Joined: 26 Nov 2002
Location: Crestview, Florida
|Posted: Mon Mar 29, 2004 5:48 am Post subject: Intuitive TIPS Blend B
|I have looked into having two independent portfolios several times. The final outcome has always been the same. This is from a different angle.
Overview and Summary
I looked at the combination of two portfolios. The first consisted only of stocks. The second of stocks and TIPS with allocations that varied with P/E10. Withdrawals were proportioned according to each portfolio's relative size. For the most part, the two portfolios were allowed to grow independently.
This has the effect of varying the overall stock allocation of the combined portfolio over time.
I included negative stock allocations in the first portfolio. In most cases, this was offset by a positive stock allocation in the second portfolio. You would not actually short stocks in the first portfolio. Instead, you would adjust the stock allocation of the overall portfolio in a complex manner.
I refer to this combination of an all-stock portfolio with a mixed portfolio as a blend. When the component other that stocks consists of TIPS, I refer to this as a TIPS blend.
This type of blend failed to improve performance.
These results favor using a standard combination of stocks and TIPS (but with allocations that vary with P/E10).
The all-stock component was a buy-and-hold allocation. Although it is possible that simple re-balancing of the two portfolios would result in a slight improvement, a better choice is likely to be much more complex. We can already look at withdrawals that are based upon both the initial balance and the current balance in any combination. It may be that something based on this plus something else similar to Gummy's Sensible Withdrawal Rate approach would be productive. With Gummy's approach, each year's (real) withdrawal amount is capped.
Portfolio A was 100% stocks.
Portfolio B was a mixture of stocks and TIPS. Its stock allocation was switched according to P/E10. Initially, the allocations were 100%-30%-0%. The thresholds were P/E10 equal to 11 and 24. The TIPS (real) interest rate was 2%. This portfolio has a Historical Database Rate of 5.2% when used by itself.
The JanSz-Chips Deluxe V2.0A was set up without using its unique features. The capital gains percentage was set to 0% (JanSz feature). Dividend reinvestments were set at 100% (Chips feature).
The original portfolio balance was $100000 to minimize the effects of round off errors. Expenses were set to 0.20%. The TIPS interest rate was set at 2.0%. Switching was selected. I examined the 30-year survivability of portfolios beginning in 1921 through 1980.
Here is what was different with this setup. Stock allocation percentages other than zero were used. Re-balancing was not used. [With switching, it is normal to put 0% into the "stock allocation" cell. Actual allocations are varied in the switching section.]
In effect, Portfolios A and B were allowed to grow independently except that they contributed proportionately to cover withdrawals and expenses. Portfolio B switched stock and TIPS allocations depending upon P/E10. This re-balanced Portfolio B internally.
I varied the percentage of stock (in $B$6) to change the relative size of Portfolio A. $B$16 was set at 2 (no re-balancing). The withdrawal rate was set at 5.3%. Portfolio B had its initial 100%-30%-0% allocations and its initial P/E10 thresholds of 11 and 24.
In all cases, increasing the size of Portfolio A made things worse. There were 4 failures without Portfolio A. There were 5 failures when Portfolio A was 10% of the combination. There were 7 failures when Portfolio A was 20% of the combination. This continued. When Portfolio A made up the entire (100%) combination, there were 18 failures.
I also looked at these conditions with annual re-balancing (i.e., with $B$16 set to 1). The results were the same for the reported allocations, but there were some minor differences at intermediate conditions. As before, increasing the stock allocation made things worse.
Varying the Upper Threshold
I set the stock percentage to 30%.
The calculator was returned to not re-balancing ($B$16 set at 2). The withdrawal rate remained at 5.3%. The lower P/E10 threshold was 11. The allocations remained at 100%-30%-0%.
I varied the upper P/E10 threshold.
The best result was when the upper threshold was 24. There were 7 failures.
There were 9 failures when the upper threshold was 11 (i.e., when there was only a single threshold and allocations of 100% and 0%). There were 8 failures when the upper threshold was 18 or 21. Otherwise, there were either 10 or 11 failures.
Varying the Lower Threshold
I kept the upper P/E10 threshold at 24. I kept the stock percentage at 30%.
I set the calculator to no re-balancing ($B$16 set at 2). I kept the withdrawal rate at 5.3%. The allocations remained at 100%-30%-0%.
I varied the lower P/E10 threshold.
The best threshold setting was 11. With it, there were 7 failures.
When the lower P/E10 threshold was 9 or 10, there were 9 failures. When the lower P/E10 threshold was 12 or 13, there were 11 failures. Other threshold choices were worse.
I varied the allocation of the all-stock portfolio (i.e., Portfolio A as a percentage of the overall, combined portfolio). I also varied the intermediate stock allocation of Portfolio B (which has stocks and TIPS).
I kept the withdrawal rate at 5.3%. I set the lower P/E10 threshold at 11. I set the upper P/E10 threshold at 24. Below the lower P/E10 threshold, the Portfolio B stock allocation was 100% (i.e., all stocks). Above the upper P/E10 threshold, the Portfolio B stock allocation was 0% (i.e., all TIPS). I set $B$16 to 2 (i.e., no re-balancing between Portfolio A and Portfolio B).
An intermediate stock allocation of Portfolio B of 30% was always at least as good as anything else with a specified allocation for Portfolio A. With a Portfolio B stock allocation of 30%, the number of failures were 4, 5, 7, 7, 7 and 10 for Portfolio A allocations of 0%, 10%, 20%, 30%, 40% and 50%, respectively.
I was able to improve results by making the Portfolio A allocation negative. However, this would have been dangerous because Portfolio B had a stock allocation of 0% whenever P/E10 exceeded the upper threshold.
The best allocations with a net short had (a) an intermediate Portfolio B allocation of 40% with a Portfolio A allocation of -20% (i.e., shorting stocks by 20%) and (b) an intermediate Portfolio B allocation of 50% with a Portfolio A allocation of -30%. Both were able to bring the number of failures at a 5.3% withdrawal rate down to zero. Neither was able to do so at 5.4%.
Negative and Positive Allocations
I examined a variety of conditions with a Portfolio A allocation of -20% but with a Portfolio B stock allocation of +20% when P/E10 was above the upper threshold. (This results in a slightly positive overall stock allocation because size of Portfolio B is increased to +120%. That is, Portfolio A + Portfolio B = 100%.)
I kept the withdrawal rate equal to 5.3%.
I varied the upper P/E10 threshold from 18 through 24. I varied the intermediate threshold of Portfolio B from 20% through 60%.
There was only one condition that did at least as well as the baseline of 4 failures with a 0% allocation for the all-stock portfolio (Portfolio A). Actually, it did better, but only by the smallest of margins possible. With a -20% Portfolio A allocation, a Portfolio B intermediate stock allocation of 50% and an upper P/E10 threshold of 21, the number of failures was 3.
I repeated this process with a stock allocation for Portfolio B of +30% when P/E10 exceeded the upper threshold. I set the Portfolio A allocation to -30%.
The baseline was 4 failures with a Portfolio A stock allocation of 0%. It was matched only when the upper P/E10 threshold was 21. The Portfolio A stock allocation was -30% and the Portfolio B intermediate stock allocations were 50% and 60%. It was never bettered.
In order to show an advantage, it is generally necessary to have a net negative stock allocation. That is, Portfolio A has a negative magnitude (or absolute value) that is larger than the stock allocation in Portfolio B when P/E10 valuations are above the upper threshold.
In only three instances was it sufficient to have a stock allocation that balanced the negative Portfolio A allocation with an equal Portfolio B allocation when above the upper P/E10 threshold. All of them had an upper P/E10 threshold of 21 and only one was better than the baseline. The baseline had a Portfolio A allocation of 0%.
All of this suggests that it is best not to have a separate, stock-only portfolio.
It is likely that the single, better result was a statistical fluke.
This particular approach, which I call the TIPS blend, did not improve Historical Database Rates. There was an isolated exception.