield was 4.41%. P/E10 was 10.0. The January dividend yield has remained below 4.00% ever since. This was 15 years before the year 2000 peak.
P/E10 was close to 15 in January of 1987 and 1989. It was 15.6 in 1991. These were 13, 11 and 9 years before the year 2000 peak.
It would be reasonable to expect dividend yields and valuations to return to their historical levels as rapidly as they rose. There is some evidence to suggest that they should return faster.
It is reasonable for us to expect P/E10 to fall to 15 by 2009 to 2013. This would lead us to estimate S&P500 dividend yields of 2.7% to 4.0%. This would lead us to anticipate that dividend investors could get yields of 4.7% to 6.0% from good companies.
That is, dividend investors should be able to get dividend yields between 4.7% and 6.0% within 3 to 7 years.
It is reasonable for us to expect P/E10 to fall to 10 by 2015. This would lead us to estimate S&P500 dividend yields of 4.0% to 6.0%. Dividend investors should be able to get yields of 6.9% to 10.4% from good companies.
Dividend investors should be able to get dividend yields between 6.9% and 10.4% in 9 years.
Projecting Dividend Growth
I have calculated the percentages that dividends have grown relative to the first year’s dividend amount. I have calculated four-year moving averages as well.
Theoretically, dividend growth should depend only on earnings growth, not on prices. It turns out that the initial earnings yields have mattered. Dividend growth has been higher when the initial earnings yield has been higher. The reason appears to have come from secondary effects: payout ratios were too high during the Great Depression and real dividends fell even though nominal dividends grew during the late 1960s and early 1970s.
The worst case conditions occurred when valuations were the highest, as they are now. We are unlikely to see dividend losses today in spite of high initial valuations.
Regardless, we shall design our portfolio assuming that stocks behave as badly today as they did in the earlier periods.
During the Great Depression, real dividend amounts fell by as much as 41% and 42% (four-year averages) with serious damage lasting beyond 20 years. Most of the damage was overcome by year 25. The worst sequences began in 1929-1932.
The story was similar in 1966-1970 but the bottom was only half as bad, down 20%.
I believe that it is sufficient to plan based on an initial loss of 20% that persists.
The current payout ratio is between 40% and 45%. This is well below the 93.7%, 99.8%, 101.1% and 89.0% levels of 1929, 1930, 1931 and 1932. In fact, current payout ratios are well below the 70.6%, 69.7%, 66.3%, 64.0% and 59.9% of 1966, 1967, 1968, 1969 and 1970.
Keep in mind that these numbers are based on the average of a decade of (real) earnings E10, not single-year earnings. By using a decade of smoothed earnings E10, we reduce the likelihood of our being caught by a surprise dividend cut. [Based on single-year earnings, today’s payout ratio is 35%.]
Satisfying the Requirement of 4%
If we could invest at 5% today, we could meet our 4% requirement. That is, if we were able to start withdrawing 5% from dividend yields, we could tolerate a persistent 20% loss and still withdraw 4% of our initial balance.
Based on DVY’s 2.95% current dividend yield, I conclude that we can get 5% (plus inflation) safely enough starting from today.
We have to consume some capital while waiting for favorable valuations.
Dividend investors can expect to get 4.7% to 6.0% in 3 to 7 years. [The dividend yield of the S&P500 index should be 2.7% to 4.0%.] If we could get 6.0%, we would need only 83.33% (which is 5%/6%) of our original balance to produce sufficient income for withdrawals of 5% of the original balance. (This would guarantee 4% under worst case conditions). We could consume the remaining 16.67% of the original balance to cover the period while waiting for the 6.0% dividend yield.
If we start with a pot equal to 16.67% of the initial balance and if we withdraw 4% of the initial balance each year, we remove 24% of the pot each year. Withdrawals from an account with 2% TIPS would last (only a little bit longer than) 4 years, 3 months.
Dividend investors can expect to get 6.9% and 10.4% in 9 years. [The dividend yield of the S&P500 index should be 4.0% to 6.0%.] If we could get 8.0%, we would need only 62.5% (which is 5%/8%) of our original balance to produce an income stream equal to 5% of our original balance. (This would guarantee 4% under worst case conditions). We could consume 37.5% of the original while waiting for dividend yields to rise to 8.0%.
If we start with a pot equal to 37.5% of the initial balance and if we withdraw 4% of the initial balance each year, we remove 10.67% of the pot each year. Withdrawals from an account with 2% TIPS would last (only a few days under) 10 years, 6 months.
On the other hand, if we could get only 6.9%, we would need 72.46% (which is 5.0%/6.9%) of our original balance to produce an income stream equal to 5% of our original balance. (This would guarantee 4% under worst case conditions.) We would be able to consume 27.54% of the original balance while waiting for dividend yields to rise to 6.9%.
If we start with a pot equal to 27.54% of the initial balance and if we withdraw 4% of the initial balance each year, we remove 14.52% of the pot each year. Withdrawals from an account with 2% TIPS would last (a few moments under) 7 years, 6 months.
At this point, developing a strategy looked doable, but marginal.
The Breakthrough
I revisited my method of projecting dividend growth. A 20% loss starting from today’s low payout ratios just didn’t sound reasonable. After all, this is with smoothed earnings, not single-year earnings.
I plotted the total growth in the dividend amount versus the payout ratio (100%*the dividend amount/the trailing ten-year average of earnings). I used 4-year moving averages.
Here are the equations with y = the 4-year moving average of the growth of the dividend amount and x = the payout ratio based on smoothed earnings E10. The first year of dividends occur in Year 1 (as opposed to Year 0).
At year 4:
y = -0.3162x+23.799 plus 30% and minus 20%.
R-squared = 0.153.
At year 8:
y = -0.9914x+79.81 plus 60% and minus 30%.
R-squared = 0.2361.
At year 12:
y = -1.0328x+95.556 plus 100% and minus 50%.
R-squared = 0.1549.
Here are the projections:
At year 4:
With a payout ratio of 40%, the dividend growth is 11% (most likely) with a range of –9% to 41%.
With a payout ratio of 60%, the dividend growth is 5% (most likely) with a range of –15% to 35%.
At year 8:
With a payout ratio of 40%, the dividend growth is 40% (most likely) with a range of 10% to 100%.
With a payout ratio of 60%, the dividend growth is 20% (most likely) with a range of –10% to 80%.
At year 12:
With a payout ratio of 40%, the dividend growth is 54% (most likely) with a range of 4% to 154%.
With a payout ratio of 60%, the dividend growth is 34% (most likely) with a range of –16% to 134%.
At today’s 40% to 45% payout ratio:
At year 4, the dividend growth will be 11% to 10% (most likely) with a range of –9% or –10% to 40% or 41%.
At year 8, the dividend growth will be 40% to 35% (most likely) with a range of 5% or 10% to 95% or 100%.
At year 12, the dividend growth will be 54% to 49% (most likely) with a range of -1% or 4% to 149% or 154%.
Starting at today’s dividend payout ratios, the worst case dividend loss is 10% at Year 4 with full recovery by Year 8. Today’s dividends are secure (relative to those of the past).
Revised Calculations
If we could invest at 4.5% today, we could meet our 4% requirement. That is, if we were able to start withdrawing 4.5% from dividend yields, we could tolerate a 10% loss and still withdraw 4% of our initial balance. We only have to protect against four years of losses.
Dividend investors can expect to get 6.9% and 10.4% in 9 years. [The dividend yield of the S&P500 index should be 4.0% to 6.0%.]
The 9-year safe withdrawal rate of 2% TIPS is 12.25%. This is 3.0625 times as large as 4.0%. Our 4% withdrawals would consume only 32.7% of an original portfolio at Year 9. This would leave us with 67.3% of the original portfolio. We would need a dividend yield of 5.94% (or 4%/0.673) to meet our requirement. Protecting against a 10% loss in the first four years brings this up to 6.6% (or 5.94%/0.9 or, alternatively, 4.5%/0.673 = 6.7%, the difference being a round off error).
A Middle Ground
Suppose that we split our portfolio into two parts: one-half in stocks yielding 3% (similar to DVY) and the other half in a 2% TIPS ladder.
The stock portfolio would yield dividends equal to 1.5% (one-half of 3%) of the initial balance. The TIPS ladder would need to produce an income stream of 5.0% to bring the total income up to 4.0% of the initial balance. (One-half of 5.0% is 2.5% of the initial balance. The total income is 1.5% from dividends and 2.5% from TIPS.)
A 2% TIPS ladder can produce 5.0% (plus inflation) for 25.5 years. At year 10, the TIPS ladder still has 67% of its original principal when withdrawing 5.0%. The portion remaining is 33.5% of the original balance (since TIPS started at one-half of the original balance).
We should be able to reinvest this to get a dividend yield between 6.9% and 10.4%. [The dividend yield of the S&P500 index would be 4.0% to 6.0%.] In terms of the original balance, the income stream would be between 2.3% (which is 0.335 times 6.9%) and 3.5% (which is 0.335 times 10.4%).
Throughout the first ten years, the dividend investor’s downside risk is 5% (i.e., from a withdrawal rate of 4.0% of his original balance plus inflation to 3.8% plus inflation) for 4 years. This assumes a 10% dividend cut from DVY (or its equivalent). After 4 years, the DVY (equivalent) income stream should have grown at least as fast as inflation. Most likely, his DVY (equivalent) income would have increased from 3.0% to 3.3% of his original DVY (equivalent purchase) price at year 4 and to 4.2% of its original DVY (equivalent purchase) price at year 8.
There is a small risk that the best yield that you can get by year 10 is 6.9%. If so, the income stream from your new stocks would be 2.3% of your portfolio’s original balance (worse case). The income stream from the DVY stocks would still be 1.5% or more (since the initial danger period has expired). The worst-case, minimum total income would be 2.3% from the new stocks and 1.5%, which totals 3.8%.
Once again, the downside risk is 5% of your 4.0% (plus inflation) withdrawal rate, which is 3.8%, over four years.
[In theory, the 6.9% income stream could itself fall by 10%. But this is assuming worst case dividend growth following worst-case dividend yields until year 10. Even then, the damage would result in a 3.6% withdrawal rate for four years, followed by a safe withdrawal rate of 4.0%+ continuing well into the foreseeable future.]
Conclusions
I met my goals.
This approach delivers 4.0% (plus inflation) far into the future. The downside risk is a 4-year reduction of 5%, which would be a withdrawal rate of 3.8% (plus inflation), followed by 4.0% or more (plus inflation).
Have fun.
John Walter Russell
January 14, 2006