NoFeeBoards.com! No Fees! No Ads! No Spam!

 squared one- from the beating a dead horse department Goto page Previous  1, 2, 3, 4  Next
Author Message
bpp
** Regular

Joined: 26 Nov 2002
Posts: 98
Location: Japan

Posted: Thu Jul 08, 2004 2:16 am    Post subject:

SG,
 Quote: 3) You have not provided any justification or cause-and-effect analysis to explain why the PE10-SWR correlation is or should be linear in the range of observed data. If you bother to look at this, you will see that it doesn't look linear and that it is better fit to a decaying exponential or to a logarithmic function . . . and by the way, the use of one of these functions changes your conclusions dramatically.

FWIW, it is not PE10 vs SWR that is supposed to be linear, it is 1/PE10 vs SWR. It does actually look pretty linear if you throw away all the data before 1923. John R. has given a plausibility argument for why this should be so:
 Quote: Consider these plausibility arguments to explain what is behind the relationship between earnings yield and Calculated Rates (and the Historical Database Rates). These are not proofs. They simply show plausibility. It makes sense for the withdrawal rate to be related to earnings yield because earnings are what supports withdrawals. It makes sense that the relationship between earnings yield and withdrawal rates should be linear, to a good first approximation, since that would be the result if there were no volatility. It makes sense that there should be some randomness in the relationship since the sequence of returns (as opposed to the total return by itself) affects the safety of a withdrawal rate. It makes sense that the relationship between earnings yield and withdrawal rates should be offset from zero since there is a return of capital. Even if a portfolio consisted only of cash (equivalents) matching inflation (with 0% real interest), the portfolio would survive 30 years with withdrawals of 3.33%.

(From the Matters of Interpretation thread on the SWR board.)

As for the 1923 cut-off, he has some plausibility arguments for that, too, though I am not sold.

Bpp

hocus2004
Moderator

Joined: 10 Jun 2004
Posts: 752

 Posted: Thu Jul 08, 2004 2:22 am    Post subject: "It has only been shown to fall into the category of "high-risk withdrawal" according to your (or JWR's) chosen criteria." This is incorrect, bpp. William Bernstein used a different methodology than the one used by JWR1945 and Bernstein also found the 4 percent number to be unsafe at the valuation levels that applied at the top of the bubble. Raddr used yet another methodology and he also found the 4 percent number to be unsafe. There is not one researcher who has done research showing a 4 percent take-out number to be safe at the valuation levels that have applied since the late 1990s. That includes intercst. He says all the time that the 4 percent number is safe, but his study was done at a time before the valuation levels we reached in the late 1990s even applied. The study has no relevance to the question of whether the 4 percent number is safe at current valuation levels. It doesn't even aim to examine the question! There is a reason why no researcher has ever come up with a finding that the 4 percent number is safe at these valuation levels. The reason is that there is no possible way to produce that result through analysis of the historical data relating to the question of what withdrawal rate is safe. The study published at RetireEarlyHomePage.com is not off the mark by just a little bit. It is off the mark by a clear country mile. The study should either be corrected or removed from the site. Today.
hocus2004
Moderator

Joined: 10 Jun 2004
Posts: 752

 Posted: Thu Jul 08, 2004 2:34 am    Post subject: "As for the 1923 cut-off, he has some plausibility arguments for that, too, though I am not sold. " I appreciate you putting forward this observation, bpp. It is clearly being put forward in a constructive spirit, as have been your recent posts at the SWR board. You are doing a lot to help get us back on the right road. The one item on which you are going very much off the rails is in your suggestion that, because the JWR1945 methodology has not been proven out, it is reasonable to continue to consider the methodology used in the REHP study as valid. This simply does not follow. The methodology used in the REHP study is analytically invalid for purposes of determining SWRs. I determined that back in 1996 thorugh my own clumsy (all that I am capable of) examinations of the historical data. If there were any who had doubts about my findings, those doubts should have been dispelled by my posting of the "What Bernstein Says" post on August 27, 2002. From that data forward, there has been no reasonable argument that the methodology used in the REHP study is analytically valid for purposes of determining SWRs. It is possible that the JWR1945 methodology will not be proven out. There is room for reasonable differences of opinion on that question. There are no grounds for reasonable differences of opinion on the question of whether the methodology used in the REHP study is analytically invalid for purposers of determining SWRs. The book was closed on that one a long, long, long, long time ago.
bpp
** Regular

Joined: 26 Nov 2002
Posts: 98
Location: Japan

 Posted: Thu Jul 08, 2004 2:39 am    Post subject: No, Hocus, none of that rises to the level of proof. Bpp
bpp
** Regular

Joined: 26 Nov 2002
Posts: 98
Location: Japan

Posted: Thu Jul 08, 2004 2:49 am    Post subject:

 Quote: The one item on which you are going very much off the rails is in your suggestion that, because the JWR1945 methodology has not been proven out, it is reasonable to continue to consider the methodology used in the REHP study as valid. This simply does not follow.

That's also not what I said.

I think I've said before, on the SWR board even, that I think the whole SWR method of taking fixed withdrawals regardless of portfolio performance is fatally flawed. You and John R. are trying to fix this approach, but I don't think it can be fixed. Try as hard as you might, you're only likely to succeed in attaching epicycles to a fundamentally flawed construction. It may look a little more plausible than the original, but it is not ultimately tenable, in my opinion.

Sorry, but for my money, that's where I see things standing.

Bpp

hocus2004
Moderator

Joined: 10 Jun 2004
Posts: 752

 Posted: Thu Jul 08, 2004 2:59 am    Post subject: "No, Hocus, none of that rises to the level of proof. " Intercst includes in his study a pledge to readers that he will aim to offer "accurate" and "up to date" information on the SWR issue. William Bernstein is a world-recognized authority on the subject of asset allocations. My post of August 27, 2002, revealed to the entire board community (including intercst) Bernstein's views on three key questions: (1) Bernstein rejects out of hand the REHP study's assumption that changes in valuation have zero effect on long-term returns, saying that such an effect exists as a matter of "mathematical certainty;" (2) Bernstein labels the results of the REHP study's methodology "highly misleading;" and (3) Bernstein looks at the historcial data to see what it really says about SWRs and finds that the number which intercst declares to be "100 percent safe" is off by a full two percentage points. Intercst has dishonored his pledge to the readers of his study. He is a con man. His continued presence on the various FIRE boards reflects poorly on each and every one of us because it shows that we have failed in our responsibility to protect those who come to these boards for honest and informed insights on the subject matter from the deceptions of intercst and other defenders of the conventional methodology. You demean yourself by even attempting to defend the actions of this individual, bpp. You have shown in your posts on other topics that you are better than that. Please post responsibly on the subject of SWRs too.
hocus2004
Moderator

Joined: 10 Jun 2004
Posts: 752

 Posted: Thu Jul 08, 2004 3:20 am    Post subject: "I think I've said before, on the SWR board even, that I think the whole SWR method of taking fixed withdrawals regardless of portfolio performance is fatally flawed. You and John R. are trying to fix this approach, but I don't think it can be fixed. Try as hard as you might, you're only likely to succeed in attaching epicycles to a fundamentally flawed construction. It may look a little more plausible than the original, but it is not ultimately tenable, in my opinion. "Sorry, but for my money, that's where I see things standing. " There's nothing to be sorry about re that statement. That statement is a reasonable statement. What you are saying here is that all SWR analysis is analytically invalid. My sense is that that is where TH comes down (more or less) and also where DataSnooper comes down (more or less). I have no problem with people saying that they reject the SWR concept out of hand. I don't agree with this view, but it is not an unreasonable view. If you reject the SWR concept out of hand, it follows that you would see the problem with other posters putting forward posts directing newcomers to the REHP study to determine what is safe. These sorts of posts are still being put forward at the various boards today. They must be challenged. We must get the word out to people who use these boards thinking that honest and informed posting is permitted at them that the REHP study has been completely discredited. You find it discredited for reasons different than the reasons for which I find it discredited, bpp. That's fine. What I need is your help getting the word out that all reasonable people who have looked at the question possess zero confidence in the claims put forward in the REHP study. I'm not trying to sign you up for any big task. I can do much of the work myself. I visit all of the various boards on a more-or-less daily basis. So I can put up posts challenging the REHP study when some uninformed poster cites it as if it were legitimate research. I can handle the heavy lifting on this thing. But I need cooperation from other posters to stop the nonsense word game stuff and the smear campaign stuff. That stuff serves to muddle the key message that we are trying to get through to people, that the REHP study falls far short of doing what it purports to do, identifying the withdrawal rate that the historical data indicates is safe. All that I really need from you, bpp, is that you put up a post from time to time when you see someone engaging in a word game or an intimidation post. If we put an end to that stuff, we will be able to fully develop the case against the REHP study and its author. Then we will be able to point newcomers to that material when we need to ask for help in getting the study taken down and in getting intercst removed from the various boards. What I am asking, bpp, is that you meet your minimal obligations to the board community. Anyone who has ever benefitted from one of the boards has a responsibility to help out when the community is under attack. After 25 months of this nonsense, I think that it is more than fair to say that we are under serious attack re this matter. Can you help, bpp?
ben
*** Veteran

Joined: 17 Feb 2003
Posts: 231
Location: The world is not enough

 Posted: Thu Jul 08, 2004 6:44 am    Post subject: He,he - have a hard time taking this serious - no offense to anybody! Anyway; rather than writing a long rant I find one of my very old posts basically saying: my 2 cents worth: 1. Intercst+other historically based SWR studies are fully valid with the limitations they included/in the historical period (with the respective valuations Etc.) they covered. This is a historical fact. 2. I/Hocus/JWR/Ataloss and practically everybody on this board DO agree that current/future record valuations DOES matter, as does a bunch of other factors (wars/inflation/interest rates/meteor strikes/corrupted board members Etc.) if trying to PREDICT/guestimate what the future SWR rate will be. 3. JWR have done some great calculations giving indications as to how starting at high valuations historically have affected the SWR. Still 4% was OK historically though - which naturally corresponds fine with Intercst/other historically based SWR studies. 4. With no mathematical certitude can the future SWR be predicted but one can decide to use JWRs fine historical nos to make his/her own guestimates/adjustments due to valuations (leaving wars/inflation/interest rates/meteor strikes/corrupted board members Etc. for another study) of the future SWR._________________Normal; to put on clothes bought for work, go to work in car bought to get to work needed to pay for the clothes, the car and the home left empty all day in order to afford to live in it...
hocus2004
Moderator

Joined: 10 Jun 2004
Posts: 752

 Posted: Thu Jul 08, 2004 6:57 am    Post subject: "Intercst+other historically based SWR studies are fully valid with the limitations they included/in the historical period (with the respective valuations Etc.) they covered. This is a historical fact." How can it be an historical fact that the SWR at the top of the bubble was 4 percent and also be an historical fact that the SWR at the top of the bubble was 2 percent? The same historical data cannot possibly say two such wildly different things.Either Bernstein got the number wrong or intercst got the number wrong. Given the fact that in 22 months intercst has not put forward a single post offering a reasoned argument that it is Bernstein who got the number wrong, and given that every single data-based post that has been put forward since that time supports the Bernstein position, it is not reasonable at this late date for anyone to claim that it was Bernstein who got the number wrong. It was intercst who got the number wrong. He should either correct the error or take the study down from the RetireEarlyHomePage.com site. He should do that today.
ben
*** Veteran

Joined: 17 Feb 2003
Posts: 231
Location: The world is not enough

Posted: Thu Jul 08, 2004 7:46 am    Post subject:

 Quote: How can it be an historical fact that the SWR at the top of the bubble was 4 percent and also be an historical fact that the SWR at the top of the bubble was 2 percent?

He,he - presuming you are talking about starting with drawals in the year 2000 bubble you can not have historical nos for 30 years showing 2% or 4% as an historical fact. (unless you have a time machine of course!)

_________________
Normal; to put on clothes bought for work, go to work in car bought to get to work needed to pay for the clothes, the car and the home left empty all day in order to afford to live in it...
hocus2004
Moderator

Joined: 10 Jun 2004
Posts: 752

 Posted: Thu Jul 08, 2004 8:02 am    Post subject: "unless you have a time machine of course! " SWRs are calculated using data available at the time the calculation is made. Here is a link to a post from the SWR Research Group board that reports what the historical data says re SWRs for recent retirement start dates. http://www.nofeeboards.com/boards/viewtopic.php?t=2657
ben
*** Veteran

Joined: 17 Feb 2003
Posts: 231
Location: The world is not enough

Posted: Thu Jul 08, 2004 8:17 am    Post subject:

JWR
 Quote: the Safe Withdrawal Rate bottomed out near 1.59%

according to your time machine I better put all in TIPs for next 30 years!

_________________
Normal; to put on clothes bought for work, go to work in car bought to get to work needed to pay for the clothes, the car and the home left empty all day in order to afford to live in it...
salaryguru
* Rookie

Joined: 26 Nov 2003
Posts: 26
Location: Mesa, AZ

Posted: Thu Jul 08, 2004 9:46 am    Post subject:

 bpp wrote: ...FWIW, it is not PE10 vs SWR that is supposed to be linear, it is 1/PE10 vs SWR. It does actually look pretty linear if you throw away all the data before 1923. John R. has given a plausibility argument for why this should be so. . . Bpp

First, recognize that this is only one of 5 points I made. Even if the data was well described by the line (and it's not) you still have at least 4 serious problems in front of you -- problem 1 being insurmountable.

Second, I have looked at the 1/PE10 vs SWR data and it does not look pretty linear if you apply standard sanity checks to the analysis.

When fitting data to curves, you can go a long way to convincing your eye that the data is linear by using 1/x, log(x), 1/y, log(y), . . .plots. Heck, almost all data looks linear if you plot it on a log-log plot.

But if the underlying force that has produced the data is causaly described as a 1/x relationship, then the variations in the data should be consistent with that. In other words, the error bars at one end of the plot should be roughly the same as the error bars at the other end. The data should not exhibit more spread (in the y-axis) as you move out. Similarly, the data should not be bunched up at one end (a lot of points at low x-values) and sparce at the other. Re-examine the 1/PE10 vs SWR plots and see what you think.

Similarly, you should be able to fit a straight line to the lower half of the data and come up with a nearly identical fit that would result from the upper half of the data. This sanity check should happen for any empirical fit, regardless of whether you are using 1/x plots or not.

Just as we can convince our eye that data is linear, we can often rationalize the reasons for that. I can easily provide rationization for why we might expect the data to approach asymptotes in both the x- and y- direction when we plot PE10 vs SWR. That data will pass the sanity checks above too. The conclusion you will arrive at regarding valuation and allocation, however,will be quite different.

_________________
-SG-
bpp
** Regular

Joined: 26 Nov 2002
Posts: 98
Location: Japan

 Posted: Thu Jul 08, 2004 2:51 pm    Post subject: Sigh. Ok, first Hocus: No, I will not join your campaign against Intercst. I have nothing personally against the guy, and don't even post to any of the same message boards that he does. His study is very simple-minded, with clearly spelled out limitations, which while it does not provide an actionable withdrawal plan (as far as I am concerned) is still useful as a minimum-bias study, and gives a basic idea of what kind of ballpark one is playing in. (I also don't find that JWR's investigations provide actionable conclusions, either, but I also don't say that his playing around with the data is completely useless, either.) As far as what a terrible human being he may or may not be, I haven't read enough of his stuff to even develop an opinion on that. I do know that he has never come to the NFB, hijacked threads, and poisoned the atmosphere with tirades against someone who does not even post here. Oh, and regarding that Bernstein quote: Bernstein has a rather hyperbolic writing style. I remember him making an analogy once to tossing lumber into a jet engine, for example. I wouldn't take his "mathematical certainty" line too literally. SG: yes, I know you made other points, and I am not trying to say that what John R. is doing is correct. I think he has gone over the edge of data-mining, and have tried to argue with him about what he can really say. He is convinced his assumptions are reasonable. Fine. It's his money, he can do what he likes with it. (Incidentally, I'm not convinced that error bars have to be the same size in all regions of the data. One could maybe try x vs y/x or something if that bothers you. But I don't think that is where the fundamental problem lies, either.) This is an internet chat board. It's supposed to be fun. Bpp
ataloss
**** Heavy Hitter

Joined: 25 Nov 2002
Posts: 559

Posted: Thu Jul 08, 2004 4:06 pm    Post subject:

 Quote: It's his money, he can do what he likes with it.

jwr is far to wise to use some datamined nonsense, he has a pension that meets his needs and invests in stocks for the next generation

_________________
Have fun.

Ataloss
salaryguru
* Rookie

Joined: 26 Nov 2003
Posts: 26
Location: Mesa, AZ

Posted: Thu Jul 08, 2004 10:01 pm    Post subject:

 bpp wrote: Sigh. . . . SG: yes, I know you made other points, and I am not trying to say that what John R. is doing is correct. I think he has gone over the edge of data-mining, and have tried to argue with him about what he can really say. He is convinced his assumptions are reasonable. Fine. It's his money, he can do what he likes with it. (Incidentally, I'm not convinced that error bars have to be the same size in all regions of the data. One could maybe try x vs y/x or something if that bothers you. But I don't think that is where the fundamental problem lies, either.) This is an internet chat board. It's supposed to be fun. Bpp

Hi Bpp,

Of course the error bars are whatever they are. But my point was simply that if the error bars increase or decrease significantly and monotonically along the curve, then the rationalization for the curve fit ought to account for the change in error bars also. The data plotted on the 1/PE10 vs SWR chart is not what one would typically hope to see if they were looking for underlying cause-and-effect relationships.

I actually played around with JWR's data and some functional curve fits that made more sense to me (ie SWR approaching an assymptote for large PE10). The fit is much better than a forced linear relationship, but it still has error bars that are much too large to be of practical value (greater than + or - 1%SWR).

The large error bars are not surprising if you believe, like I do, that PE10 probably captures only a small part of the observed variation in SWR. PE10 is certainly not an independent variable and probably not even be a primary contributor to the historicaly observed variation in SWR. This is why a principle componenet analysis would be useful. I can think of a number of economic metrics that might play a significant role in the historical SWR results.

Fun . . . huh? That sounds like a good idea.

_________________
-SG-
hocus2004
Moderator

Joined: 10 Jun 2004
Posts: 752

ben
*** Veteran

Joined: 17 Feb 2003
Posts: 231
Location: The world is not enough

 Posted: Fri Jul 09, 2004 2:31 am    Post subject: He he - hocus you crack me up!! Can't believe the energy you put into this! I would however wish that you did it on the SWR board as the scroll wheel on my mouse just fell off! _________________Normal; to put on clothes bought for work, go to work in car bought to get to work needed to pay for the clothes, the car and the home left empty all day in order to afford to live in it...
hocus2004
Moderator

Joined: 10 Jun 2004
Posts: 752

 Posted: Fri Jul 09, 2004 2:43 am    Post subject: "Can't believe the energy you put into this! " I will be starting a thread at the SWR Research Group board in coming days in which I will demonstrate the power that these boards have to help middle-class workers achieve their most important life years. When FIRE discussion boards permit honest and informed on-topic posting, they are capable of doing wonderful things for large numbers of people. These boards matter. They are worth saving. "I would however wish that you did it on the SWR board as the scroll wheel on my mouse just fell off!" My preference is that discussions of the SWR matter be held at the board that was created for the purpose of facilitating such discussions. But I do not have magic powers that permit me to force other posters to put up their SWR posts over there. If you would like to take on the task of asking that posters put their SWR posts up at the other board, I think that would be just great.
ataloss
**** Heavy Hitter

Joined: 25 Nov 2002
Posts: 559

 Posted: Fri Jul 09, 2004 5:28 am    Post subject: good grief, I am even losing interest in this too bad th was driven off the board although now that he isn't around to defend himself I see that this is being recharacterized as a victory over a defender of the convential methodology _________________Have fun. Ataloss
 Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First
 All times are GMT - 9 HoursGoto page Previous  1, 2, 3, 4  Next Page 3 of 4

 Jump to: Select a forum General----------------NFBTown CenterTest Posts Personal finance----------------FIRE BoardIndex Funds BoardSWR Research GroupLBYMGeneral FinanceNewbies Board Community----------------WorkspacePolitically CorrectYour HealthGummy Stuff
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum