An earlier blog entry described the background of my recent correspondence with Michael Kitces on safe withdrawal rates (SWRs). Set forth below is the text of an e-mail that I sent to Michael on August 25 .
Thanks so much for crafting that e-mail. We do not agree on everything, but we agree on a great deal. And on several of the points re which we do not agree, I think you are making points that need to be given careful consideration.
As I mentioned earlier, I intend to post both your e-mail and my e-mails as blog entries (please let me know if there are any comments that you would like to keep private). Your comments are going to be a big help to my readers. I hold strong beliefs re SWRs and I worry at times that my blog presents too unbalanced a take. I of course need to state my own views at the level of confidence that I hold on them. But I am grateful when someone is able to offer some reservations about my views as effectively as you do in the words below. That will cause those who tend to agree with me to think things through more carefully and that is obviously a good thing for all concerned. I’m less likely to change my mind given how much effort I have already put into understanding this stuff, but your comments at least prompt me to do some rethinking as well, and that too is of course a very good thing. So — thanks, man!
I don’t think it is right to say that Russell is “theorizing.” He uses the same data that the Old School researchers use. He does not employ Monte Carlo techniques. So he is not speculating about what may happen different in the future. He is REPORTING what the SWR is presuming that stocks perform in the future somewhat as they always have in the past. That’s just what the Old School researchers purport to be doing.
Both cannot be doing this and coming to such different conclusions! You are absolutely right to say that John’s findings are only as good as his model. The same is true of the Old School researchers. One of the models is wrong, and the findings produced by that model is doing the aspiring retirees who use it to plan their retirements great harm. I honestly do not see any way that that conclusion can be sidestepped.
Before I go on, please let me note that I do not mean by those words to show any disrespect for the researchers who constructed the earlier methodology. Nothing could be farther from the truth. Bill Bernstein has described the Trinity study as “breakthrough research.” I view that comment as being right on the mark.
That said, though, I strongly believe that the methodology used in the Old School studies is analytically invalid. Both things are so. The Old School studies took us to levels of understanding that were not possible before they came out and those new levels of understanding opened the door to the further advances achieved with the New School studies.
New research always builds on old research. It is by collectively saying the words “I” and “Was” and “Wrong” that humankind moves forward day by day.
To determine who is right, you need to look at the two models used to report what happened in the past and form an assessment as to which model is proper. That’s what this comes down to.
There’s a paradox in the Old School studies that you made mention of in your study. The paradox is that the take-out percentage that is declared “safe” for the fellow who retired in 2000 is also declared “safe” for the fellow who retired in 2002 even though the fellow who retired in 2002 has only half the assets because of the drop in prices suffered during those years. This makes zero sense, Michael. It is absurd and the Old School researchers need to come to terms with the analytical error that made this absurd paradox a reality of SWR research for all these years now.
It cannot possibly be “safe” for the fellow who retired in 2000 to take double the money out each year (because he is applying the 4 percent withdrawal rule to a portfolio value double the size of the one to which the 2002 retiree is applying the 4 percent rule). A model that produces such absurdities is a gravely flawed model.
The flaw is the failure to adjust the SWR for the effect of valuations.
There are two things that affect the SWR: (1) the return pattern that happens to come up (which cannot be known at the time of retirement); and (2) the valuation level that applies at the start date of the retirement (which is known). At a P/E10 level of 19, an 80 percent stock-allocation portfolio has an SWR of 4.1 percent. That obviously doesn’t mean that nothing above 4.1 percent will work. It means that 4.1 works in the worst-case scenario. In a best-case scenario (the most favorable returns sequence we have seen in the historical record), a withdrawal rate of 7.1 percent works. So the range of possibilities covers three percentage points of withdrawal rate.
The same three percentage points of withdrawal possibilities applies at the other valuation levels (since the range of possible returns sequences is the same). But the starting point of the three percentage points of withdrawal possibilities (it is this point on the spectrum of possibilities that we call the “safe withdrawal rate”) is NOT the same. At a P/E10 of 26, the range is from 3.12 to 6.13. At a P/E10 of 14, the range is from 5.41 to 8.41. At a P/E10 of 8 (which we may be seeing in the not-too-distant future), the range is from 9.13 to 12.13.
The strategic implications are huge. It’s not just that the Old School studies report the SWR wrong (although that in itself is a big deal as it is likely going to cause great amounts of human misery). What the historical data is telling us is that people should be changing their stock allocations when prices change dramatically. The long-term value proposition of stocks is just too different at dramatically different valuation levels to justify a strategy (Passive Investing) in which the investor stocks to a single stock allocation no matter what.
When we debate SWRs, what we are really debating is whether Passive Investing makes sense or not. I certainly do not say that people who believe in Passive Investing should not put forward their views. We of course need to hear those views. But the reality today is that the Old School studies are claiming that Passive Investing is “safe.” They are using the Passive Investing model (without even knowing it in most cases) to determine how we should construct our retirement plans and then declaring that is is “safe” to not even consider the possibility that the Passive Investing model is gravely flawed. This is arrogant, dangerous, and irresponsible. I don’t say that the intent of the researchers is to be those things. I say that the consequence of their too easy acceptance of the merits of the Passive Investing model is to fall guilty of generating claims that in fairness can be so characterized.
The problem would`largely go away if the researchers added caveats saying: “This study reports what is safe in the event that the Passive Investing model holds up; those who do not have confidence in the Passive Investing model should review the New School studies, which are rooted in an entirely different model of understanding how stock investing works (I call the alternate model “Rational Investing,” because it is rooted in an acceptance of the obvious [to me, at least] reality that valuations affect long-term returns). If such a caveat were included, investors using the studies would know what they were getting when they took them into consideration.
As things stand today, the investors making use of these studies view them as “scientific.” The common belief is that the studies are just reporting what the numbers say. The debate that rages today below the surface of much discussion of most financial planning discussions (whether valuations truly affect long-term returns or not and whether long-term returns are thus to a large extent predictable) is not mentioned. Most investors are not aware that, when they read the Old School studies, they are reading findings rooted in one particular school of thought only, a school of thought that very much appears (at least to me) to be well on its way to being entirely discredited.
I’d like to move away from discussion of the numbers for a moment to illustrate the critical importance of this last point. We have been discussing the flaws in the Old School SWR studies for over six years at the Retire Early and Indexing discussion-board communities. We have had thousands of people respond in extremely positive and constructive ways. We have also had a good number respond in hostile and abusive ways. Here is a link to an article at my site in which community members ask that the abusiveness that has so marred our discussions (resulting in bans on honest SWR posting at numerous boards!) be reined in:
Many of the people who have come to place their confidence in the Old School studies have reached a point where they have given up on the idea of being able to present their views in a civil and reasonable way. Why? What does this say about the methodology used in these studies? What it says is that the methodology gives the appearance of being science when in reality it is very much not science. The Old School methodology is rooted in an implicit acceptance of the Passive Investing model. People need to know this. When people use those numbers to plan their retirements, they are buying into an investing philosophy that may or may not hold water (I strongly believe that the Passive Investing model does not hold water). This should be stated up front as a caveat.
Again, I do not say that people should not be permitted to say that Passive Investing holds up. Just as it would be dishonest for me to say that I believe in Passive Investing, it would be dishonest for those who believe in it to say that they do not. But retirement advice that reflects only one point of view on how investing works (a point of view that has certainly not been proven to be the right one) should not be presented as “science.” These studies need to be questioned. For them to be effectively questioned, there needs to be a widespread understanding that they reflect a particular point of view, one that is itself not science.
You said earlier that you view the Old School and New School approaches as different research pathways. There’s a sense in which that is right, but there is also a sense in which it understates the problem we are facing today. The two approaches do indeed explore different pathways. What needs to be better understood is that the two pathways cannot both be proper pathways. Either valuations affect long-term returns or they do not. If they do not (or if for some reason effective predictions of long-term returns are not possible), the Old School studies are right and the New School studies are wrong. If valuations do affect long-term returns (and effective predictions are to at least some extent possible), the New School studies are right and the Old School studies are wrong.
Either there is a need to make an adjustment for valuations in determining the worst-case scenario or there is not. If there is a need to make such an adjustment and the adjustment is not made, millions of middle-class retirees will likely suffer busted retirements in days to come as a result. The financial planning community should be doing all that it possibly can to prevent this frightening scenario from becoming a reality. No?