Future Proceedings on the SWR Matter

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

Future Proceedings on the SWR Matter

Post by hocus »

In my last post on the safe withdrawal rate (SWR) matter, which appeared on the "SWR Investigations" thread, I ventured forth my explanation of why the idea that valuation levels affect SWRs has proven so controversial. My belief is that publication of the book "Stocks for the Long Run" produced a sea change in the understanding of the risks associated with stocks.

In earlier times, the general understanding had been that stocks are a high-return asset class, but that they are also a high-risk asset class. Therefore, you often heard recommendations that people invest 50 percent of their assets or so in stocks, not necessarily more. "Stocks for the Long Run" made a remarkable claim and offered historical data to support it. The claim was that, so long as you held stocks for time-periods in excess of 20 years, it was possible to avoid the risk previously thought to be associated with this asset class. Yes, stocks could go down, but over the long term stocks were shown to always be the best investment class.

This claim is the intellectual underpinning for the intercst assertion that it is "irrational" to go with something other than a 74 percent stock allocation. It is also the claim that underlies his assertion that, if you elect to invest in other asset classes, it will take longer to achieve a safe early retirement. These assertions make sense if you believe that stocks always provide the best returns in the long term. They do not make sense if you do not believe that.

My assertion is that William Bernstein's book "The Four Pillars on Investing" reveals that the "Stocks for the Long Run" claim is invalid at the levels of overvaluation that were reached in the late 1990s. Stocks affect returns as a matter of "mathematical certitude," and returns affect SWRs. Do the math to determine the SWR for a high-stock portfolio in the year 2000, and the number you come up with is 2 percent. There were other asset classes offering a SWR higher than that in 2000 (30-year TIPS were paying 4 percent). So it was not "irrational" to retire with something other than a 74 percent stock allocation. It did not take "longer" to achieve a safe retirement with some alternate stock allocation percentage.

I believe this is big news. I believe it has all sorts of compelling implications. Most important for aspiring early retirees is that it reveals that most SWR studies that have been done to the present date are seriously flawed. Most existing studies do not include an adjustment for valuation, and, without such an adjustment, the numbers produced are invalid. The allocations recommended are not "safe." A year 2000 retiree planning to live on $40,000 who went with a 4 percent withdrawal rate would have retired with $1,000,000 less in assets than what he needed for a safe plan, according to what the historical data says.

I believe that thie valuation claim has proven controversial because the mindset that was encouraged by the "Stocks for the Long Run" claim has become deeply embedded in most people's thinking about stocks and risk. The true significance of Bernstein's statement is its implication that the "Stocks for the Long Run" claim is not longer true (or, at a minimum, that is was not true in the year 2000 and that it may not be true again in times to come).

I believe that it is important that we reach a consensus on this fundamantal point. Either SWR analysis that does not include an adjustment for valuation levels is valid or it is not. We really should try to resolve the matter one way or the other. There are lots of aspects of the SWR question that we could examine subsequent to doing that. But I see the valuation question as a root question, a building block for further learning efforts.

I propose that I take a four-month break from posting so that I can catch up on some other work. I recommend that community members with an interest in bringing this controversy to a satisfactory resolution consider using the time between early June and early October to read Chapter Two of "The Four Pillars of Investing." It is probably a good idea to take in Bernstein's full argument to develop a full appreciation of the points in contention, I believe.

I propose that in early October we engage in an effort to reach a consensus on this one aspect of the SWR question, and that we plan to devote up to three months to that effort.

Presuming that we reach a consensus that valuation levels do indeed affect SWRs, I propose that in January we invite four experts on questions of stocks and risk to join us for one night each of discussion of the SWR matter at this board. The four experts I propose we invite for "An Evening on the NoFeeBoards.com FIRE Board" are: (1) William Bernstein; (2) Peter Bernstein (no relation to William); (3) Scott Burns; and (4) gummy.

My idea is that we would, as part of the October through December discussions, identify questions that we would like to pose to these four experts. Each would of course be free to elect to comment on the question or not. I believe that, if we demonstrated as part of the materials we submitted to them with our invitation that we had done our homework before requesting their help, we would be able to get at least two of the four to participate. I believe that a statement from even two of these four would be of significant value. A statement agreed to by three of four or four of four would be that much better, of course.

We would not seek any statement at all unless we came to a consensus that valuation levels affect SWRs. Presuming that we did, the statement sought would be something along the lines that "valuation levels always affect the calculation of SWRs, and it is invalid to claim knowledge of the SWR by making reference to an analysis that does not include an adjustment for valuation levels." We might well request an additional statement that there is nothing "wrong" with issuing a study that does not include an adjustment for valuation levels so long as the study contains language clearly indicating this limitation in its analytical approach.

I believe that this approach to future proceedings would serve to focus our SWR discussions in a constructive way. By no means would it exhaust the aspects of the SWR question that could be profitably explored over time. But it would bring the core controversy to resolution; we would produce a definitive finding as to whether the claim that has been put forward re valuation is a valid one or not. Doing that would serve to set things up for profitable exploration of a host of additional questions, I believe.

I believe that the core matter in dispute here is too contentious to be resolved by engaging in the ordinary sort of discussion board debates. The question is a big one with far-reaching implications. We need to bring in widely recongnized experts to see whether they are willing to verify any conclusions that we adopt in a tentative way. Before we request that these experts give of their time, however, I believe that we are under an obligatiion to do the best we can to reach a consensus among ourselves and to organize and craft any questions that we intend to present to them for their consideration.

Does anyone have any thoughts as to whether the procedure outlined above seems likely to lead in a constructive direction?
User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

Post by ataloss »

My assertion is that William Bernstein's book "The Four Pillars on Investing" reveals that the "Stocks for the Long Run" claim is invalid at the levels of overvaluation that were reached in the late 1990s. Stocks affect returns as a matter of "mathematical certitude," and returns affect SWRs. Do the math to determine the SWR for a high-stock portfolio in the year 2000, and the number you come up with is 2 percent.

I hope this response doesn't sound too defensive. I honestly have no objection if people want to come forward and say "this housing assumption of yours is nuts."

My good friend Bob is pretty sure that you have completely misunderstood Bernstein wrt to "mathematical certitude". His understanding is consistent with what JWR1945 posted. Bob hopes that you will value his honest opinion. I don't like posting this but Bob said that you would prefer this to having an unchallenged misconception. Maybe Bob is wrong. You might consider discussing your interpretation of Bernstein with someone you trust.

FWIW, I think that it makes sense to plan for a lower withdrawals in an era of lower returns.
Have fun.

Ataloss
User avatar
ben
*** Veteran
Posts: 231
Joined: Mon Feb 17, 2003 4:00 am
Location: The world is not enough

Post by ben »

Bob keeps making so much sense to me! I love Bob!! :lol:

That said; have anybody actually MADE the study of SWR including the 3 tier valuations at starting year? (low P/E below 10, medium P/E (around 10), high P/E (above 10) - or whatever levels found fair).

Is is hard to do?
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...
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

My good friend Bob is pretty sure that you have completely misunderstood Bernstein wrt to "mathematical certitude".

The question on the table is whether your friend Bob is willing to have the discussion or not. Ultimately, the question is whether your friend Bob is willing to join in an effort to ask Bernstein himself what Bernstein meant by his statement.

I am willing to have the discussion. I am asking to have the discussion. I am saying that, if we have the discussion, Bob will come around.

His understanding is consistent with what JWR1945 posted.

JWR1945 has independently calculated the SWR for the year 2000 and come to a number very close to the number that Bernstein came to. I believe that JWR1945's number was 2.3 percent. Bernstein's was 2 percent. Intercst's was 4 percent.

I don't like posting this but Bob said that you would prefer this to having an unchallenged misconception.

It saves everyone a lot of time when people post frankly. It is only by acknowleding the sources of disagreement that we can have any hope of resolving them. Bob steered you right on this one.

You might consider discussing your interpretation of Bernstein with someone you trust.

This comment cuts two ways, Ataloss. There are comments that you have posted that suggest a distrust of me.

We are talking about a mathematical calculation here. You put the data for the factors that apply to the question into a calculator and the calculator tells you the result. There should not be too much room for differences of opinion on the core issue.

Both Bernstein and JWR1945 performed the calculation and they independently arrived at similar results. I think they both performed the math properly. The result that they both came to has enormous implications for the project of this board, and I think that we should be making plans to talk through those implications.

If there are some who do not understand why Bernstein and JWR1945 did the math the way they did, we need to take questions on the matter until every last question has been answered to the satisfaction of the community member asking it.

But it would be a mistake for the board to duck the matter because it is too "controversial." It is extremely controversial, and there is a reason for that. The fact that it is so very controversial tells you that it is so very important. When people get this worked up over a mathematical calculation, there is a reason.

There is no other issue that this board could discuss that has more profound implications for its effort to achieve the goals for which the board was established.
wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer »

I'm not looking for reams of pages, but can someone post the exact quote (keeping it short - for the typist's and argument's sake) from Bernstein - that 2% was the SWR for a 75/25 portfolio in 2000. Which seems to be what you are saying, hocus. Non?
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 »

Can someone post the exact quote (keeping it short - for the typist's and argument's sake) from Bernstein - that 2% was the SWR for a 75/25 portfolio in 2000.

The quote is on Page 234 of the book "The Four Pillars of Investing."

It reads: "A particularly bad returns sequence can reduce your safe withdrawal amount by as much as 2 percent below the long-term return of stocks. Recall from Chapter Two that it's likely that future real stock returns will be in the 3.5 percent range, which means that current retirees may not be entirely safe withdrawing more than 2 percent of the real starting values of their portfolios per year!"
wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer »

I'll take it line by line -

A particularly bad returns sequence can reduce your safe withdrawal amount by as much as 2 percent below the long-term return of stocks.

Not sure why he is relating SW(R?) to long-term return (I assume he means real return). Returns and SWRs are two different animals and there is a relationship, but not lock step. For example, the return could be X%, but a steady X%, above 4%, and, although lower than the long-term real return from US equities (6.5%-7%) it could lead to a higher SWR. That's the charm of REITs. Volatility is the downfall (and savior) of equities.

Recall from Chapter Two that it's likely that future real stock returns will be in the 3.5 percent range,

he makes plausible assumptions to get here (assumptions with which I definitely agree - i think raddr makes similar ones) but he is definitely not saying this with "mathematical certitude." At least not in the quote (maybe I am being unfair - I confess to not having read the book).

which means that current retirees may not be entirely safe withdrawing more than 2 percent of the real starting values of their portfolios per year!"

retirees will never run out of money if they withdraw 2% of their boy real starting values. As long as there is still some amount in the kitty. If he means inflation adjusted from the first year, then he may be right and that would be a) a very strong statement, b) would seem to directly contradict the "traditional" approach, and c) support hocus' contention 100%. No? Bob?

Although we can engage in semantics, the clear implication of the "traditional" method was that it should provide a reliable guide going forward. (The transcript that hocus puts so much confidence in has periodic, Nixonian gaps exist which make me less sanguine about relying on it as the final arbiter of what was said when.) What is definitely clear is that there was virtually no reference to valuations when I used to check in with some frequency.

Thoughts?
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 »

Bernstein: Recall from Chapter Two that it's likely that future real stock returns will be in the 3.5 percent range,

Wanderer: He is definitely not saying this with "mathematical certitude."

The thing he is saying is true as a matter of "mathematical certitude" is that valuation levels affect returns. We all know that returns affect SWRs. The inescapable conclusion is that valuation levels affect SWRs. Our task becomes--how do we include in our calcuations recognition of this thing that we know as a matter of "mathematical certitude" to have an influence on the question we are trying to answer?

Bernstein does it through use of the Discounted Dividend Model (DDM), a methodology that he describes as superior to the historical sequence methodology used in most existing SWR studies. It is the DDM methodology that provides him the 3.5 number.

I don't think of the 3.5 number as a prediction. I think of it as a data point. Just as the historical sequence methodology uses data points to generate a result, so does the DDM methodology. Bernstein didn't pull the 3.5 number from the air. The number is provided by making use of the superior methodology.

He is not making predictions. He is using data to make assessments of the probabilities of various future possibilities. That's how you determine the SWR, you look to data to make assessments of the probabilities of various future possibilities.

The 3.5 percent figure is the number you need to do a valid calculation of the SWR for a retirement beginning in the year 2000. I am saying that an analysis that does not include this step is an invalid analysis.

Now, Bernstein does not directly say that existing SWR analyses are invalid. If he had said that, I presume that we would not be having this discussion. It is hocus who is making that claim, no one else that I know of.

What Bernstein says directly is close to what JWR1945 says. He says directly that the historical sequence methodology is "sometimes misleading." JWR1945 is saying that accounting for valuation is the superior approach, so he is saying something very similar.

I am saying that the necessary logical implication of Bernstein's claim that valuation levels always affect returns (and thus SWRs) is that the historical sequence model is not only inferior but out-and-out invalid. I am going one step beyond what Bernstein said directly. I am saying that, if you accept what he said directly, you are led inevitably to the further conclusion that the right way to calculate SWRs is to consider the effects of changes in valuation levels.

This is why I am suggesting that we invite Bernstein to discuss the matter with us and comment on the question at hand. I would like for the community to work toward development of a clear and concise statement as to what we think is the proper way to do SWR analysis, or at least a clear identification of our concerns and uncertainties on the question, and ask Bernstein and some others to comment on it. I would like him to expand on the comments he has already put forward. I would like him to address this question of whether a logical follow-up to what he has already said is that valuation levels must be considered in a valid SWR analysis.

If someone says to you "Tell me the sum of these four numerals--2, 3, 7, and 8," is it valid to say "the sum is 10?" If you are challenged on that response, and you respond that "the numerals 7 and 3 most certainly do add up to 10, I added properly!" is that a valid response?

I say no. You didn't answer the question that was asked. A valid answer to the question "What is the SWR at such and such a date?" is one that looks at all factors that bear on the question. The valuation level that applies on such and such a date is a crtiical factor. Ignore that factor, and it is not possible to arrive at the correct answer to the question asked.

(Maybe I am being unfair - I confess to not having read the book).

Bernstein is saying things that are shocking departures from what you usually hear in discussions of stock investing. I think it would be a big help if more people would read the book. At least Chapter Two, which is the one that explains how and why he calculated the SWR at 2 percent.

When you add up 2, 3, 7, and 8, you get 20. When you only add up only 7 and 3, you get 10. "10" is not just an "inferior" answer to the question asked, it is outright wrong. I am saying that "2 percent" is not just a "superior" statement of the SWR in the year 2000. I am saying that the answer "4 percent" is outright wrong.

If both answers are right enough to be considered valid, then it is possible for one valid SWR to call for savings of $1,000,000 less to support annual spending of $40,000 than another valid SWR. That difference is so large that it calls into question whether SWR analysis really tells you anything. Accept that answers that vary that widely are both valid, and you have a tooln that is so elastic as to be virtually useless.

I am saying that, when you discover that a factor always effects the result, it is not just a "nice to include" item, it is a "must include" item. The proper response to learning that valuation levels always affect SWRs is to adopt a practice of always giving recognition to this factor in your calculations.
galagan
* Rookie
Posts: 5
Joined: Thu Mar 20, 2003 3:00 am
Location: Billings, MT

Post by galagan »

hocus wrote: JWR1945 has independently calculated the SWR for the year 2000 and come to a number very close to the number that Bernstein came to. I believe that JWR1945's number was 2.3 percent.


Could someone please point me to JWR1945's calculations and discussion of his methodology?

thanks,
dan
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

Could someone please point me to JWR1945's calculations and discussion of his methodology?

Here is a link to a recent post in which JWR1945 discusses his earlier analysis. I don't have ready access to a link to the analysis itself.

http://www.nofeeboards.com/boards/viewt ... ght=#p6777
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

For galagan. This is from raddr's thread about Of tainted history, data torture, and faux SWR's which he started on Sun, Dec 15, 2002 at 2:59 pm CST. This is a copy of my (JWR1945) post:

[It was for] Andrew61

There are several ways to back up what raddr and the others have said. One way to understand what is going on is to use dory36's safe withdrawal rate calculator at: www.capn-bill.com/fire/ It is easy to use. It uses the same information as the Retire Early Home Page calculator (but without the new monthly breakouts). Look at the detailed results for a variety of withdrawal rates and time periods. Use 4.5%, 5%, 6% withdrawal rates, withdrawal periods of 30, 40 and 50 years and stock allocations of 50%, 65% and 80%.

The results are clear and visual. Failures are in red. Failures are grouped together. They are not randomly distributed. They cluster around the early 1900s, the Great Depression years (1929 and the 1930s) and the 1960s and early 1970s. I do not know the details about the early 1900s. It is easy to identify the crash of 1929 and the Great Depression. It is also easy to recognize the effects of the high inflation/stagflation years.

Adding to raddr's pie analogy, consider two sets of data. In the first set you record stock prices every minute inside of a single 5-hour period. That has 300 data points. In the second set you use weekly stock prices published in Barron's over a 5-year period. That has 260 data points. The information in each set is different. It is not simply 300 data points versus 260.

Various safe withdrawal rate approaches and their calculators are complementary. I have already posted an overview on this board with my post FAQ starter kit. I wrote that in support of hocus's the Safe Withdrawal Rate investigations. They are scheduled to start in mid-January of next year.

As for making projections in this era of historically high valuations, I prefer to rely on stock market earnings (using Professor Shiller's ten-year averages and reflected in his P/E10 data). My hypothesis (i.e., guess) is that stock market earnings support the true safety of withdrawal rates. The peak of the historical range of the SP500 P/E10 ratio is just below 25. There have been only a couple of brief exceptions prior to the 1990s. My hypothesis is that the actual safe withdrawal rate will be close to (25/the current SP500 P/E10 value when over 25)*(the historically based numbers around 4%). That is, the earnings (averaged over the last ten years) that support a 4% withdrawal rate correspond to a P/E10 of 25.

Here is another thought. The SP500 P/E10 peaked at 44 in 2000. If my hypothesis is correct, a safe withdrawal rate for 2000 would have been around 2.3%.

This is just one of the many issues that we will address next month.

Have fun.

John R.
User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

Post by ataloss »

My hypothesis is that the actual safe withdrawal rate will be close to (25/the current SP500 P/E10 value when over 25)*(the historically based numbers around 4%). That is, the earnings (averaged over the last ten years) that support a 4% withdrawal rate correspond to a P/E10 of 25.


Could this apply to times of undervaluation? Looking at gummy's Maximum Rate of Withdrawal © I would guess that the market was undervalued in 1950 resulting in a 9.75% MRW © for 40 years




( a 10.2% p/e ratio would make your formula work)
Have fun.

Ataloss
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

I have just posted my comments about William Bernstein's numbers on the Great SWR Investigation-Part 1 thread. I have finally figured out exactly what he did.

http://nofeeboards.com/boards/viewtopic ... 6953#p6953

As for a night at the nofeeboards.com, count me in. Our statistical idiots are light years ahead of the experts.

Have fun.

John R.
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

As for a night at the nofeeboards.com, count me in. Our statistical idiots are light years ahead of the experts.

I agree, JWR1945. Some of the statistical work that has been done at this board is highly significant. We are the experts in this field. When there are circumstances in which it appears that we can provide useful guidance to other experts by sharing what we have come up with with them, and that we can benefit ourselves from seeking some guidance from those other experts, I think it makes sense to pursue something like this.

If we can pull it off, I think it will be a lot of fun. Thanks for those encouraging words.
User avatar
ElSupremo
Admin Board Member
Posts: 343
Joined: Thu Nov 21, 2002 12:53 pm
Location: Cincinnati, Ohio

Post by ElSupremo »

Greetings FIRE folk :)

This is just a suggestion but if you all go through with your "Night at NoFeeBoards.com!" I would be willing to put up a temporary board just for the occasion. Everyone who wants to participate can post there and when it's all said and done I would move all the threads over to the regular FIRE board. I think the event would warrant something along those lines. I would also like to make mention of this on our home page.

Anyway as always let me know if I can do anything to help. I think it's a fantastic idea!
"The best things in life are FREE!"

www.nofeeboards.com
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

Let me know if I can do anything to help. I think it's a fantastic idea!

Thanks so much, ES. Your endorsement of this idea is a great note on which to take my leave of posting for a few months while I attend to some other business. I am becoming confident that we really are going to be able to make this happen.

There are a lot of details to work out. I think it is a mistake to ask someone like Bernstein to help out without thinking through carefully just what it is we aim to gain from his participation. But I believe that in this case it has the makings of a win-win. He obviously has a great interest in the subject matter and this board community has developed some powerful insights on SWRs.

I believe that your idea of using a separate board for the "A NIght With William Bernstein on the NoFeeBoards.com FIRE Board" discussion is a good one. I also agree that we should note the event on the home page. I'll try to think of some other promotional possibilities.

If Gummy would be willing to make note of the event at his site, I am sure that would help bring some people in for the event. We will probably want to have someone put up a post at the Motley Fool board letting that group know. There might be some other things we could do.

The most important thing is to formulate clear and concise and significant questions. We probably should provide a short list of agenda items with the invitation, with an understanding that there will be follow-up questions asked during the board discussion. We might want to put together a package of our best posts for delivery to the four experts along with our invitations to the "A Night With" discussion events. That would give them a good idea of what the board is about and show that we already possess a good bit of background on the issues.

I don't plan to do much more on this until October, when I expect to have more time for posting open up to me. But I will be thinking it over when I go on runs and bike rides, and perhaps will come up with some additional thoughts. Perhaps some others too will have suggestions as to how to make the event nights a success.

This idea excites me. I have maintained for a long time that The Great Debate could lead to good things if energies were channeled in a more constructive direction. I see the prospect of a series of special event debates with SWR experts as potentially a most positive development for the broad community of people with an interest in the subject matter of this board.

Thanks again for your endorsement of the idea, ES. I see this as a great way to make use of the internet's power to serve as a resource for helping aspiring early retirees share ideas and learn together.

Catch the Wave!
WiseNLucky
** Regular
Posts: 84
Joined: Tue Nov 26, 2002 3:59 am
Location: Florida

Post by WiseNLucky »

I think it's a fantastic idea!


I agree. And I bet it will allow us to beat our previous record for most users logged on at one time. 8)
WiseNLucky

I just wish everyone could step back and get less car and less house then they want, and realize they don't NEED more. -- NeuroFool
therealchips
*** Veteran
Posts: 174
Joined: Sat Jan 04, 2003 4:00 am
Location: Henderson, Nevada, USA

Post by therealchips »

John R said:
There are several ways to back up what raddr and the others have said. One way to understand what is going on is to use dory36's safe withdrawal rate calculator at: www.capn-bill.com/fire/ It is easy to use. It uses the same information as the Retire Early Home Page calculator (but without the new monthly breakouts). Look at the detailed results for a variety of withdrawal rates and time periods. Use 4.5%, 5%, 6% withdrawal rates, withdrawal periods of 30, 40 and 50 years and stock allocations of 50%, 65% and 80%.

Thanks, John, for the address. I had forgotten Dory36's work. I ran his calculator with a one million dollar retirement stash, no social security (just for comparison purposes) and no corporate pension (because Dory doesn't allow for such input anyway), 75% allocation to the stock market and 25% in short term interest-bearing accounts, with a withdrawal rate of 4% and a 30 year life-span of the portfolio. The results look like this:
You have proposed a withdrawal of 4.00% of your starting portfolio. . . Your success rate is 100%. . . A withdrawal of about $45,300.00 is the highest withdrawal that would have survived 95% of the periods tested, using the equity split and other criteria you entered. This would be a withdrawal starting at about 4.53% of your starting portfolio, with adjustments for inflation.

Pardon my confusion about raddr's work and whatever others you refer to, but if Dory says that the historical record indicates that 4.5% is safe, how does that result back up raddr's work and what part of that work? :?:
He who has lived obscurely and quietly has lived well. [Latin: Bene qui latuit, bene vixit.]

Chips
User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

Post by ataloss »

Most important for aspiring early retirees is that it reveals that most SWR studies that have been done to the present date are seriously flawed. Most existing studies do not include an adjustment for valuation, and, without such an adjustment, the numbers produced are invalid

I found this excellent response to the valuation question:

intercst : The 4% withdrawal rate is not for all time periods. It is for the single worst 30-year pay out period out of the 100 periods examined between 1871-2000. The 4% withdrawal rate is not based on a mix of data from high, low and fairly valued stock periods. It is based on the single 30-year period that produced the lowest withdrawal rate.

If you retired in any of the other 99 30-year time periods, your safe withdrawal would have been higher. The existing REHP study already has examined cases where a retirement commenced during periods when stocks were high, low, and "fairly valued" -- the 4% withdrawal rate for a 30-year pay out period is the worst case of all of those 100 periods examined.

5/15/02
http://boards.fool.com/Message.asp?mid=17225029
Have fun.

Ataloss
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

therealchips
Pardon my confusion about raddr's work and whatever others you refer to, but if Dory says that the historical record indicates that 4.5% is safe, how does that result back up raddr's work and what part of that work?

My comments were regarding statistical confidence limits. Is the effective number of independent data points the same as the number of start years (for sequences lasting thirty years)? The answer is no. Dory36's visual display makes that clear. An even stronger assertion had been made about a different calculator. raddr showed that it was false.

This is raddr's thread:
Post subject: Of tainted history, data torture, and faux SWR's dated Sun Dec 15, 2002 at 2:59 pm CST.
http://nofeeboards.com/boards/viewtopic.php?t=179

This discussion had to do with the statistics when using the historical sequence method. intercst had updated his Retire Early Safe Withdrawal Rate Study to produce outputs for retirements starting in every month from 1870 until very recently. He then claimed to have a tremendous number of data points for calculating confidence limits.

raddr's point was that the effective number for calculating confidence limits was very low because of a tremendous amount of overlap among sequences. Making monthly calculations does very little in narrowing down the uncertainty of the answers. The variance from N data points decreases by 1/N (or the standard deviation decreases by 1/[the square root of N]) only if all of those data are independent.

(IIRC, somebody...possibly raddr...has used the Monte Carlo method to determine the effective number of independent data points. IIRC, it turned out to be in the neighborhood of 9 to 13.)

When you look at Dory36's outputs, you see that there are very few independent answers. If every start year were independent (as intercst was trying to claim), then the failures (which are displayed in red) would have been distributed randomly. They are not. They are clustered together.

That is the context. Estimating the safety of different withdrawal percentages is a different issue.

Have fun.

John R.
Post Reply