Rebalancing Strategies During Retirement: results for fixed withdrawals

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
embwbam
Posts: 85
Joined: Mon May 11, 2020 2:28 pm

Rebalancing Strategies During Retirement: results for fixed withdrawals

Post by embwbam »

In this post, I outlined my plan to test a rebalancing strategy related to Prime Harvesting, with a small additional rule to buy the dip. There are some suprises in the results that I wanted to share.

In this experiment, I used fixed rebalancing, 5/25 rebalancing bands, Prime Harvesting, and the new Prime Harvesting modified to buy stocks when the markets are low.

100% Maximum Safe Withdrawal Rates, 50 year retirements

Code: Select all

    Method           | MSWR | Median End Portfolio
    ---------------------------------------------
    Rebalance Fixed  | 3.3% | $2.770M
    Swedroe 5/25     | 3.4% | $5.952M
    Prime Harvesting | 3.4% | $2.380M
    Prime Buy Stocks | 3.4% | $2.578M
98% Maximum Safe Withdrawal Rates, 50 year retirements

Code: Select all

    Method           | MSWR | Median End Portfolio
    ---------------------------------------------
    Rebalance Fixed  | 3.5% | $2.370M
    Swedroe 5/25     | 3.7% | $5.071M
    Prime Harvesting | 3.7% | $2.018M
    Prime Buy Stocks | 3.8% | $2.171M
96% Maximum Safe Withdrawal Rates, 50 year retirements

Code: Select all

    Method           | MSWR | Median End Portfolio
    ---------------------------------------------
    Rebalance Fixed  | 3.5% | $2.370M
    Swedroe 5/25     | 3.7% | $5.071M
    Prime Harvesting | 3.8% | $1.929M
    Prime Buy Stocks | 3.9% | $2.080M
We can see that other rebalancing methods provide a consistent bonus to MSWR over fixed rebalancing. At a 96% success rate, the modified Prime method provides a huge 0.4% bonus to MSWR.

I was suprised that the new method has a higher MSWR than anything else. I thought buying the dip would only increase median portfolio size (which it hardly does). I was also surprised to see that 5/25 rebalancing bands provide about half the same benefit to MSWR (+0.2% vs +0.4%) while providing a much higher median end balance.

Other notes from the experiment
  • 60/40 was optimal for the Prime strategies, at higher stock allocations they are only barely superior to rebalancing, because the bonds-first strategy doesn't have enough time to work.
  • 120% / 80% thresholds had the best results for the prime strategies.
  • Results are similar for 40 year retirements
  • We can't effectively test 60 year retirements, because it reduces the number of bad starting years in the data set
What's next?
I've started playing with Amortized Based Withdrawals to see how they combine with these results. I'll post about it soon!


Prime Harvesting

Prime Harvesting is a method created by Michael McClung, and works as described here. In short, you withdraw from bonds and never rebalance, until stocks swing high enough to reach a threshold (120% of inflation-adjusted starting value). You then sell above that threshold to buy bonds.

It works on the principle that you start with enough bonds to survive sequence of returns risk, and by the time you run out the average return of the stock market has at least begun to kick in.

Prime Harvesting Modified to Buy Stocks

I modified the above by adding a second threshold at 80% of inflation adjusted starting value. When stocks went below that threshold I sold bonds to buy them to bring them back to the threshold. See the preivous post for more information

Links and More Information Raw Results (MSWR, Success%, Median End Portfolio)

Code: Select all

    Years: 50

    Rebalance Fixed
    ----------------
    ("RateResult",3.300%,100.000%,$2.770M)
    ("RateResult",3.400%,97.849%,$2.541M)
    ("RateResult",3.500%,97.849%,$2.370M)
    ("RateResult",3.600%,92.473%,$2.186M)
    ("RateResult",3.700%,91.398%,$2.028M)
    ("RateResult",3.800%,90.323%,$1.812M)
    ("RateResult",3.900%,82.796%,$1.496M)
    ("RateResult",4.000%,79.570%,$1.299M)
    ("RateResult",4.100%,74.194%,$1.114M)
    ("RateResult",4.200%,66.667%,$0.871M)
    ("RateResult",4.300%,63.441%,$0.744M)
    ("RateResult",4.400%,60.215%,$0.559M)
    ("RateResult",4.500%,56.989%,$0.374M)

    Swedroe 5/25 bands
    ------------------
    ("RateResult",3.300%,100.000%,$6.289M)
    ("RateResult",3.400%,100.000%,$5.952M)
    ("RateResult",3.500%,98.925%,$5.645M)
    ("RateResult",3.600%,97.849%,$5.347M)
    ("RateResult",3.700%,97.849%,$5.071M)
    ("RateResult",3.800%,93.548%,$4.833M)
    ("RateResult",3.900%,92.473%,$4.621M)
    ("RateResult",4.000%,91.398%,$4.404M)
    ("RateResult",4.100%,88.172%,$4.176M)
    ("RateResult",4.200%,83.871%,$3.752M)
    ("RateResult",4.300%,78.495%,$3.419M)
    ("RateResult",4.400%,76.344%,$3.137M)
    ("RateResult",4.500%,73.118%,$2.850M)

    Prime Harvesting
    ----------------
    ("RateResult",3.300%,100.000%,$2.477M)
    ("RateResult",3.400%,100.000%,$2.380M)
    ("RateResult",3.500%,98.925%,$2.276M)
    ("RateResult",3.600%,97.849%,$2.183M)
    ("RateResult",3.700%,97.849%,$2.018M)
    ("RateResult",3.800%,96.774%,$1.929M)
    ("RateResult",3.900%,92.473%,$1.840M)
    ("RateResult",4.000%,92.473%,$1.762M)
    ("RateResult",4.100%,91.398%,$1.687M)
    ("RateResult",4.200%,87.097%,$1.611M)
    ("RateResult",4.300%,86.022%,$1.497M)
    ("RateResult",4.400%,81.720%,$1.405M)
    ("RateResult",4.500%,75.269%,$1.236M)

    Prime Harvesting 2
    ------------------
    ("RateResult",3.300%,100.000%,$2.664M)
    ("RateResult",3.400%,100.000%,$2.578M)
    ("RateResult",3.500%,98.925%,$2.493M)
    ("RateResult",3.600%,98.925%,$2.398M)
    ("RateResult",3.700%,97.849%,$2.262M)
    ("RateResult",3.800%,97.849%,$2.171M)
    ("RateResult",3.900%,96.774%,$2.080M)
    ("RateResult",4.000%,92.473%,$1.958M)
    ("RateResult",4.100%,91.398%,$1.870M)
    ("RateResult",4.200%,90.323%,$1.734M)
    ("RateResult",4.300%,87.097%,$1.640M)
    ("RateResult",4.400%,83.871%,$1.548M)
    ("RateResult",4.500%,75.269%,$1.436M)
Post Reply