vineviz wrote: ↑Sun Jan 23, 2022 8:37 pm
McClung wrote: "A Rising Glidepath strategy showed a moderate improvement over traditional rebalancing for some realistic retirement cases when based on equivalent stock-bond averages, but this improvement was not consistent over broader testing."
He's describing what I pointe out earlier: the rising glidepath doesn't
consistently improve outcomes, and when it has in the past the effect has been small.
The phrase "rising glidepath" has two meanings: a specific strategy in that 2013 paper, and a more general meaning of any strategy where the percentage of equities goes up over time. I think we all agree that the specific strategy from the 2013 paper is a weak performer. Even Pfau and Kitces have moved on to bond tent.
McClung evaluates a number of different withdrawal strategies. Most of them, and all of the top performers, increase stocks over time. Perhaps the simplest one that could be called a bond tent is "bonds first": when withdrawing for living expenses, take from bonds until they're gone and you're at 100% stocks. Other than that, move between stocks and bonds, including never rebalancing.
Code: Select all
Fixed AA | Bonds First | Prime
US worst case 4.0% | 4.4% | 4.4%
UK worst case 3.0% | 3.7% | 3.7%
Japan worst case 3.3% | 3.5% | 3.8%
US 90% 4.7% | 5.0% | 5.2%
UK 90% 3.8% | 4.2% | 4.2%
Japan 90% 4.6% | 5.2% | 5.6%
(Figures 29 through 32)
The worst case, as you say, is only a single time period in the entire history. However, bond tent (bonds first) outperforms all 3 countries. To look at other time periods, we can look at the 90%. Again, the bond tent outperforms.
Figure 37 compares worst case, 90%, 80% and 50% for the 3 countries, plus some simulation results, and using large cap stocks instead of total stock market. Although there are a few cases where the fixed asset allocation does better, the bonds first strategy almost always beats it.
He concludes by saying:
Backtesting clearly shows that eliminating this occasional tilt toward a high stock percentage increases known risk. ... Although it’s tidier and more comforting to have a fixed stock-bond ratio, it comes at a cost. ... Despite the data, retirees can still choose to maintain a bond floor, selling stocks as needed to maintain a minimum bonds percentage. ... While this is a viable option for retirees who are uncomfortable with a low bond percentage, it’s important to understand that across four datasets (SBBI, Shiller, UK, Japan) this consistently lowered performance during difficult retirement periods. On the other hand, when combined with a variable-withdrawal strategy, the loss in income is divided across many years without a major impact — this is briefly revisited in Chapter 10. Bounding the lower bond level isn’t recommended based on the data, but an otherwise strong retirement plan can handle it.
And to the point about other periods, we would expect bond tent to outperform in those times too, since the bear markets were shorter than the 16 year one from 1966 - 1982. Only a 20+ year bear market would make bond tent perform worse.
Is the effect large enough to be worth using bond tent over a fixed AA with annual rebalancing? That's a personal choice. If you're a "no one was ever fired for buying IBM" type, then probably not. In fact, you might even want to stick with age in bonds or some other decreasing glide path, as used in target date funds. Personally, I'm planning to use a bond tent, although with a bond floor of maybe 25% or 20%.