randomguy wrote: ↑
Mon Dec 02, 2019 8:05 pm
Do you really think that the difference between 94.1% and 95.1% isn't with in the margin of error of the model?
I agree with you that SWRs don't show a significant difference -- quite possibly within the margin of error of the model. There are other metrics, though, that shed more light.
For instance, we can compare Certainty Equivalent Withdrawals under a variable withdrawal strategy:
Numbers bigger than $0 means a bond tent "wins" for that retirement cohort. Numbers lower than $0 means a bond tent "loses" for that retirement cohort. It is immediately clear that bond tents usually
lose. That's not necessarily a bad thing. After all, we're usually okay forgoing some upside to reduce the downside in the worst cases. So how do bond tents fare in the worst cases?
They do better in 1906-7, 1928-1931, 1965-1970. On the surface that's kinda exactly what we want. But "winning" doesn't mean it was significant. Here's the annual withdrawals for 1966 and 1969:
We're usually talking about 1-2% more income, just a few hundred dollars a year. It seems unlikely to make or break a retirement.
Overall, tents win 17% of the time and lose 83% of the time. When they win, the median gain is only $691 a year (from a $1,000,000 portfolio). When they lose, the median loss is $3,268 a year.
The only time that bond tents ever amount to anything was 1929, where they resulted in +12% more income. That's definitely something you'd notice.
If you don't like CEW and variable withdrawals, you can also use a coverage ratio with constant dollar withdrawals but the result is basically the same
So really bond tents don't protect against "sequence of returns" the way you might naively expect it. That's because the way most people talk about sequence of returns is vague. There are all kinds of bad sequences and bond tents only protect against one kind: sudden, large stock crashes.
This is easiest to see when we look at (real) portfolio values in the 1929 scenario (massive, sudden stock crash) versus the 1966 scenario (low returns over a long period of time)
It seems to me that a bond tent is mostly about giving up approximately 8-10% of your income in almost every scenario in order to do better in the event of an 70%+ stock crash in early retirement. Investing is all about tradeoffs, so I wouldn't call anyone crazy if they decided to make that tradeoff for themselves. But I'm less clear that is something that should be "recommended" as a "default" for people who don't know about they are trading off.