Post
by hocus » Thu May 29, 2003 2:07 am
The more good things that I found and that others found, the more hostile and dogmatic the opposition became...along with stronger assertions that valuations don't matter.
JWR1945 has it right.
The history of this is that I researched SWRs on my own in 1996. When I saw the intercst study in 1999, I knew right away that it was flawed by its failure to consider the effects of changes in valation levels. I wanted to post on the question, but I could tell from how intercst responded to questions that it would cause a good deal of friction for me to tell the board what I knew. My primary concern was for the growth of the board. So I kept my mouth shut.
Wanderer was driven off of the board in February 2002. It alarmed me how few were willing to go public in support of him despite his great popularity. I saw it as a sign that what one poster had referred to as the "cult of personality" re intercst had grown so dominant that the board's continued usefulness as a learning resource was in question.
I determined that the most constructive response was to do something to generate more on-topic debate. I thought that an examination of the realities of SWRs would do that. I was not nearly as clear in my understanding of the realities of SWRs then as I am today (not that I am 100 percent clear on all points today). So I was tentative in my initial post. I framed it as a question. I was asking whether the board thought it might be a good idea to consider whether changes in valuation levels affect SWRs.
The initial board response was extremely positive, as JWR1945 suggests. There were dozens of posts in which people said that the discussion was the most exciting that any one had brought to the board in many months. Ironically, it was the great support that several of my posts generated (two had in excess of 80 recs, making them two of the five most popular posts of the entire year) that caused all the trouble. It was the intercst response to the positive board reaction to the matter I brought to the table that sent things in a downward spiral. As the personal attacks escalated, the more reasonable community members left the board or fell into silence. The learning process was aborted. I don't believe that this happened by accident.
JWR1945 is also right that the initial thread ended with an apology from me to both intercst and the board in general. The mistake that I made was on a small technical issue, but it was a mistake and I thought that an apology was required. The apology does not mean that I was wrong about the valuation question in general. The mistake was on a narrow point.
Changes in valuation levels always affect SWRs, I believe. But there are times when this is relatively easy to see and there are times when this is relatively hard to see. In cases like the year 2000, there is just no question. In 2000, the valuation level was at a point it had never reached in the 130-year time-period examined in the intercst study. It is simply not possible for a study that did not even take into account those sorts of valuation levels could possibly get the SWR right for that year. On this point I am dogmatic. There is no grey area. The study is wrong.
In years in which the valuation level is within the range of valuation levels covered by the study, I still believe that the failure to take changes in valuation levels into account is a serious flaw. But it is not as serious as in a year like 2000, and it is not nearly as easy in those cases to explain the reason why there is a problem
To see the concern re these sorts of years, you need to look at the study's treatment of SWRs with a less than 100 percent safety assessment. The study reports results for plans aiming at 95 percent safety or 90 percent safety or whatever. If a plan is assigned 95 percent safety, that means that, in 5 out of 100 start years, the retirement plan went bust. Intercst is saying that this means that someone retiring with that plan has a 95 percent chance of success.
I am saying that he is guilty of a logical fallacy. He is presuming that the years resulting in busted retirements are randomly distributed, that you have an equal chance of going bust no matter when you retire. I am saying that is not so, that the start years leading to busted retirements are bunched in years of high valuation.
If that is so, you have a greater than 5 percent chance of a busted retirement if you retire in a high-valuation start year. The converse implication is that you have a less than 5 percent chance of going bust if you retire in a low-valuation start year. By failing to account for changes in valuation levels, you cause people to have unreasonably high expectations of safety in some circumstances and unreasonably low expectations of safety in other circumstances.
Neither of these results is a good thing for aspiring early retirees. You obviously want to achieve your desired safety level. If you are seeking 95 percent safety, and in fact the data tells you that you do not have it when you retire in a high-valuation start year, you could still retire with 95 percent safety in that year, but you would need to lower your withdrawal rate to do so. The study should tell people this. The other side of the story is that, in low-valuation start years, you could retire with a higher SWR and still possess the level of safety you were seeking. A study not taking account of changes in valuation levels is wrong in different directions at different times, but it is always wrong, in my view.
What I needed to make an apology about is that in my posts in the first thread, I had not been careful to note that not all cases examined in the study have fail possibilities. In the 100 percent safe scenario, there are no start years causing busted retirements so what I am saying above as a technical matter is not so. I should have pointed that out. I knew that there were 100 percent safe scenarios, of course. My mistake was that I was focused on the point I was trying to explain, and overlooked the need to note the caveat re its application.
This does not necessarily mean that there is nothing wrong with the intercst approach in deaiing with scenarios where there is 100 percent safety. My personal belief is that there is probably still a flaw resulting from the failure to account for changes in valuation levels. My belief is that the problem in uncovering it is that there is not enough data in the study to bring it to light.
I beleive that if you ran a Monte Carlo analysis for the 100 percent safe scenarios, you could generate start years leading to busted retirements and thereby show the effect as you now can for scenarios assigned less than 100 percent safety assessments. I do not know enough about the technical aspects of statistical analysis to know if this is true for sure. Raddr would probably know.
JWR1945 is right that he is the one responsible for reigniting the debate. I intended the "My Plan" post to be my kiss-off to the board. You may not see it by reading the words, but there was steam coming out of my ears when I wrote that post. I felt that the board had been tricked, and that people were going to suffer serious life setbacks as a result, but that there was nothing I could do about it.
At that time I made a decision to start a new Motley Fool board when time opened up for me to do so. Me and JWR1945 agreed to exchange e-mails in the interim. I crafted the "Coin Toss" post solely for his eyes in an e-mail I sent to him. He told me that he thought I was wrong in my conclusions about the impossibility of holding a reasoned debate on the board, and ultimately convinced me that I should give the board another try. I submitted the "Coin Toss" post to the full board, it generated great enthusiasm, and the monster had been brought back to life.
The questions I have raised are valid ones deserving of serious consideration. Not all of the points are easy to understand, but the claims I am making about SWRs are important to the future of the Retire Early movement. If I am wrong in what I am saying, I should be proven wrong. That's a completely OK result. Alternatively, I should be proven correct. In no circumstances should the issue just sort of lay out there, unresolved.
My primary objection to what happened at the other board is not that I was attacked in extremely personal ways. I can deal with that. My objection is that the discussion was aborted. The questions raised were not resolved in a reasonable way. My contention is that this was deliberate. I believe from what I have seen in his posts that intercst came to understand that his approach is flawed and deliberately poisoned the debate to shut it down. Things got so out of hand with the various semantical games that no one could make sense of the arguments.
That is what I want to avoid here. I don't object at all to people challenging my points so long as the challenges are sincere ones and not exercises in pure word play. We should adopt a practical perspective. What is it that an aspiring early retiree needs to know when planning to hand in the resignation notice? Whatever approach best answers that question is the best approach. It is the approach that does the job that needs to be done that is best.
If we had perfect data on all aspects of the question, there would be one clear answer to the question "What is the safe withdrawal rate?" We don't have that. So as a practical matter we need to make compromises. There are some areas in which one person's view of what is the best compromise will not be the same as another's, and we should leave open the option of a variety of approaches in those areas.
There are other areas, however, where there is a right and wrong way to go. Would anyone say that an SWR analysis using the historical sequence methodology that does not include the Depression years or the 1970s is a valid analysis? I say no. I don't say that people should be forbidden from doing that analysis. It might tell you some interesting things. But I do not think that the claim that the result it produced is the SWR is a fair claim. Others may disagree, but that is where I am coming from.
I think that the result of that analysis might be of some use, but that ithat sort of analysis should be called something other than SWR analysis. It is not telling you what it really "safe." Perhaps it's as "Analysis of Allocaton Options for Those Comfortable Taking Risks with Their Retirement Plans." People can call it what they want so long as they do not call it SWR analysis. I think it is a bad thing for the future of SWR analysis if people do studies like that and call them SWR analysis. We can't stop them, but I think we should do what little we are able to do to discourage them.
I feel the same way about studies that do not account for changes in valuation levels. The results might provide you with some insights, it may be worth doing such studies. But such studies do not tell you the true SWR. They leave out a variable with a critical effect on the answer to the question you are posing. I do not think that such studies should be termed SWR analysis, but something else. I have no objection to disagreement on this point, of course. I am just saying what I think.
Regardless of whether such studies are valid or not, however, there is simply no question but that it was a fatal error for intercst to fail to make an adjustment for the fact that the valuation level in the year 2000 was one never seen in his 130 years of data, or at the absolute minimun to note that this caused problems with the analysis. There is no reasonable view that can be put forward that it is possible to obtain the right result from the historical sequence methodology in that circumstance. It is the faillure of the historical sequence methodology to account for valuation levels higher than ever experienced before that caused Bernstein to refer to the historical sequence methodology as "sometimes misleading."
A misleading result is not a good result. When you are offering advice on money matters, putting forward a misleading result has potentially grave consequences. Using a 4 percent SWR rather than a 2 percent SWR for a retirement beginning in the year 2000 would cause an early retiree to be $1 million off in his accumulated savings from what he needed to support a retirement at the recommended stock allocation percentage. A $1 million mistake is not a rounding error. It is a serious, serious mistake.
I do not fault Intercst for making the mistake. Presumably, he did not realize at the time he did the study that the historical sequence methodology does not work in some circumstances. But I fault him for not acknowleding the mistake when it was brought to his attention. Instead of thanking me for bringing the matter to the board's attention, he engaged in prohibited posting practices to poison the discussions and have the poster asking valid questions removed from the board.
The result is that people making use of the Motley Fool board to learn about early retirement will continue to suffer serious life setbacks because they are not being exposed to both sides of the story. People should be allowed to hear both sides and make up their own minds. They cannot. Only one side of the story can be voiced in an effective way at that board. The integrity of the board experience there has been severely damaged.
I don't want to dwell on the Motley Fool experience in our discussions here. But we should all try with great effort to avoid the mistakes that did so much damage over there. The key is that we all adopt an attitude that the goal never be for one side to "win" the debate and the other to "lose." The goal should be that we all learn. If we pursue that goal, there is no way that this will not end up being a positive experience for this community.
We need to begin to understand that intercst statements cannot be taken at face value. I said in an earlier post that I have always been a straight shooter. Well, the opposite is true of Intercst. He has never been a staright shooter. He crafts his posts to contain just enough surface plausibility to trick people who have not been able to study a matter in depth, but just enough deception that those trying to explain a concept fairly will not achieve their communication goals. This can be demonstrated by making reference to the REHP Post Archives.
My hope is to be able to discuss the SWR issue here without making undue mention of the history of the Motley Fool discussions. But if I am challenged with beliefs that people acquired from reading the TMF board or with statements made there, I am left with no choice but to address the credibility of the people who put forward those ideas or who made those statements.
I am prepared to go into the credibility issue in depth when I return to regular posting in October. But I am also happy to aim to avoid such discussions to the extent possible. I leave it to the board community as a whole to indicate which it thinks is the best way to go. My goal is to advance the communty's understanding of the realities of SWRs. I don't care one way or the other about the personalities involved. It doesn't trouble me to reveal to you in detail what intercst has done (because he is a public figure who did what he did knowing that it could cause great harm to people's hopes for a safe retirement), and it doesn't concern me to let that matter rest. I will do whatever appears to best serve the goal of advancing an understanding of the subject matter of the board.
)
And looks like Hocus agrees too. So let's move on. Could somebody do a SWR study with a (simple 3-level?) show of starting (of withdrawel) p/e valuation levels?
I've launched a few duds in my day but that may be one of the biggest. 
I'll keep hoping. 
