Yesterday’s blog entry reported on an e-mail sent to me by Academic Researcher Wade Pfau on March 2, 2011. My response, sent later the same day, is set forth below.
That all sounds super.
I’ll mention something that I’ve been working on for a long time and that I think is important and that I think you were touching on in some of your posts on the Bogleheads thread re SWRs.
I believe that there is an element of this that is random — the nature of the particular returns sequence through which reversion to the mean is achieved. Those who used the 4 percent rule for retirements beginning in 1996 were doing something risky — but after 10 years passed without a crash, they entered safe territory. It is only retirements that experience a crash within the first 10 years that fail. No one knew in advance what sort of return sequence was going to turn up. But we saw a lucky one for the 1996 through 2006 period and those retirees are now okay.
This throws lots of people. People look at the fact that a retirement that began when the P/E10 was 33 (1929) survived with a 4 percent withdrawal and conclude that a 4 percent withdrawal is safe even at times of insanely high valuations. No! If you look at the returns sequence that occurred from 1929 through 1939, you see that it was a bit more favorable for retirees than most we have seen in the record.
People have a hard time accepting that because the results were so poor. But the results were not poor because there was a bad returns sequence. Results were poor because valuations were insane. The somewhat lucky returns sequence saved an extremely dangerous situation from turning out worse than it did.
I believe that it would be a huge help to be able to characterize what constitutes a good returns sequence for different types of investors (what is good for retirees is not the same as what is good for young investors). John and I were working at the time he died on a calculator (“The Returns-Sequence Reality Checker”) that would permit people to plug in returns sequences (either ones we have seen in the historical record or made-up ones that they want to test) and see what sorts of results they produce over different time-periods (the 30-year results are of course affected by the starting-point valuation level of the time-period examined).
I have put the project on hold for a long time. I am thinking that I need to get back to work on it because it is looking more and more important as things move forward. My problem has been that I cannot handle the numbers side work on my own.
I found a fellow (Sam) who helped me develop a prototype (it’s not finished but I believe he is willing to finish it if I ask). The guy is a totally great guy. But I have been reluctant to take on the cost of putting it in calculator form because I cannot be sure of all his formulas (he’s certainly trying to do it right but I am not sure that he always understands the questions that I am asking — I am a numbers dunce and I speak a different language). John knew me well enough that we got to a point where we could figure each other out. That gave me the confidence to put calculators at my site even though I do not understand the formulas (but only the concepts) that drive them.
If you are willing, I would like to finish the prototype with Sam and then have you review it to see if it really does what it is supposed to do and names sense. If you gave a sign-off, I would be comfortable spending the money to have it done as an actual calculator (the prototype would be an Excel spreadsheet).
I don’t want to take you away from your other work and I of course understand if this is something that you do not want to get involved in. But it might spark some thoughts on your side too. If you tell me that you would be willing to take a look at this, I would redouble my efforts to see if Sam and I could finish the prototype. I can show you the unfinished prototype now if you feel that that would give you a better idea of the concept being explored here.
Wade responded the same day.
In reference to my discussion of how our inability to know in advance what sort of return sequence will turn up adds an element of randomness to the analysis, he said: “I agree with this, though I know you have not been popular when you
say it. This is similar to your drunk driving analogy, which I agree with.”
Wade referred me to a paper titled “Will 2000-Era Retirees Experience the Worst Retirement Outcomes in U.S. History? A Progress Report after 10 Years.”
He said that he would be happy to look at the prototype for the Returns Sequence Reality Checker. His e-mail states: “I’d been meaning to look more at your calculators, but just haven’t gotten around to it yet. I know people accuse you of using imaginary numbers, but to me that sounds like the definition of Monte Carlo simulations. So I’m not sure whether criticisms of your calculators have merit or not. I can’t really take anything DRiP Guy says at face value.”
Finally, he reported that Dan Moisand was writing an article about Wade’s research on safe saving rates. He described Dan as an influential financial planner who had served as a past president of the Financial Planning Association. “So getting his approach about this is definitely a step in the right direction.”