Switching Survey

Research on Safe Withdrawal Rates

Moderator: hocus2004

Post Reply
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Switching Survey

Post by JWR1945 »

Switching Survey

I have made a brief survey using two-level three-allocation switching to determine its potential.

Switching increases the 30-year Historical Database Rate from 3.9% to 5.0% for a portfolio of stocks and commercial paper.

Details

I collected this data using my full-up modified version of the Retire Early Safe Withdrawal Calculator (version 1.61). I refer to it as the JanSz-Chips Deluxe V1.0A version. This could have been done on any of my modified versions that include two-levels of P/E10 switching thresholds with three-allocations of stocks and commercial paper.

I also made extensive use of some of my early data summary improvements.

I limited my examination to retirement portfolios begun from 1921 to 1980.

I set the initial balance at $100000 to minimize the effects of round-off errors. I set the expenses to 0.20% of the portfolio balance. I did not use my special JanSz-Chips modifications. That is, I set the capital gains percentage at 0% and the dividend reinvestments at 100%. I examined portfolio life times of 30 years.

I made my allocation choices abrupt. The low threshold allocation was 100% stocks. The middle allocation was 50% stocks and 50% commercial paper. The high threshold allocation was 0% stocks.

I used an 80% stock / 20% commercial paper allocation to establish a baseline without switching.

The Surveys

For my initial survey I set the withdrawal rate to 5.0% of the initial balance (plus adjustments to match the CPI). Taking advantage of Mike's earlier findings, I set the high threshold equal to 20 and varied the low threshold from 5 through 20. (When the high threshold equals the low threshold, there is only one overall threshold. In this part of the survey, that single threshold was at P/E10 = 20.)

Code:

Code: Select all

P/E10   Failures
 5    17
 6    17
 7    17
 8    16
 9     6
10    6
11    2
12    3
13    3
14   15
15   15
16   15
17   10
18    7
19   12
20   12
In the case of a single threshold at P/E10 = 20, there was also a portfolio that failed by the 20th year. Having a few early failures is typical of 100%-stock portfolios.

Next, I varied the upper threshold from 12 through 28 (and I looked at 40 and 80 as well) while keeping the lower threshold at 12. I set the lower threshold at 12 because it was in the middle of three good thresholds. This tends to lower the sensitivity to an exact threshold setting. The best choice historically would have been 11.

Code:

Code: Select all

P/E10   Failures
12   10
13   7
14   9
15   9
16   8
17   5
18   3
19   8
20   3
21   0
22   4
23   4
24   4
25   6
26   6
27   6
28   6
40   6
80   6

From these two surveys, we see that lower thresholds of 11 and 12 are interesting and that upper thresholds of 20 and 21 are interesting.

I examined each of those conditions at withdrawal rates of 5.0% and 5.1%

Code:

Rate Lower Upper Failures
5.0% 11 20 2
5.0% 11 21 0
5.0% 12 20 3
5.0% 12 21 0
5.1% 11 20 3
5.1% 11 21 1
5.1% 12 20 5
5.1% 12 21 2


The data favors a lower threshold of 11 and an upper threshold of 21.

Both 11 and 12 thresholds have zero failures at a 5.0% withdrawal rate when the upper threshold is 21. At a 5.1% withdrawal rate, a threshold of 11 has the slightest advantage possible. An upper threshold of 21 is consistently favored over a threshold of 20. Under this set of conditions, the sensitivity of the upper threshold is greater than that of the lower threshold.

The Historical Database Rate is 5.0% when the upper threshold is 21 and the lower threshold is either 11 or 12.

Looking at an 80% stock / 20% commercial paper portfolio under identical conditions except that there was no switching, there were zero failures at a withdrawal rate of 3.9% and one failure at 4.0%. The Historical Database Rate without switching is 3.9%.

Remarks

This only scratches the surface. Quite a bit can be done to reduce sensitivities. I doubt seriously that a 100%-50%-0% stock allocation is an optimal combination although Mike has provided evidence (that I have verified to my own satisfaction) that 0% is likely to be right at today's valuations (i.e., when P/E10 is 20 or above).

We have not looked at anything using TIPS or ibonds. I expect to find that they produce Historical Database Rates well in excess of 5.0%.

A special acknowledgment goes to intercst for developing the Retire Early Safe Withdrawal Calculator and to BenSolar for encouraging him to introduce the P/E10 switching capability.

Have fun.

John R.
Post Reply