JanSz wishes to know about combining Gummy's Sensible Withdrawal Rates with the Switching of portfolio allocations. I am not able to answer JanSz's question fully. I am able to provide some relevant information. The combination appears to be quite powerful.
I have modified the question. In its original form the withdrawal rate would start at 2% of the initial balance (with adjustments to match inflation) and increase later at half of the portfolio's growth (averaged over 5 years). I interpret that as being in terms of what to do today because today's Safe Withdrawal Rates are only slightly above 2%. For me to evaluate switching, I am forced to look into historical sequences and extract Historical Database Rates. I am not aware of any current Monte Carlo calculators that switch according to valuations.
The parallel problem in the Historical Database would require a withdrawal rate of (roughly) 3% of the initial balance (with adjustments to match inflation). In both cases we are talking about withdrawals that are only slightly above dividend yields at the start of one's retirement.
JanSz described the portfolio as being the S&P 500 index and money market funds. It is tempting to substitute TIPS (or ibonds) since the returns would be better. I decided to use commercial paper instead (which is almost identical to most money market funds). We know that switching alone has been only slightly better than not switching in establishing Historical Database Rates. However, this problem is somewhat different. The initial withdrawal rate is being set at a level that would have comfortably produced a 100% safe Historical Database Rate. I wanted to see if there was something more to learn about switching in general. I am pleased to report that there is. Switching is even more beneficial than we had thought.
Now for the bad news.
I had to do almost everything by hand. The result is that I have gathered information for only one start year: 1966. That year is the one with the worst Historical Database Rate for an 80% stock allocation. The dividend yield in January of that year was 2.94%. That year was somewhat unusual in that the yield was less than the real interest rate of commercial paper (3.52%). It was more common for dividends to be higher, but only slightly so.
Instead of averaging 5 years of growth, I sampled the withdrawal rate relative to the current balance once every 5 years. If the new withdrawal rate had dropped, indicating real growth in the portfolio's balance, I increased the current rate by half of the difference between the two. If the new withdrawal rate increased, I left it unchanged. The calculator increased withdrawal amount according to inflation. All read outs were in terms of nominal dollars (i.e., without adjustments for inflation). That meant that I had to re-enter data with new nominal balances and new withdrawal percentages and read outputs from the new year.
Data Inputs: Details
All of my calculations were made on the Retire Early Safe Withdrawal [Rate] Calculator, version 1.61, dated November 7, 2002, without any additional changes by me. It is available for downloading from http://www.retireearlyhomepage.com/ .
My inputs were for an initial portfolio balance of $100000, with switching selected and an initial withdrawal rate of 3.000%. My switching threshold was 12.0. Below threshold, my stock allocation was 80% and above threshold it was 20%. All other values were left at their default settings. This includes an expense ratio of 0.20% of the portfolio's assets. (For those unfamiliar with the details of the calculator, F19 and B20 and I20 work. I19 and F20 do not. There are only two levels of switching in the current version. There are provisions for three. All outputs are in terms of whole dollars. Setting the initial balance to $100000 reduces round-off errors.) The defaults include an inflation adjustment that matches the CPI. The calculator uses commercial paper for the fixed income component when the switching of portfolio allocations is selected.
Test Conditions
The tables show the portfolio balances (in nominal dollars), withdrawal amounts (in nominal dollars) and withdrawal rates based upon the current balances in increments of 5 years. The basic portfolio maintains a withdrawal rate of 3% of the initial balance (plus adjustments for inflation). The switching-only portfolio maintains a withdrawal rate of 3% of the initial balance (plus adjustments for inflation). The sensible withdrawal portfolio includes increases in the withdrawal percentage when there is real growth in the portfolio's balance. It does not include switching. The combined portfolio includes switching and increases in the withdrawal percentage when there is real growth in the portfolio's balance. As indicated earlier, the increase is half of the difference between the withdrawal rate from five years earlier and the most recently calculated withdrawal rate. Both rates are in terms of their (then) current balances, not in terms of the initial balance.
Portfolio Balances
Code: Select all
Year basic switching sensible combined
1966 100000 100000 100000 100000
1971 107542 116034 107542 116034
1976 94101 129545 94101 129545
1981 130088 193784 130088 193784
1986 182615 302617 182615 302617
1991 303081 522077 298563 513238
1996 374275 615050 348632 577727
2001 747079 881016 662145 789924
Code: Select all
Year basic switching sensible combined
1966 3000 3000 3000 3000
1971 3754 3754 3754 3754
1976 5246 5246 5246 5246
1981 8208 8208 8208 8208
1986 10340 10340 10931 11578
1991 12698 12698 15648 16928
1996 14566 14566 18111 19418
2001 16416 16416 27404 28891
Code: Select all
Year basic switching sensible combined
1966 3.000 3.000 3.000 3.000
1971 3.491 3.235 3.491 3.235
1976 5.575 4.050 5.575 4.050
1981 6.310 4.236 6.310 4.236
1986 5.662 3.417 5.986 3.826
1991 4.190 2.432 5.241 3.298
1996 3.892 2.368 5.195 3.361
2001 2.197 1.086 4.139 3.066
With a (P/E10) switching threshold of 12.0, the 1966 portfolio switch positions are high for 1966-1974, low for 1975-1986 and high for 1987-2002. The stock allocations were 20% in 1966-1974 and 1987-2002. The stock allocations were 80% in 1975-1986.
The real interest rates of commercial paper in 1960-1965 were 3.25%, 1.20%, 2.72%, 2.17%, 2.45% and 3.49%. The dividend yields in 1960-1965 were 3.22%, 3.26%, 2.94%, 3.29%, 3.01% and 2.93%.
The real interest rates of commercial paper in 1966-1970 were 3.52%, 2.09%, 2.52%, 3.65% and 2.93%. The dividend yields in 1966-1970 were 2.94%, 3.41%, 3.08%, 3.02% and 3.50%.
Conclusions
1. Switching improves the behavior of the portfolio balance. It has been established elsewhere that Historical Database Rates are very close, with or without switching, when using commercial paper. By dropping the initial withdrawal rate to 3%, we have eliminated a troublesome dip in the portfolio balance at ten years.
2. The basic portfolio and the switching-only portfolio have the same withdrawal amounts. They are both 3% of the initial balance. The switching-only portfolio has consistently higher portfolio balances.
3. Withdrawals in 1966, 1971, 1976 and 1981 were all identical because the market was depressed. Beginning in 1986, the sensible withdrawal portfolio and the combined portfolio both increased payouts. The combined portfolio grew faster because switching improved overall performance.
4. The withdrawal rates of the both the basic portfolio and the sensible withdrawal portfolio exceeded 6% of their current balances. The sensible withdrawal portfolio stayed above 4%. With switching, the peak withdrawal rate was 4.2% of the current balance. The switching-only portfolio fell gradually to 1.1% in 2001. The combined portfolio fell gradually to 3.1% in 2001.
5. The combination of switching and sensible increases in the withdrawal rate is quite powerful. Switching eliminates the downside. Sensible withdrawals increase the upside.
Summary
These conclusions are based upon a single sequence in the historical record. Translating the cause and effect relationships to exact numbers cannot be done. The general conclusion to JanSz's question is that starting at a 2% withdrawal rate in today's market, switching and making sensible increases in withdrawal amounts (above inflation) is highly likely to succeed into the indefinite future even when commercial paper is the fixed income component. As mentioned earlier, there are attractive alternatives to commercial paper.
This analysis is based upon a series of posts. Together, they lead to this general conclusion.
Have fun.
John R.