A Concise Explanation of the SWR Dispute

Financial Independence/Retire Early -- Learn How!
Post Reply
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

A Concise Explanation of the SWR Dispute

Post by hocus »

This post is intended as a response to Dagrims and Andrew61 and Galagan and CrazyLegs883 and StubbleJumper and a few others who in recent weeks have requested a concise explanation of the core dispute in the debate about safe withdrawal rates that has been going on for some time at the Motley Fool's Retire Early Home Page board and that is now being discussed here from time to time as well.

The post linked below sums up the problem.

http://boards.fool.com/Message.asp?mid=18237425

Intercst says here that "I've made no assumption about the [statistical validity of the] data" used in the SWR study. He notes that he has not tried to determine "Is it skewed in some way?" Thus, he is not able to assign a statistical confidence interval" to his claim that the SWR is 4 percent. All he really knows for sure is that "the study does what it says it does."

Think about that last statement for a moment. How hard is it to prepare a study that "does what it says it does?" Not hard at all so long as you do not limit yourself to statements that can be supported by a valid use of data. I could issue a study this afternoon claiming that I will show by reference to data that the earth is flat, and, so long as I permit myself the use of data with no scientific validity, the study would indeed "do what it says it does." I could then claim "100 percent certainty" that my study does what it says it does. That wouldn't really prove that the earth is flat, would it?

The thing that keeps a researcher honest in his claims, the thing that makes a study science and not conjecture, is the use of valid data. Intercst has not determined whether his data is suitable for answering the question examined in the study, so he has not engaged in a process that merits the designation "science." By using whatever data he wanted to use, regardless of whether it was a valid way to answer the question or not, he has presented his personal opinion on safe withdrawal rates, nothing more.

Several posters at this board have made attempts to determine how statistically valid the intercst study is, and the answer that has come back is, "not very." There are probably some circumstances in which a 4 percent withdrawal rate applies. But it appears that that has been the exception in recent years rather than the rule. In any event, there are few circumstances in which the "optimal" stock allocation is 74 percent. In most cases, it appears that the optimal allocation is something close to 50 percent.

The bottom line is that the intercst study doesn't provide you with much help in your task of determining the safe withdrawal rate for your investment plan. What are we to do now that we know that, toss the intercst study in the trashbin? I don't think that's such a hot idea. Despite its lack of statistical validity, the intercst study does provide some valid insights. It is not the final word on the question it examines, but so long as you understand the study's limitations, it can prove helpful in your effort to put together a successful plan. You would be a fool not to make use of it.

The logical thing to do is to make use of the intercst study, but to supplement what you learn from it with insights obtained from other sorts of studies that shed light on the questions that the intercst study ignores. Members of the REHP board community have raised lots of questions they would like to know more about--whether real estate could be an effective Retire Early investment, whether it makes sense to diversify by investing in international stocks or small cap stocks, whether it is possible to benefit from taking a little off the table in times of extreme Price/Earnings ratios, and so on. Exploration of these questions would bring the REHP board to life because it would permit for an increase in on-topic posting, something that has been in short supply there for a long time now.

But intercst and some others who post at that board do not believe that such discussions should be permitted. The argument they offer is that most studies looking at alternative investment stratgies are in some way imperfect. Perhaps all the data needed is not available, or perhaps it is available going only 20 or 30 years back, or whatever.

It is true that most of the data we need to look at is imperfect in some way. However, most of this data possesses at least as much statistical validity as the intercst study data, and much of it is more valid than that data. So it is a double standard to say that the REHP board may debate the intercst recommendations but not the data supporting alternatives to the intercst recommendations.

Posters at the REHP board are permitted to put up posts describing alternative investment strategies. The problem comes when they try to engage other community members in discussions of the merits of these strategies. At this point, the group that seeks to limit debate to the intercst recommendations engages in four posting practices prohibited under the TMF posting rules--disruption, deception, personal attack, and ridicule. Most sensible posters stop participating in the threads once this happens, and the community members seeking a reasoned discussion of alternatives are blocked from achieving their goals.

What is needed is an expression of community will at the REHP board to permit fair discussion of a variety of investment strategies. I have tried a number of approaches to bring that about, and have not yet been successful. The next thing I want to try is to have this board develop a Frequently Asked Questions (FAQ) statement on SWRs, and then take that FAQ statement to the REHP board and request that the community there endorse it as their own FAQ. My hope is that, with a FAQ statement to point to as support, REHP community members seeking a discussion of investing alternatives will have greater success obtaining a fair hearing for their ideas.

The key issue in dispute is, should there be a double standard at the REHP board, so that only the intercst study is exempt from a general rule prohibiting debate on data that possess some statistical imperfection? Or should the board be run by Motley Fool rules, which would allow any poster with insights or data to offer that would help other community members plan successful early retirements to put his or her ideas or data forward for the full community's consideration?
StubbleJumper
* Rookie
Posts: 2
Joined: Wed Dec 11, 2002 3:00 am

Post by StubbleJumper »

This post is intended as a response to Dagrims and Andrew61 and Galagan and CrazyLegs883 and StubbleJumper and a few others who in recent weeks have requested a concise explanation of the core dispute in the debate about safe withdrawal rates that has been going on for some time at the Motley Fool's Retire Early Home Page board and that is now being discussed here from time to time as well.


Hocus:

I do not recall asking for any such explanation. Personally, I have no particular problem with the SWR study that dominates the conversation over at the Fool. It is an interesting exercise which provides a rough idea of where one should start in estimating a SWR.

I do, however, have concerns about how the study has become gospel. It ain't perfect. Everybody should recognize that. However, there are some adherents who defend the dogma to the core. They insist that it provides the one right answer on both SWRs and asset allocation. There is an absolute failure to recognize the possibility and probabilty that we could experience something in the future that is nastier than the Great Depression.

I appreciate the rough results of the study -- they are valuable. However, I reject its underlying premise that one should invest only in S&P500 and fixed income. There is a very strong argument that we ought to attempt to diversify our holdings a little more...

I also reject the hypothesis that the future cannot be worse than the past. Spouting nonsense about asteroids, hemorrhoids or nuclear war is not productive. We have just experienced a period record valuation. In our lifetime we have witnessed the effect of record valuation on the Japanese stock market. Of course, I am comparing apples and oranges to a certain extent, but there does exist an advanced, industrialized economy that is experiencing an event in its financial markets that may eventually be worse than the Great Depression. I have played a little bit with the Japanese financial data, and somebody who quit his job with Toyota in 1988 and set up a 75/25 portfolio of the Nikkei Index and Japanese fixed income is in deep trouble now if he has been taking an inflation-adjusted 4% withdrawal....as near as I can tell, his portfolio is unlikely to last 20 years, let alone 30 or 40. I'm not saying that the US is likely to experience anything similar.....but perhaps we ought to recognize that there exists a definite possibility of something nasty happening over the next ten years. This is not the time to hide behind dogma.

I prefer to approach the subject with thoughtful uncertainty. The things that I do not know about SWRs vastly overwhelm those which I do know....


StubbleJumper

Ah, but I was so much older then, I'm younger than that now.
User avatar
BenSolar
*** Veteran
Posts: 242
Joined: Mon Nov 25, 2002 5:46 am
Location: Western NC

Post by BenSolar »

but there does exist an advanced, industrialized economy that is experiencing an event in its financial markets that may eventually be worse than the Great Depression. I have played a little bit with the Japanese financial data, and somebody who quit his job with Toyota in 1988 and set up a 75/25 portfolio of the Nikkei Index and Japanese fixed income is in deep trouble now if he has been taking an inflation-adjusted 4% withdrawal....as near as I can tell, his portfolio is unlikely to last 20 years, let alone 30 or 40.


No doubt that the current Japanese bear market is worse than the Depression in terms of safe withdrawal rate. A 4% inflation indexed withdrawal is almost certainly bust now or this year. 14 years it lasted. This thread at TMF discusses it. http://boards.fool.com/Message.asp?mid= ... e#18209072 .

Ben
"Do not spoil what you have by desiring what you have not; remember that what you now have was once among the things only hoped for." - Epicurus
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

I do not recall asking for any such explanation. Personally, I have no particular problem with the SWR study that dominates the conversation over at the Fool. ...I do, however, have concerns about how the study has become gospel.

OK, StubbleJumper. I was not trying to say that you have any particular oppsosition to the study itself. I have used the study myself and have found value in it. So I am not trying to cause any others to find fault with the study proper.

My goal is to get others to find fault with the idea of using the study as, in your word, "gospel." I see it as one useful tool among many that could be used to provide insight into the safe withdrawal rate, not by any means the final word on the subject.

You were one of the people I had in mind when I crafted the post. You made some outstanding points in the discussion that followed in the wake of the "Community Rules!" post from two weeks ago. I think that the key to resolution of this mess is to get some fresh perspectives on it, and I was happy to see you offering some thoughts on this matter on the other board.

I've also been encouraged by the comments of Dragrims, Andrew61, crazylegs883, and galagan. I believe that there are many people in the REHP community who have far more reasonable views than the views of the posters there who tend to dominate debate on this question.

I noted that a post by Andrew61 in the "New FIRE Board to Prepare FAQ on SWRs" thread obtained in excess of 25 recs. It's a lot more than just me that sees value in exploring the realities of SWRs, despite claims of people like ResNullius and GolfWayMore to the contrary.

Anyway, thanks for clarifying your views, StubbleJumper. I hope you will be able to participate in the effort to write a FAQ on SWRs that begins January 13.
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

No doubt that the current Japanese bear market is worse than the Depression in terms of safe withdrawal rate.

I don't think that's necessarily so. Not because the situation for stock investors in Japan has not been terrible. But because the way that the intercst study assesses the damage done by the U.S. crash in 1929 is artificial and far understates the effect that the crash had on safe withdrawal rates of retired investors.

If you retired in early 1929 with a 74 percent stock allocation, you saw three-quarters of your portfolio lose close to 90 percent of its value in the following years. The idea that most investors in those circumstances would continue to hold a 74 percent stock allocation until prices recovered is just absurd. There is no historical data that I have ever seen lending support to this fantastic assumption.

In the real world, most investors who went with such high stock allocations lowered them when prices fell by 90 percent. The lowering of stock allocations caused them to miss out on much of the great gains that stocks experienced in subsequent years. It is a flaw of the intercst study to count these gains in its determination of the SWR since most investors following the study's recommended stock allocation strategy did not experience these gains.

The true safe withdrawal rate is the amount that you can take out of your portfolio each year with virtual certainty that your portfolio will survive until death. Counting gains that you are not likely to enjoy in the real world does not help in the effort to determine what this number is.

At the recent bull market top, the DOW was at close to 12,000. A 90 percent drop (the worst case scenario in the past 130 years) would bring it to something near 1200. For a 4 percent withdrawal to apply, there would have to be zero investors with a 74 percent stock allocation who lowered their allocation in response to a fall to DOW 1200. If there were any investors who lowered their allocations, it would be the results obtained by those investors that would determine the worst case scenario.

To presume that none of the investors with 74 percent stock allocations would lower their allocations in response to a fall to DOW 1200 is sheer folly, in my opinion.
User avatar
BenSolar
*** Veteran
Posts: 242
Joined: Mon Nov 25, 2002 5:46 am
Location: Western NC

Post by BenSolar »

hocus wrote:
For a 4 percent withdrawal to apply, there would have to be zero investors with a 74 percent stock allocation who lowered their allocation in response to a fall to DOW 1200.


I agree whole heartedly in the need to emphasize the likelihood that investors will bail out in a severe bear. It is a very big risk, one that we have seen in action already in this bear, even though the severity is not even close to 1929.

However, I don't think that you can say there have to be zero investors who strayed from a given allocation. Some in any population will inevitably change their investment plans, even if they are widely diversified away from stocks. Sometimes the change helps, sometimes it hurts. Doesn't mean we can't model what would have happened without the tweak.

I think that in reality, even the most well diversified ER will probably be taking some emergency measures in the event of a 1929 style meltdown. That is, if they are not as rich as Creosus and could lose 90% and still withdraw at 1% to live on. :wink:

B.
"Do not spoil what you have by desiring what you have not; remember that what you now have was once among the things only hoped for." - Epicurus
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

However, I don't think that you can say there have to be zero investors who strayed from a given allocation.

BenSolar:

To measure what the result would be if a single investor lowered his stock allocation would give you the safe withdrawal rate with a 100 percent safety level assigned to it. I think it would be perfectly reasonable to use a number with only a 95 percent or 90 percent safety level.

But I am all but certain that, with a 90 percent drop in stock prices (the worst case scenario in the past 130 years), more than 10 percent of those investors who retired with 74 percent stock allocations would take some money out of the market. So it would not be valid to use the SWR numbers from the intercst study without making an adjustment, even in times in which valuations were within the ranges experienced during the 130-year time-period he examined.

The tendency of investors with large stock allocations to take money out of stocks during big downturns has been demonstrated by the historical data. So you can't just ignore it when trying to determine the safe withdrawal rate. It may not be possible to assess this factor with absolute precision, given the available data. But it is better to be roughly right than definitely wrong. A study that ignores this factor altogether cannot produce the right number, in my view.

I think that in reality, even the most well diversified ER will probably be taking some emergency measures in the event of a 1929 style meltdown.

Yes, but I believe that those with higher stock allocations will be pressed to make bigger changes than those with more moderate stock allocations. So you have to incorporate this factor into your analysis of what the "optimal" allocation is. If you don't, you are factoring in the good side of a high stock allocation--the high growth rate--and leaving out the bad side--the high volatility in prices that forces sales at the worst possible time. That can't be the right way to go about developing a tool for comparing the merits of the various asset classes, can it?
User avatar
BenSolar
*** Veteran
Posts: 242
Joined: Mon Nov 25, 2002 5:46 am
Location: Western NC

Post by BenSolar »

hocus wrote:
Yes, but I believe that those with higher stock allocations will be pressed to make bigger changes than those with more moderate stock allocations. So you have to incorporate this factor into your analysis of what the "optimal" allocation is. If you don't, you are factoring in the good side of a high stock allocation--the high growth rate--and leaving out the bad side--the high volatility in prices that forces sales at the worst possible time. That can't be the right way to go about developing a tool for comparing the merits of the various asset classes, can it?


I agree that the tool kit needs to include education about the risks you discuss. I'm just not convinced of the best way to do it.

Gotta run, take care.

Ben
"Do not spoil what you have by desiring what you have not; remember that what you now have was once among the things only hoped for." - Epicurus
eurotrash01
* Rookie
Posts: 3
Joined: Wed Dec 11, 2002 5:46 am

Post by eurotrash01 »

hocus said:

"
If you retired in early 1929 with a 74 percent stock allocation, you saw three-quarters of your portfolio lose close to 90 percent of its value in the following years. The idea that most investors in those circumstances would continue to hold a 74 percent stock allocation until prices recovered is just absurd. There is no historical data that I have ever seen lending support to this fantastic assumption.
"

I have to concur here. The studies at the REHP don't fully detail the "pain" well enough because they use year end numbers. The worst of the 73-74 bear market is not expressed by year end 1972 to year end 74. Like this year, there was a nadir in the fall of 74 which was very real for those watching their portfolios. There is nothing sacred about 12/31 other than that it is a convenient point to observe where the data is likely to be cleaner (more people track this.)


Euro
wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer »

I have to concur here. The studies at the REHP don't fully detail the "pain" well enough because they use year end numbers. The worst of the 73-74 bear market is not expressed by year end 1972 to year end 74. Like this year, there was a nadir in the fall of 74 which was very real for those watching their portfolios. There is nothing sacred about 12/31 other than that it is a convenient point to observe where the data is likely to be cleaner (more people track this.)

bogle said it best: psychology kills more portfolios than economics.

it has been telling to read the increasingly frequent posts by even the most ardent us equity stalwarts, and others, those who enjoyed the 18/18 orgy to the fullest, on such arcane topics as real estate, bonds, value investing, ex us equities, storage real estate, and the like. tough (us equity) times will do that. i know we would have been far more concerned if it hadn't been for fairly market cap weighted diversification.

my particular fave, however, is hearing the gnashing of teeth over folks' most important asset: their physical health. listening to the health care insurance challenged of tender years, who self selected for removal from the work force after less than 2 decades, twist in the wind always lightens up my day. i liked seattle pioneer's approach - don't retire until you can afford the health care bill. maybe the investment corollary is: don't invest in equities until you are sure you can afford the bill (ie volatility).

w
regards,

wanderer

The field has eyes / the wood has ears / I will see / be silent and hear
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

Maybe the investment corollary is: don't invest in equities until you are sure you can afford the bill (ie volatility).

That's a great way of putting it, wanderer. There is a natural tension present in informed portfolio allocation analyses. You want the long-term growth that is generally associated with volatile investment classes, but you also want the safety generally associated with the less volatile classes. It's usually a mistake to put all your chips on growth or all your chips on safety. The goal is to strike the appropriate balance between the two that makes the most sense for an investor with your particular life goals and financial circumstances.

Safe withdrawal rate analysis is a powerful tool for dealing with this tension. It allows you to determine what percentage of your portfolio you can afford to invest in the volatile/high-growth classes while still maintaining the level of safety for the overall portfolio that you require. This is a tremendously valuable thing to know.

But you can never figure it out if you insist on doctoring the numbers to give you a pre-ordained result. The only way for the tool to work is to just allow the data to speak for itself.. If you end up not liking what the data says, you can always follow your own investment preferences regardless of what your SWR analysis tells you. But there is no advantage at all to going through a SWR analysis skewed in such a way that all it does is support the allocation that looked good to you before you performed the analysis.

At every stage of his analysis (I don't mean just what's in the study, but also the commentary on the study offered at the REHP board), intercst skews his analysis in favor of stocks. I could give many examples of this, but I'll offer just one to get the basic point across.

You can generally increase the SWR on volatile asset classes by diversifying among them. If the SWR for 80 percent stocks is 3 percent, and the SWR for 80 percent real estate is 3 percent, you might be able to get up to an overall SWR of 4 percent by having 50 percent stocks, 30 percent real estate, and 20 percent fixed income. I am just making up the numbers I am using here to ilustrate the point. But the point is a valid one and an important one. Diversification pulls the SWR up.

Intercst skews every analysis he performs in favor of stocks. He ignores things that pull the SWR for stocks down, and he ignores things that pull the SWR for alternate asset classes up. The result is that his SWR analysis cannot help you with the most important benefit that a valid SWR analysis provides--comparing the relative merits of the various asset classes.

The whole point of the exercise is to determine, How much should I put in stocks and how much should I put in other classes? With intercst, it seems that the purpose is to provide himself with some sort of psychological comfort that a stock allocation that has never been safe in the past will somehow work out OK in the future.

One of the things that attracted me to the intercst web site when I discovered it was intercst's focus on SWR analysis. In my own research, I had found the SWR concept to be of great value, and I was excited to see someone else backing me up on this. I was also fascinated to see that he came up with a withdrawal rate for his plan--4 percent--that was precisely the one I decided on for my plan after years of research. I was disapointed in the flaws I saw in the way he set the study up, but my presumption was that these were just mistakes which anyone can make when doing research on a question that hasn't yet been subject to much scrutiny.

I am now coming to believe is that the limitations in the study were not "mistakes" at all. It appears from a lot of his recent posts that intercst knew all along of the statistical invalidity of his study, and he went ahead trying to get people to place unwarranted confidence in it anyhow.

My guess today is that he knew he wanted to go with a 74 percent stock allocation before he put together his first spreadsheet. I lean towards thinking that the study was a means of lending support to an investment allocation that was a personal preference of the author, and that that is the reason why all the considerations that would have produced undesired results were left out of the equation.
Post Reply