stock market return projections
Posted: Tue Feb 11, 2003 6:27 pm
stock market return projection
I have separated this post from the Mean Reversion Equivalence thread. Many relevant observations and comments started there. But they digressed from the original purpose of that thread, which was to praise raddr's Monte Carlo model. This discussion is both simple and complex.
A. Overview
This excursion started when I objected to using the Gordon Model to project stock market returns. The required assumptions do not apply in today's market. The key issue is this: the stock market payout ratio has decreased in recent years. This has accelerated earnings growth. I prefer to go back to the Dividend Discount Model and make adjustments to it directly.
First, I confirmed BenSolar's findings. Then I projected the growth rate of the overall stock market.
B. Details
My original observation was based on an eyeball evaluation of graphs of the real earnings amount and the real dividend amounts from Professor Shiller's data. There seems to be a divergence around mid-1948. I looked at historical point values (single year amounts) for the ratio of real dividends to real earnings or payout ratio, the ratio of the real dividend amount to the price (S&P 500 index value) or dividend yield and the inverse of P/E10. I looked at numbers for January of the years 1881, 1901, 1921, 1941, 1961, 1981 and 2001. I chose to start with 1881 since it had the first entry for PE10.
The payout ratios were stable. They were 54.5%, 62.6%, 66.8%, 63.9%, 60.6%, 42.1% and 33.4%. Roughly speaking, today's payout ratio is one half of the older values.
Dividend yields were 4.28%, 4.27%, 7.11%, 6.38%, 3.26%, 4.66% and 1.21%. The values of the inverse of PE10 were 5.41%, 4.78%, 19.52%, 7.19%, 5.41%, 10.80% and 2.70%. If numbers such as 19.52% and 10.80% concern you, remember that they correspond to 1921 and 1981. Both were great years for making investments.
Comparing dividend yields to the values of the inverse of PE10, the ratios fluctuate greatly. They are 79%, 89%, 36%, 89%, 69%, 43% and 45%. The pre-1960 dividend yields were roughly 80% of the inverse of PE10. The number 36% corresponded to 1921 and it represented a rare buying opportunity.
C. Projections
I have made two projections.
a) The first is to double today's dividend yields to match historical payout ratios. I then adjust prices downward to two thirds of Professor Shiller's January 2001 numbers since the S&P 500 has fallen about that far. The S&P 500 dividend yield on January 2001 was 1.21%. Scaling (multiplying by 2 and again by 3/2), the equivalent historical dividend yield is 1.21%*2*1.5 = 3.63%.
For the growth term, I use the 1.1% historical long term growth rate of dividends, not earnings. The faster rate of earnings growth is already accounted for in the decrease in the payout ratio. This faster rate is the effect of retaining a higher percentage of earnings. We have already included a scale factor to adjust for the change of the payout ratio. Dividend growth is the proper choice.
This results in a projection of the overall stock market return of 3.63% + 1.1% = 4.74% (after adjusting for inflation).
b) The alternative approximation is to take 80% of the inverse of PE10 to estimate the dividend yield. In January 2001 this was 2.70%. I then scale by the same 3/2 factor to reflect the decrease in the (index value or) price of the S&P 500 in today's market. This produces an equivalent historical dividend yield of 3.24%. I add the 1.1% historical long term growth rate of dividends. My projection becomes 3.24% +1.1% = 4.35%.
I prefer the second projection. At first blush, its numbers are based on components that seem to fluctuate too much to be of value. But it is only the price component of PE10 that fluctuates rapidly. The earnings component is a ten year average. In addition, the fluctuations are in the right direction. They tell you that stocks do better if you buy when prices are low.
D. Comments
My projections are 4.35% and 4.74% real growth overall. This is will support a withdrawal rate of 4% provided that you can overcome the problem of volatility. These numbers are 1.76% and 2.15% below the historical 6.5%.
These projections are likely to convey accurate information since they are based on crude approximations. It is obvious when these approximations fail. This is much better than having precise calculations that hide a subtle error.
Have fun.
John R.
I have separated this post from the Mean Reversion Equivalence thread. Many relevant observations and comments started there. But they digressed from the original purpose of that thread, which was to praise raddr's Monte Carlo model. This discussion is both simple and complex.
A. Overview
This excursion started when I objected to using the Gordon Model to project stock market returns. The required assumptions do not apply in today's market. The key issue is this: the stock market payout ratio has decreased in recent years. This has accelerated earnings growth. I prefer to go back to the Dividend Discount Model and make adjustments to it directly.
First, I confirmed BenSolar's findings. Then I projected the growth rate of the overall stock market.
B. Details
My original observation was based on an eyeball evaluation of graphs of the real earnings amount and the real dividend amounts from Professor Shiller's data. There seems to be a divergence around mid-1948. I looked at historical point values (single year amounts) for the ratio of real dividends to real earnings or payout ratio, the ratio of the real dividend amount to the price (S&P 500 index value) or dividend yield and the inverse of P/E10. I looked at numbers for January of the years 1881, 1901, 1921, 1941, 1961, 1981 and 2001. I chose to start with 1881 since it had the first entry for PE10.
The payout ratios were stable. They were 54.5%, 62.6%, 66.8%, 63.9%, 60.6%, 42.1% and 33.4%. Roughly speaking, today's payout ratio is one half of the older values.
Dividend yields were 4.28%, 4.27%, 7.11%, 6.38%, 3.26%, 4.66% and 1.21%. The values of the inverse of PE10 were 5.41%, 4.78%, 19.52%, 7.19%, 5.41%, 10.80% and 2.70%. If numbers such as 19.52% and 10.80% concern you, remember that they correspond to 1921 and 1981. Both were great years for making investments.
Comparing dividend yields to the values of the inverse of PE10, the ratios fluctuate greatly. They are 79%, 89%, 36%, 89%, 69%, 43% and 45%. The pre-1960 dividend yields were roughly 80% of the inverse of PE10. The number 36% corresponded to 1921 and it represented a rare buying opportunity.
C. Projections
I have made two projections.
a) The first is to double today's dividend yields to match historical payout ratios. I then adjust prices downward to two thirds of Professor Shiller's January 2001 numbers since the S&P 500 has fallen about that far. The S&P 500 dividend yield on January 2001 was 1.21%. Scaling (multiplying by 2 and again by 3/2), the equivalent historical dividend yield is 1.21%*2*1.5 = 3.63%.
For the growth term, I use the 1.1% historical long term growth rate of dividends, not earnings. The faster rate of earnings growth is already accounted for in the decrease in the payout ratio. This faster rate is the effect of retaining a higher percentage of earnings. We have already included a scale factor to adjust for the change of the payout ratio. Dividend growth is the proper choice.
This results in a projection of the overall stock market return of 3.63% + 1.1% = 4.74% (after adjusting for inflation).
b) The alternative approximation is to take 80% of the inverse of PE10 to estimate the dividend yield. In January 2001 this was 2.70%. I then scale by the same 3/2 factor to reflect the decrease in the (index value or) price of the S&P 500 in today's market. This produces an equivalent historical dividend yield of 3.24%. I add the 1.1% historical long term growth rate of dividends. My projection becomes 3.24% +1.1% = 4.35%.
I prefer the second projection. At first blush, its numbers are based on components that seem to fluctuate too much to be of value. But it is only the price component of PE10 that fluctuates rapidly. The earnings component is a ten year average. In addition, the fluctuations are in the right direction. They tell you that stocks do better if you buy when prices are low.
D. Comments
My projections are 4.35% and 4.74% real growth overall. This is will support a withdrawal rate of 4% provided that you can overcome the problem of volatility. These numbers are 1.76% and 2.15% below the historical 6.5%.
These projections are likely to convey accurate information since they are based on crude approximations. It is obvious when these approximations fail. This is much better than having precise calculations that hide a subtle error.
Have fun.
John R.