Yesterday’s blog entry described two emails that researcher Wade Pfau sent me on December 18, 2010. It also set forth the text of my response to the first of those e-mails. The text of my response to the second of those e-mails is set forth below.
Thanks for putting an apology in the new post. That’s fine. I think it is especially good that you mention my name in the new post. It is 100 percent wrong that people posting at that board feel intimidated re posting my name. It obviously is not only you who have felt that intimidation. Lots of good and smart people have felt it and the discussions there obviously would be much more productive if people did not feel that. By using my name, you help others to get over their feelings of intimidation.
Drip Guy is a Super-Goon. I am not saying that you should not respond to him. The points he made in his post needed to be addressed. I am just pointing out that he has a long history of posting with bad intent on numerous boards. It would be wonderful if you got somewhere with him but I don’t see the odds as being good.
The point that matters most from my perspective is where Drip Guy says: “That is, if you had chosen to blindly withdraw 4% per year from a nest egg, it would have lasted at least 30 years until depletion, with 95% conf.” It is this way of saying things that is the cause of most of the confusion, in my assessment. It is of course an objective historical fact that 4 percent always worked in the past. 4 percent is the Historical SURVIVING Withdrawal Rate. There is certainly no question in my mind re that.
What I do not get at all is how he can say that a 4 percent withdrawal could be taken “with 95% conf.” Looking back, we know that 4 percent worked. So, looking back, you could take 4 percent with 100 percent confidence. But a determination of what is safe is by definition a forward-looking concept. The entire idea of having confidence in something only makes sense if you do not know what is going to happen. So the confidence percentage must take into consideration the factors that were unknown to the retiree at the time be began his retirement.
We know the range of possible returns starting from any possible valuation level. What we do not know is what particular returns sequence is going to pop up. Say that the retirement began in 1929. In 1929, 50 percent of the returns sequences that we have seen in the historical record permitted a 4 percent withdrawal to work and the other 50 percent caused a 4 percent withdrawal to fail. So it seems to me that the confidence level for a retiree taking 4 percent for a retirement beginning in 1929 is 50 percent, not 95 percent.
Yes, that retirement worked. But not because it was safe. A retirement that has only a 50 percent chance of working out cannot be retroactively characterized as “safe.” It was a risky retirement plan that worked because of good luck (a good return sequence happened to pop up). I am not able to think of any other field of human endeavor in which we determine what is “safe” by seeing what worked in two or three tests.
Say that we wanted to find out whether drunk driving is safe or not. We test 100 drivers driving 20 miles. 98 of the drivers are sober. 2 are drunk. The 98 drive the 20 miles without incident. The 2 get in major accidents and live but are crippled for life. Would we conclude from this that driving drunk is “safe”?
That’s what we are doing with the Old School SWR studies. A 4 percent withdrawal usually works with lots of room to spare. But it always means trouble for retirements that begin at times of high valuations. Prior to the 1990s, there were
only two times of super-high valuations in the modern record. On both of those occasions, 4 percent barely worked (although it worked with lots of room to spare on all occasions when valuations were not super-high). No, the 4 percent withdrawal did not bring death in either of the two tests (that is, the portfolios did not go to zero). But it came awful close in both cases. Is this not telling us that 4 percent is a high-risk withdrawal rate for retirements beginning at times of super-high valuations?
It seems to me that the Old School SWR methodology is telling us the Historical Surviving Withdrawal Rate. That’s a very different concept than the SAFE withdrawal rate. To determine what is safe, we need to look at the factors that determine what is safe. The record shows that the single most important factor is the valuation level that applies on the day the retirement begins and yet the Old School studies contain no valuations adjustment whatsoever. How could they possibly tell us what is safe if they do not even look to the factors that determine what is safe? This is why I say that these studies are “analytically invalid.”