Current Research

Here is an index with links to my previous Current Research efforts.

Current Research Index

Current Research G: Dow Jones Utiltities

Unclemick sent a Letter to the Editor requesting that I use the Dow Jones Utilities Index in my research.

His suggestion has already led me to several advances regarding dividend-based strategies.

Here is a link to the Letter to the Editor that started this line of investigation.

January 11, 2006 Letters to the Editor

Here are links to a few of these earlier articles.

Dividend-Based Design Example
Dividend-Based Design Outline
Dividend Sound Bite
Lessons from the Dow Jones Utilities

I have built the Unclemick 01 Calculator. I expect to use it to gain additional insights. Refer to the SWR Calculators section for details.

SWAT2 Equivalent

I have made a brief survey using the equivalent of SWAT2, replacing the S&P500 with the Dow Jones Utilities.

I used my Unclemick 01 Calculator (i.e., the Deluxe Calculator V1.1A08aUM01). I varied stock and TIPS allocations between 25% and 75%, depending upon the P/E10 level of the S&P500. I used 2% TIPS. I set expenses equal to 0.20%. I adjusted withdrawals to match inflation according to the CPI. I rebalanced portfolios annually.

I examined 30-year Historical Surviving Withdrawal Rates.

I increased withdrawal amounts in increments of 0.1%. I determined the lowest withdrawal rates at which one, six and twelve failures occurred. I restricted my investigation to historical sequences beginning in 1929-1980.

For purposes of comparisons, I also examined what would have happened with the SWAT2 portfolio when using February S&P500 data. We would expect this baseline portfolio to be very similar to SWAT2, but with slightly imperfect P/E10 timing. This baseline portfolio uses the same sequences (1929-1980) as the new portfolios. It includes my latest update of S&P500 data.

Optimized Values

With the Dow Jones Utilities, I ended up selecting allocations of 75%-40%-25% with P/E10 thresholds of 12 and 20.

SWAT2 has allocations of 75%-40%-25% with P/E10 thresholds of 11 and 21.

With the Dow Jones Utilities (and 2% TIPS), the withdrawal rates were:
First failure: 4.5%.
Sixth failure: 4.8%.
Twelfth failure: 5.3%.

With the SWAT2 (with the S&P500 and 2% TIPS) with February data, the withdrawal rates were:
First failure: 4.7%.
Sixth failure: 5.0%.
Twelfth failure: 5.3%.

Observations

The best allocations and thresholds are very close. This is the kind of thing that we can expect with other switching portfolios as we look into the future. Results should be similar, but not identical, to those extracted from the past.

Using P/E10 from the S&P500 for switching works reasonably well with the Dow Jones Utilities. We have seen similar behavior when using individual segments of the market (i.e., Gummy slices). This confirms the notion that you can measure the valuation of the overall market. It makes sense to talk about stocks in general.

The results are very close.

You can learn about the sensitivity of switching algorithms by comparing the Dow Jones Utilities results with the SWAT2 (S&P500) results. They are not critically sensitive. They tolerate small differences.

Region of Optimal Performance

The region of optimal performance is very broad. All of the following Dow Jones Utilities conditions perform virtually the same as the optimized portfolio.

Middle allocation: 30% or 40%.
Lower P/E10 Threshold: 12 or 13.
Upper P/E10 Threshold: 12, 13, 14, 18, 19 or 20.

General Optimization

In my original optimization I used Benjamin Graham’s constraint: stock and bond allocations between 25% and 75%. This led to the following tradeoff region:

DJU and 2% TIPS.
75%-[30% or 40%]-25%.
P/E10 thresholds: [12 or 13] and [12-14 or 18-20].

I selected the following condition:

DJU and 2% TIPS.
75%-40%-25%.
P/E10 thresholds: 12 and 20.

This resulted in the following performance:

First failure: 4.5%.
Sixth failure: 4.8%.
Twelfth failure: 5.3%.

My Latest Optimization

In my latest optimization, I allowed the full choice of four thresholds and three intermediate allocations. The allocation below the lowest threshold is always 100% stocks. The allocation above the highest threshold is 0% stocks. My latest optimization led to the following conditions:

DJU and 2% TIPS.
100%-100%-30%-0%-0%.
P/E10 thresholds: [12 or 13]-[12 or 13]-20-20.

It has the following performance:

First failure: 5.0%.
Sixth failure: 5.3%.
Twelfth failure: 5.5%.

An alternative choice would have been:

DJU and 2% TIPS.
100%-100%-40%-0%-0%.
P/E10 thresholds: [12 or 13]-[12 or 13]-20-20.

This has the following performance:

First failure: 4.9%.
Sixth failure: 5.3%.
Twelfth failure: 5.7%.

Observations

Notice that the optimal conditions are almost identical. In spite of having a choice of three intermediate allocations, the optimal conditions have only a single intermediate allocation. The optimal allocations are 100%-[30% or 40%]-0%. The P/E10 thresholds are [12 or 13]-[12 or 13]-20-20. That is, the allocation is 100% whenever P/E10 is below 12 or 13 and 0% whenever P/E10 is above 20. In between, the best allocation is either 30% or 40%.

The optimal portfolio choices changed only by replacing Benjamin Graham’s constraint of 25% and 75% with 0% and 100%.

Removing Benjamin Graham’s constraint improved the withdrawal rate at first failure by 0.4% to 0.5%. It improved the withdrawal rate at the sixth failure by 0.5%. It improved the withdrawal rate at the twelfth failure by 0.2% to 0.4%.

Have fun.

John Walter Russell
February 1, 2006