Notes starting October 21, 2006
Notes updated on November 22, 2006.
Earnings Growth Adjustments
I posted this at the Morningstar Income & Dividend Investing board.
I have looked at making adjustments to my calculations by introducing an earnings term or an earnings growth term.
I started out by making a prediction based on the initial P/E10. I compared it to what has happened historically.
That is: Residual = Actual Historical Results (returns or Historical Surviving Withdrawal Rates) - predicted Results based on P/E10.
I have looked at average earnings over 1, 2, 3, 4 and 5 years. I have looked at scaled earnings E3/E5, delta E3/E5 and delta (E3 or E5)/E10. Normalized earnings growth of the form [Ex(t)-Ex(t-lag)]/[Ex(t)+Ex(t-lag)].
All of my results so far have been negative.
[I examined years 1923-1980. There is a consistent negative slope associated with using earnings Ex directly. I associate this with the calendar more than anything else. The relationship is weak. In addition, this same data may be showing different effects for Bull Markets and Bear Markets.]
Morningstar Web Site
Capitalization Weighting of What?
This exposes the underlying fallacy related to Sharpe’s Theorem that capitalization weighting matches the return of the market as a whole. It incorporates Dr. Hussman’s observation that money can never enter nor leave the market as a whole.
What has been missing? What is the critical oversight? Dividends.
Capitalization Weighting of What?
More about Earnings Growth Adjustments
I have continued to look at earnings growth adjustments for my calculations. I have continued to compare predictions based on the initial P/E10 with what has happened historically. That is: Residual = Actual Historical Results (returns or Historical Surviving Withdrawal Rates) - predicted Results based on P/E10.
This time, I looked at percentage earnings yields 100Ex/P over 1, 2, 3, 4, 5 and 10 years. I looked normalized changes in the percentage earnings yields delta 100Ex/P with x=1, 2, 3, 4 and 5. These normalized percentages have the form [100Ex/P(t)-100Ex/P(t-lag)] / [100Ex/P(t)+100Ex/P(t-lag)]. In this case, lags are chosen so that the earliest term occurs ten years earlier than the latest. That is, when using x=1, the lag is 10 years. When x=2, the lag is 9 years. And so forth.
My results are negative.
R-squared was less than 0.10 except for 100E1/P versus residual returns at Year 10, 100E1/P versus residual returns at Year 15, 100E1/P versus HSWR80T2 residuals. None of the normalized changes in percentage earnings yields delta 100Ex/P resulted in an R-squared level as high as 0.10.
In a few instances, I calculated a second order polynomial (quadratic).
With the Year 5 residuals, a second order polynomial with 100E2/P had an R-squared value of 0.0994. The implication is that there spread in residual returns is a little bit larger when the percentage earnings yield is very high. The returns are outstanding when P/E10 is low. But my calculations are a little bit less accurate.
With Year 10 residuals, the first order 100E1/P equation has an R-squared value of 0.1099. The second order 100E1/P equation has an R-squared value of 0.1237. The slopes are positive. The actual returns are likely to be a little bit higher than normally calculated using P/E10 when the single-year P/E1 is unusually low (and 100E1/P is unusually high).
That is, if this year’s P/E1 is lower than normal, it is likely to show up at Year 10 as a slight benefit relative to P/E10 based predictions.
With Year 15 residuals, the (first order) 100E1/P equation has an R-squared value of 0.125. The slope is positive. If this year’s P/E1 is lower than normal, it is likely to show up at Year 15 as a slight benefit relative to P/E10 based predictions.
With HSWR80T2 residuals (in 30-year Historical Surviving Withdrawal Rates), the first order 100E1/P equation has an R-squared value of 0.1058. The second order 100E1/P equation has an R-squared value of 0.252. The slopes are positive. The actual returns are likely to be a higher than normally calculated using P/E10 when the single-year P/E1 is unusually low (and 100E1/P is unusually high). However, the second order term indicates an increase spread in Historical Surviving Withdrawal Rate predictions. The calculated rates are less certain when the single-year P/E1 is very low (and 100E1/P is unusually high). The odds favor a boost in the Safe Withdrawal Rate as compared to using a P/E10 calculation by itself, but without offering anything close to a guarantee.
Mindless Comparisons to Index Funds
How often have you read that actively managed funds underperform a suitably chosen index? It is a lie.
Mindless Comparisons to Index Funds
New Standards for Financial Reporting
During my professional career, I was never allowed to get away without doing the following. Why do we allow "financial experts" to get away with anything less?
This is based on my “It’s about time...” series of notes.
I am posting this note for a second time because of its importance.
New Standards for Financial Reporting
System Engineering
Most of my professional work was in system engineering. You can see evidence of this throughout this site.
System Engineering
Today’s Skill
Today’s skill is intermediate-term market timing.
Today’s Skill
P/E10 Updates
Professor Robert Shiller has stopped updating the S&P500 spreadsheets at his main web site.
Professor Shiller has opened up a new site to advertise the updated version of Irrational Exuberance. He keeps the Excel spreadsheets at this new site up to date.
NOTE: the Stock Returns Predictor shows the current relationship between P/E10 and S&P500. I keep it updated.
The earnings term E10 grows with time. P is the current level of the S&P500 index. You must press the CALCULATE button after adjusting the slider.
Professor Shiller's Web Site
Professor Shiller’s Irrational Exuberance Web Site
Short Term Price Fluctuations
I recently realized that I can say a lot more about short-term price fluctuations than I had previously thought. Here it is.
Short Term Price Fluctuations
I have constructed more tables that tell us about short-term price fluctuations. Rob Bennett has written a Letter to the Editor that puts everything into context.
More about Short Term Price Fluctuations
November 7, 2006 Letters to the Editor
I have constructed tables of conditional probabilities. They may be easier to understand. They show what happened to prices when P/E10 fell within specified ranges during the years of 1921 through 1980.
Short Term Price Fluctuations (Continued)
Simplified Retirement Trainer with Dividends A
I have added dividends to the Simplified Retirement Trainer A.
Simplified Retirement Trainer with Dividends A
Late Letters
Oops! I was late in connecting my November 7, 2006 Letters to the Editor to the Letters button on the side of the page. Enjoy three recent letters.
November 7, 2006 Letters to the Editor
Basic Estimates and Refinements
I talk about refinements separately from estimates. This example may help you understand the distinction.
Basic Estimates and Refinements
Earlier Notes
Visit Notes Starting July 10, 2006 for links to all earlier notes:
Notes through August 21, 2005
Notes through November 29, 2005
Notes through January 13, 2006
Notes Starting from January 14, 2006
Notes through April 18, 2006
Notes through June 12, 2006
Notes starting June 13, 2006
Notes Starting July 10, 2006
Notes Starting July 10, 2006 covered the following topics:
It is about time..., Time and the Gordon Model, It is about time...continued, It is about time...more, The Copie Index, It is about time...number six, Compact Variable Terminal Value Rate Calculators, Orders of Magnitude, Using Stock Return Predictions, Eye Opening Calculations with Compact CVTVR L, New Standards for Financial Reporting, A Tip about my Yahoo Briefcase, Rational Pessimism and Tobin’s q, Super Variable Terminal Value Rate SVTVR Calculators, What does “3% + inflation” mean?, Tobin q Survey, Turning Points, A Helpful Theorem, Year 15 Calculator A, Designing a 45-Year Retirement, Retirement Trainer.
Notes Starting July 10, 2006
Notes starting August 25, 2006 covered the following topics:
Retirement Trainer, P/E10 Predictions, Bulls, Bears and P/E10 Predictions, P/E10 Predictions Revisited, Great Article, Playing with the Toy, More Fun with the Toy, Why Dividends Are Better, Improving the Retirement Trainer, Great Fun with the Improved Retirement Trainer, Accumulation and the Retirement Trainer, Dividends versus Capital Appreciation, E10 or D10?, Type 2A Bull Bear Retirement Trainer, Simplified Retirement Trainer A, More PE10 Predictions.
Notes starting August 25, 2006
Notes starting September 16, 2006 covered the following topics:
My Retirement Trainers Work!, Demonstration a1 to a8, Retirement Planning Insights, Retirement Trainers and Accumulation, Learning the RIGHT Lessons, The Wrong Lessons, Denial Is Expensive, My Yahoo Briefcase, Dollar Cost Averaging at Year 15, The Next Recession, Interesting Web Site, Market Timing--What Works and What Doesn't, I Saw My Doctor Today, Capitalization Weighted Stock-Bond Allocations, Explosive Earnings Growth, More about Earnings Growth, Why Is Today's Investing Advice So Poor?, Another Note about Earnings Growth.
Notes starting September 16, 2006