Research Highlights

This is a research web site. I started this site in 2005. My research extends back more than 5 years. My articles appear in the sequence written, which is not always the best for presentation. Here is an overview of some of my more important findings.

1. My first effort was to determine Price Adjusted Safe Withdrawal Rates. It resulted in many articles, looking at the issue from a variety of vantage points. It has culminated in the Year 30 SWR Retirement Risk Evaluator, the Year 15 SWR Retirement Risk Evaluator and the Stock Return Predictor as well. [Rob Bennett and I built these as a joint effort.] If you know the current level of the S&P500 index, you can calculate Safe Withdrawal Rates with any ending balance percentage that you want. You can see a full presentation of the probabilities associated with different withdrawal rates. You can adjust your investment allocations between the S&P500 and TIPS. The Stock Returns Predictor shows what is likely in ten year increments up to year 60.

2. During my early investigations, I found that Professor Robert Shiller’s P/E10 was the best measure of valuations. More precisely, it is the inverse, the percentage earnings yield 100E10/P, that works best. The Gordon Model and other forms of the Dividend Discount Model support this finding from a theoretical standpoint. I found that individual segments of the S&P500 index respond as well. P/E10 is adequate for them as well.

3. I established a tough to beat baseline early in my research. TIPS ladders do the job. You can see how powerful they are by looking at the TIPS Table.

4. I found repeatedly that rebalancing is a bad idea as long as you are able to measure valuations. The drag from inferior investments does little to compensate for giving away the upside from your winners. The special case in which rebalancing provides protection is when stock valuations are especially high, as they are today. Even then, reducing your stock allocation works better.

5. I tried changing allocations according to valuations. I call this switching. It was successful, significantly improving Safe Withdrawal Rates. I looked at a wide variety of algorithms. Later, I found that training could produce even better outcomes. This led to the Scenario Surfer. It is an advanced Monte Carlo simulator that uses two forms of Mean Reversion.

6. I examined strategies that take advantage of intermediate timing based on valuations (delayed purchases). This shows great promise. The key is to recognize the risk that more favorable valuations may fail to appear. Although unlikely to take so long, we can wait up to 20 years before an intermediate timing strategy fails.

7. I extended the understanding of the Gordon Model. I found that it is most accurate for time frames between 5 and 15 years and best at 10.

8. In my research of the Sortino Ratio, I found that using monthly data strips off the effect of valuations beyond one or two months. I found that using annual data strips off the effect of valuations beyond one or two years. This explains why so much academic research has failed to address valuations properly. Easily available monthly data gives many data points, but little in the way of long-term information.

9. I found that dividend strategies make a lot of sense. I found that managing the income stream works wonders. A dividend blend easily raises the continuing withdrawal rate above 5%, far above the 4% claimed (incorrectly) to last 30 years in early traditional Safe Withdrawal Rate studies. A dividend blend takes excess cash generated by a high yield, lower growth investment to cover the middle years while an initially lower yielding, high dividend growth investment catches up to fill in the later years. I have built a variety of spreadsheets to help.

Have fun.

John Walter Russell
January 19, 2008