A comment said they’d be curious to hear more about where I’ve personally landed with PMT-based retirement withdrawals. The basic PMT formula offers two main levers for personalisation — the rate chosen and the time-frame to operate over.
=pmt(rate, period, 0, 1)
- My rate is based on expected returns of my actual portfolio. For each major asset class I generate an “simple expected return”.
- The period is the average of (115 minus my partner’s age) and (our joint life expectancy at the 90th percentile). I use my partner’s age because they are younger than me and I am virtually guaranteed to die sooner.
A bit more on each in turn….
When I originally talked about the rate parameter I mentioned three of the many possibilities
- Siegel & Waring use the current real TIPS yield, updated annually.
- Steiner suggests using 4.5%, a number he provides (I’m not sure how he arrived at it), also updated annually.
- VPW uses the long-run historical real returns of global stocks & bonds applied to your actual portfolio weights. It does not update annually.
Flavors of PMT-based withdrawals, part 2 of 2: Expected returns
Last time I said that PMT calculations for retirement need two inputs — the number of years of payout and the expected…
A fourth option, via Gordon Pye, is to use 8%, which he offers as a “don’t preemptively cut spending” philosophy.
Pye v VPW
Back in 2008/9 Gordon Pye published some articles (which eventually turned into a book) about his Retrenchment Rule…
I think Pye’s approach actually has a lot of merit for someone who is close to normal retirement age and finds themselves unexpectedly retired due to health, age-discrimination, or simply industry-wide changes. That said, I’m nowhere close to normal retirement age.
I think VPW’s approach of looking at your actual portfolio makes a lot of sense. I think Siegel & Waring’s idea of updating things every year as the world changes also makes sense. I extend VPW’s approach by estimating an expected return for each asset class I own and then adding them all up to get an estimated expected return for my entire portfolio. If my estimated expected return for Emerging Markets is 6.8%, and EM makes up 20% of my equities, and equities make up 90% of portfolio, then they will contribute 1.23% of estimated expected return.
=6.8% * 20% * 90%
For equities, I just use 1/PE for my estimate. For bonds and REITs I just use the current yield for my estimate. Those give nominal returns, so I also subtract off a flat -2% (the Federal Reserve’s target number for inflation) to account for inflation and get estimated real returns.
(I use PE instead of CAPE because it is trivial to find the PE for every mutual fund, ETF, and asset class in existence. I don’t worry too much about whether we’re talking about trailing PE or forward PE or whatever.)
These estimates don’t need to be perfect. They just need to be a bit better than not estimating at all. Right now, it means I’m using a rate of 3.75%. If I were using VPW, it would have me using a rate of 4.7%. If I were using TIPS yields, it would be 0.95%.
Doing it this way gives me two benefits:
- I don’t think we’re in some massive bubble and the world is going to end. But there’s a decent case to be made that the future decade will have somewhat muted returns. The shift from 4.7% down to 3.75% feels like it is taking that into account.
- By updating this once or twice a year, you do get some smoothing of withdrawals. We actually just saw this back in December/January where the market dropped but expected returns went up.
For a little while I did a more complicated “estimate expected returns” where I would hunt down a fair number of “2019 Expected Returns Report” from places like BlackRock and GMO and BNY and whatnot. And then average them all together under some kind of “wisdom of the crowds” forecast. I eventually decided that was way too much effort and just making a kind of false comfort.
Anyway, again we’re just trying to get in the general ballpark, not to be perfectly accurate.
We’ve also seen various approaches to determining “how long” the manage the draw down over.
VPW goes to age 95 (But for who? Me at age 95? My partner at age 95? There is an age gap of several years between us, so it isn’t a trivial difference.) Siegel & Waring initially suggest “to age 120” (since that’s how old the oldest known human was) before offering a compromise of “use the average of ‘to age 120’ and current remaining life expectancy”.
I’ve adopted Siegel & Waring’s approach, though I use 115 instead of 120 for the upper limit, for two reasons.
- It just seems crazy to ignore actual life expectancies, especially when they are so easy to find calculators for online. I’ve chosen the 90th percentile for me & my partner because that feels “good enough”, especially since I’m already averaging it with a bigger number to account for “worst case longevity”.
- VPW uses age 95 and longinvest repeatedly suggests that real retirees should annuitise before they get close to that. That’s good advice if you live someplace where you can buy an annuity. There are no annuities where I live, so I need to “roll my own annuity” by being more conservative with the period.
I use 115 instead of 120 because when I look at lists of “oldest people ever” it seems like a more plausible cut off. Only 2 people have ever lived more than 117 years.
List of the verified oldest people - Wikipedia
The oldest person ever whose age has been independently verified is Jeanne Calment (1875-1997) of France, who lived to…
And one of them, Jeanne Calment has a question mark. One researcher has claimed that Jeanne Calment actually died in 1934 and her daughter claimed her identity to avoid paying inheritance taxes. According to this claim, Calment was actually only 99 when she died. That would mean only 1 person ever lived more than 117 years.
I use the average because it (slightly) front-loads spending to when we’re younger and more active.
The net result
I am fairly conservative with the longevity largely out of concern for ensuring there’s enough for my partner. My partner will not receive Social Security or any other government-sponsored pension. My partner will likely outlast me by a fair number of years. My partner would not be able to earn an income that would cover anything like our current standard of living and I expect that as the years pass their human capital will erode further.
I am slightly conservative with the rate due to slight concerns about lowered returns over the next decade and the reality as an early retiree that (especially when combined with the above about my partner) my portfolio needs to last six or seven decades, not the two or three that “normal” retirees face.
The reality is that, through a combination of being fortunate in life, living in Vietnam where the cost of living is relatively low, and having a paid-off house, our “normal living expenses” are quite low relative to what PMT says we could be withdrawing.
I use the results of the PMT calculation mostly as a guide for how much “free cash flow” we have for discretionary spending on vacations, house renovations, giving to family, charitable contributions, and so on.