SWR as a Tool

Research on Safe Withdrawal Rates

Moderator: hocus2004

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

SWR as a Tool

Post by JWR1945 »

In its primitive form, the Safe Withdrawal Rate tool had these critical attributes:
1) All investments were normalized to a very high level of safety.
2) The Safe Withdrawal Rate of each was determined in as objective a manner as possible, preferably using mathematical calculations.
3) The upward potential of each investment was determined (even if it was only a crude approximation).
4) The degree of safety of each investment was determined (even if it was only a crude approximation).
5) Judgments, rules of thumb and guesses were kept external to the determination of Safe Withdrawal Rates.
6) Individual preferences and personalities were kept external to the determination of Safe Withdrawal Rates.
7) All factors in the historical database with significant information regarding long-term investment performance were considered.

Later, hocus applied the Safe Withdrawal Rate numbers to determine his investment allocations. These criteria applied:
1) His nest egg should remain the same or gradually increase with time. It was OK if there were occasional losses as long as they were only temporary.
2) He planned on maintaining his portfolio indefinitely (as opposed to a fixed number of years).
3) He did not want to abandon long-term growth entirely just to obtain the highest levels of safety.
4) He addressed the growth potential of his investment choices from a long-term perspective. He viewed a portion of his holdings in his bonds and certificates of deposit as only a means of preserving capital. They were to be deployed later in growth-oriented investment classes for the long-term when conditions became favorable.
5) He examined the health of his portfolio and its allocations in detail at the start of each year.

This was the original tool. Normalizing to a high level of safety and making objective calculations of each Safe Withdrawal Rate were key.

The Safe Withdrawal Rate tool is not a calculator. We have supplemented it with information from our calculators. We have introduced confidence limits. We try to extract as much information from the historical data as possible. We like to look at the data from many different vantage points.

Have fun.

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

Post by JWR1945 »

Hocus has been remarkably consistent in the description of his SWR Tool. It is only recently that we have reached enough clarity so as to describe it properly.

There have been three or four different things that people have called Safe Withdrawal Rates. The necessary clarity to describe the SWR Tool properly was not available prior to the existence of this board. In fact, I have used the term improperly as recently as the beginning of this year. I initially called the Calculated Rate, which is also known as the Zero Balance Withdrawal Rate, a Safe Withdrawal Rate. It is not. To guarantee a high level of safety, it is necessary to use the lower confidence limit relative to the Calculated Rate.

Some people confuse the term SWR Tool with the calculators that support it. Instead of debating the point, hocus has chosen to add clarity by changing the name of his invention to the Data Based SWR Tool.

Make no mistake about it. Hocus invented the Data Based SWR Tool back in 1996. I have contributed by adding clarity and supporting information drawn from our calculators.

Have fun.

John R.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

I much appreciate you putting up these two posts, JWR1945. The most important work before us today is to put an end to the unwarranted disparagement of the Data-Based SWR Tool that defenders of the conventional methodology have come to rely on as their primary strategic thrust. The logic seems to be that, if we can detroy the reputation of the person who brought us the tool, we can destroy confidence in the tool. It is an ugly strategy, it is beneath our normal community standards, and it does great damage to the hopes of aspiring early retirees to achieve their most important life goals. Your two posts above help to set the record straight for those trying to make sense of these discussions.

I want to add a third post from you to the thread because there are many here now who were not here when you first posted it last October. This one too sets the record straight on the "Did hocus offer anything of value?" question. I noted in an earlier post today that I don't claim to be an investing expert. What I am is an SWR expert. Your post below sets forth the testimony of a Numbers Guy that that particular claim of mine is an accurate one.

I didn't set out in late 1995 determined to become an SWR expert. All that I set out to do was to discover what the historical data says is safe. The process I followed was similar to the one that BenSolar told us the other day that he followed in more recent days. He became interested in the valuation question as a result of the discussion board debate that I kicked off, he checked out Shiller's work and some other stuff, played with the numbers, and came to the conclusion that he had better change his allocation before he suffered some serious life setbacks.

That's how it was with me. I wasn't out to "get" intercst. I had never heard of intercst at the time. I was out to craft a successful Retire Early plan. Knowing the SWR for stocks helps in doing that. So I looked into the question. While doing that, it became clear to me that the valuation level in effect at a particular point in time has a critical effect on the withdrawal rate that survives for the 30-year historical sequence following from that point in time. So I naturally took that factor into account in the work I did in this area.

You can't get the right answer to the question of "what is safe?" without taking changes in valuation into account. If you ignore a critical piece of data when doing an analysis, the answer you get is wrong. That is why all the analyses of SWRs that have been done since the late 1990s show SWRs for high-stock-percentage portfolios that are not even remotely in the neighborhood of the number reported in the REHP study.

When I say that I am an SWR expert, I am not saying that I am a statistical expert. I am in fact a a statistical dunce. What I bring to the table is a more sophisticated understanding of the SWR concept than other community members. I expect that others are going to catch up to me and pass me in days to come. I want to see that happen as soon as possible. When people become more knowledgeable than me, I teach less and learn more.

Right now, a lot of the posts that go up on the SWR question make otherwise smart posters sound like they don't know what they are talking about. We demean ourselves with a lot of the gibberish we have posted in the past 25 months. The SWR is a data-based construct. The SWR cannot be determined by taking opinion polls. The SWR can only be determined by looking at the historical data. The SWR is whatever the data says it is.

I hope that more people will begin acknowledging these realities in days to come. There is a lot of exciting stuff that we do not yet know about SWRs, and we should all be working together getting about the business of learning it. We can't move on to the advanced stuff until we reach consensus on the ABCs. It should not be a controversial statement to say that the SWR is what the data says it is. That is as basic as it gets. So we should try real hard to reach agreement on that and get that one behind us. I am embarassed for the community that we have not yet managed to pull that one off.

Here's the JWR1945 post from last October:

JWR1945 Legend
Joined: 26 Nov 2002
Posts: 1030
Location: Crestview, Florida
Posted: Sun Oct 26, 2003 3:18 pm Post subject: Why I support hocus
------------------------------------------------------------------------
Why I support hocus

1. Several people claim not to understand why I support hocus as strongly as I do. They deride his claims to be the person most knowledgeable about Safe Withdrawal Rates. They see that as arrogance, not fact. They dismiss his contributions in keeping the research alive, as if his efforts were insignificant.

2. Most people accept the notion that hocus has been following the topic for a long time. Most people will concede that he really does have 40 notebooks filled with articles and other reference material that he has collected over the years. Many people will accept the idea that hocus has been making use of Safe Withdrawal Rates in his personal financial planning.

3. Many people seem to reject the idea that hocus has a lot of ideas and thread starters that he has not yet presented. I think that he does have them. I do not think that they are all finished products. I do think that they all contain something useful and worth developing.

4. From my personal experience, there have been numerous times when I have thought that I had addressed something adequately, but I did not. At times I have been surprised to see hocus hang in there on what seemed to be a minor detail. Yet, as the discussions have gone back and forth, there has emerged something of importance, an insight of considerable value.

5. Thus it was with The Coin Toss post. Hocus had asked me for my comments as he developed his thoughts. After repeated refinements, came up with a true gem. And later, when he finally succumbed to my repeated requests to post once again, he pulled it out of his files and posted it at the Motley Fool, where it received much acclaim.

6. What truly impresses me is the consistency that hocus points us to something of value. Yes, I do believe that hocus is the most knowledgeable person when it comes to Safe Withdrawal Rates. I have seen too much evidence to believe otherwise. And, of course, I would not have engaged in any research on the matter if it had not been for hocus's contribution in keeping the discussion alive. Nor would I have become aware of the efforts of the others on these boards.

Have fun.
bpp
** Regular
Posts: 98
Joined: Tue Nov 26, 2002 6:46 am
Location: Japan

Post by bpp »

John R.,

You seem to me like someone who generally knows what he is talking about. You also seem to have taken it upon yourself to explain what Hocus is talking about to everyone. Some things I don't understand about this tool:
1) All investments were normalized to a very high level of safety.
"Normalize"? The tool somehow rescales investments to unity or something? I don't understand the meaning of this.
2) The Safe Withdrawal Rate of each was determined in as objective a manner as possible, preferably using mathematical calculations.
Does this mean that each asset is treated separately? The SWR is not calculated for the portfolio as a whole? Why is it done this way? Seems you would get a super-conservative answer this way (no reduction in volatilities from mixing uncorrelated assets).
7) All factors in the historical database with significant information regarding long-term investment performance were considered.
What are the factors (PE10, dividend yield...?)?. More importantly, how do you know that you got all of them?

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

Post by JWR1945 »

bpp wrote:1) All investments were normalized to a very high level of safety.
and
"Normalize"? The tool somehow rescales investments to unity or something? I don't understand the meaning of this.
This explains hocus's initial question that set off such a furor back in May 2002. It does not explain the subsequent reaction. He asked what were the Price-Adjusted Safe Withdrawal Rates of stocks. He had rejected the idea that it was always 4%, regardless of price (or valuation).

In his own personal planning, he had not previously needed to know anything more than the fact that they carry risk, especially at bubble level valuations. His alternatives carried no risk at all along with attractive interest rates, making this part easy.

Hocus was wanting to know how to allocate money that would be coming from his 7% (nominal) interest rate certificates of deposit as they matured. He wanted to buy stocks. He was trying to find out from others what the Safe Withdrawal Rate of stocks were. He was asking for assistance. He needed numbers that would insure a high level of safety.
2) The Safe Withdrawal Rate of each was determined in as objective a manner as possible, preferably using mathematical calculations.
and
Does this mean that each asset is treated separately? The SWR is not calculated for the portfolio as a whole? Why is it done this way? Seems you would get a super-conservative answer this way (no reduction in volatilities from mixing uncorrelated assets).
I doubt that hocus ever rejected the notion of calculating the Safe Withdrawal Rate of his portfolio as a whole. It was more of a matter of what we can calculate.

There is also the process of applying Safe Withdrawal Rate information. That is where personal opinions, judgments, guesses and rules of thumb are introduced. The final result can be as conservative or as risky as suits the individual.
7) All factors in the historical database with significant information regarding long-term investment performance were considered.
and
What are the factors (PE10, dividend yield...?)? More importantly, how do you know that you got all of them?
It is difficult to phrase this precisely. All factors known to have significance were considered to the extent possible. Nothing has been rejected out of hand.

The most important, single element missing from alternative approaches was the fact that prices matter. It may seem obvious, but it was repeatedly rejected out of hand as being untrue.

As time has progressed, we have been able to extract additional information from the historical database. P/E10 and dividend yield were both successful. We have looked at Tobin's Q and regular (single-year) P/E and comparisons with interest rates, but they have not turned out to be as good.

In addition, while the conventional methodology is critically dependent upon extreme conditions that have produced worst case results, we have been able to make use of all of the data. Look at my posts on using the earnings yield (of P/E10, which is 1/[P/E10]) to come up with Calculated Rates along with their lower confidence limits, which are the Safe Withdrawal Rates. Those curves make use of every element in the historical database.

Have fun.

John R.
bpp
** Regular
Posts: 98
Joined: Tue Nov 26, 2002 6:46 am
Location: Japan

Post by bpp »

Hi John,

Thanks for the reply. Still some confusion remaining:
1) All investments were normalized to a very high level of safety.
and
"Normalize"? The tool somehow rescales investments to unity or something? I don't understand the meaning of this.
This explains hocus's initial question that set off such a furor back in May 2002. It does not explain the subsequent reaction. He asked what were the Price-Adjusted Safe Withdrawal Rates of stocks. He had rejected the idea that it was always 4%, regardless of price (or valuation).
Sorry, I still don't understand. I think this may be a simple English comprehension problem for me: I don't know what it means to "normalize an investment." To me, the word "normalize" means to rescale something. I can picture, for example, comparing investments using expected return divided by variance (as in an MVO). Is it something like this? Or something like dividing price by some measure of valuation? (For example, looking at PE10.) But if that's the case, what does it then mean to "normalize to a very high level of safety"?

Or is this some other meaning of the word "normalize," as in normalizing trade relations, or something else?
I doubt that hocus ever rejected the notion of calculating the Safe Withdrawal Rate of his portfolio as a whole. It was more of a matter of what we can calculate.
I don't get this. He is using historical data, right? Can't he simulate a mixed portfolio just as easily as he can a single asset class? But he always seems to make a point of calculating the SWR for each asset separately. It seems to be a unique feature of his approach; I've never seen anyone besides him discuss the SWR of individual asset classes in the case of a mixed portfolio. Is this not a significant feature of the tool?
It is difficult to phrase this precisely. All factors known to have significance were considered to the extent possible. Nothing has been rejected out of hand.

The most important, single element missing from alternative approaches was the fact that prices matter. It may seem obvious, but it was repeatedly rejected out of hand as being untrue.
Ok, this part I think I understand. I just worry that there is not some other element that has not been missed. There is the famous example of dividend yields of stocks versus bonds, and how that popular measure of valuation became useless after the 1950s. Other factors took over. How certain can you (or Hocus) be that something like that won't happen again with any of the currently fashionable valuation measures, for either better or worse?

Cheers,
Bpp
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

"How certain can you (or Hocus) be that something like that won't happen again with any of the currently fashionable valuation measures, for either better or worse? "

We are in the early stages of our work on this. My expectation is that we will continue refining and enhancing the tool for many years to come.

That said, I don't want to understate the importance of what we have already accomplished. We have demonstrated beyond any reasonable doubt that the numbers generated by the conventional methodology studies are inaccurate. That's a big deal. We used to think the conventional methodology SWR was a safe take-out number. Now we know that at current valuation levels it is a high-risk number.

That's a breakthrough insight. We need to build on it, and we could use the help of as many community members as are willing to help out to do so. But the initial insight was a breakthough insight. The Retire Early movement entered a new era on May 13, 2002. We entered an era in which we have begun to talk about Retire Early investing strategies in more realistic ways.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

bpp wrote:Sorry, I still don't understand. I think this may be a simple English comprehension problem for me: I don't know what it means to "normalize an investment." To me, the word "normalize" means to rescale something. I can picture, for example, comparing investments using expected return divided by variance (as in an MVO). Is it something like this? Or something like dividing price by some measure of valuation? (For example, looking at PE10.) But if that's the case, what does it then mean to "normalize to a very high level of safety"?
To normalize to a very high level of safety answers this question:

What withdrawal rate can I plan on using with a very high confidence of not running out of money within a specified number of years (e.g., 30 years)?

This restates the risk and reward trade off that we are all familiar with into terms that retirees need to know. It is not the same as saying that the long-term real return of the stock market is 6% to 7%. Downward volatility can eliminate a retiree's portfolio long before conditions are favorable enough to get back to the long-term return.

In fact, the Trinity Study and the Retire Early Study made that clear.

The key is that a retiree makes withdrawals. He is interested in more than an initial balance and a final balance when there are no withdrawals.

The problem with the earlier studies is that they stopped at a single number solution (e.g., 4%). They did not look into cause and effect nor did they look for failure mechanisms. They did not search for ways to monitor one's portfolio. If their studies had an unknown flaw, they would not provide any kind of warning.

We have shown that current valuations and dividend yields are well outside of the historical range used in those studies. The older 4% number is not safe.

Our research has shown that the first decade (actually, eleven years) is the critical period. Usually, but not always, a portfolio will have grown enough to assure future success or already be in danger within eleven years. That kind of information had not been known earlier.

Notice that this is not simply normalization by dividing by a standard deviation. Nor is it simply a scaling according to P/E10 (although that was part of my first approach). In some cases, such as with dividend based design, we can transfer the concern from a long-term projection (i.e., the long-term stock market return) to a short-to-medium term projection (i.e., the growth and stability of dividends).

Have fun.

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

Post by JWR1945 »

bpp wrote:I don't get this. He is using historical data, right? Can't he simulate a mixed portfolio just as easily as he can a single asset class? But he always seems to make a point of calculating the SWR for each asset separately. It seems to be a unique feature of his approach; I've never seen anyone besides him discuss the SWR of individual asset classes in the case of a mixed portfolio. Is this not a significant feature of the tool?
No. We have reliable information only for the S&P500 index and commercial paper.

We do not have reliable information for other asset classes. They provide us with some useful information about inflation indexed securities (TIPS and ibonds) held to maturity. It is possible that we could modify our existing calculators to tell us something useful about traditional treasuries (notes and bonds) when invested intelligently as opposed to being treated as single-year trading vehicles.

They do not provide us with reliable information on other asset classes. With REITS the problem is that the relevant history is short. With some others, such as high income stocks, the appropriate information has not been broken out (and there would be time-related issues as well). There are problems with the remainder as well.

Have fun.

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

Post by JWR1945 »

bpp wrote:Ok, this part I think I understand. I just worry that there is not some other element that has not been missed. There is the famous example of dividend yields of stocks versus bonds, and how that popular measure of valuation became useless after the 1950s. Other factors took over. How certain can you (or Hocus) be that something like that won't happen again with any of the currently fashionable valuation measures, for either better or worse?
This is an advantage of our approach. We try to identify failure mechanisms. We try to send out warnings early if our analysis has broken down. Frequently, we translate the issue of long-term safety into terms that we can evaluate in the near-term (such as the quality of dividends).

We cannot guarantee that we will find everything. But we try.

Have fun.

John R.
bpp
** Regular
Posts: 98
Joined: Tue Nov 26, 2002 6:46 am
Location: Japan

Post by bpp »

Hi John,

Ok, so what I'm getting is that the "single-asset SWRs" are just the byproduct of not having any long-term data for anything except S&P500 and commercial paper, and are not an essential feature of the tool. Also, you don't claim that you have found all the factors that impact on the SWR calculation yet; this is reasonable-sounding, as I don't see how, in principle, one ever could anyway.

As for "normalizing to a very high level of safety," that seems to be a way of saying that you are constantly seeking out new failure modes?

Hi Hocus,
We have demonstrated beyond any reasonable doubt that the numbers generated by the conventional methodology studies are inaccurate.
If I've understood John R.'s replies above, that seems to overstates the case a bit. If you can't guarantee that you've found all the bad things that could happen, it also stands to reason that you might have missed some good things that might happen, no?

Bpp
User avatar
salaryguru
* Rookie
Posts: 26
Joined: Wed Nov 26, 2003 7:14 am
Location: Mesa, AZ

Post by salaryguru »

JWR1945 wrote:. . . We have shown that current valuations and dividend yields are well outside of the historical range used in those studies. The older 4% number is not safe.

. . .

The fact that valuations based on backward looking metrics are high is well-known.
That fact does not logically imply your conclusion in the second sentence.

Your reasoning is flawed here. Empirical descriptions can be very useful when the the cause-and-effect relationship is too complex to describe analytically. But you cannot justify the application of an empirical relationship outside of the range of data used to establish that description. Further, you cannot justify even the functional form of the empirical relationship without cause-and-effect models.
-SG-
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

"...and are not an essential feature of the tool."

The defining characteristic of the Data-Based SWR Tool is that it is objective. The conventional methodology tool purports to be objective, but in reality it is not (at least not after May 13, 2002). When a researcher deliberately leaves out of his or her analysis a factor known as a matter of "mathematical certainty" to affect the result of the question being examined, he is allowing his personal preferences for what he wants the number to be to influence his research scheme.

I am rejecting that approach. I am not at all saying that we know every factor that affects SWRs today. My assessment is that we are in the third or fourth inning of this thing. We have a long way to go before we can reasonably claim that the work of this board community is finished.

"that seems to overstates the case a bit. If you can't guarantee that you've found all the bad things that could happen, it also stands to reason that you might have missed some good things that might happen, no?"

It's possible that we have missed some good things. We are under an obligation as data-based SWR analysts to make appropriate adjustments when we learn of mistakes we have made at earlier stages of our understanding of the SWR concept. By no stretch do I view the current tool as the last word in SWR analysis.

I did not start out with a desire to find a SWR of 2 percent for stocks when they were at high valuations. I had most of my money in stocks at the time I started looking at the SWR question. I am satisfied if the number is 2 or 3 or 4 or 5 or 6. It just makes no difference to me. I just want to know the realities so that I can devise my investing strategies accordingly.

What I want to know is what the number is when you look at all of the historical data known to have an effect on the question being examined. It is not an overstatement for me to say that the results of the conventional methodology studies are inaccurate. It's not possible to get the right answer when excluding a key factor.

There is no possible reasonable justification for deliberately excluding a factor known to have a large effect on the question being examined. Methodologies that do so are analytically invalid for purposes of answering that question.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

"you cannot justify...."

You cannot justify the practice of some community members to continue telling visitors to the various boards that the historical data reveals a 4 percent take-out number from a high-stock-percentage portfolio to be safe at current valuation levels. The historical data simply does not support this claim.

There is not one study that has been prepared since we reached the price levels reached in the late '90s that shows a SWR of anything close to 4 percent. That includes the REHP study. That study was prepared in the mid-90s.

Would intercst find a SWR of 4 percent if he redid his study today? I can't say for sure. Personally, I doubt it.

If he has confidence in the findings he put forward in the mid-90s, he should respond in a reasoned manner to the questions that have been put forward re his methodology. Why not include a new section in his study discussing what I have brought foward, what JWR1945 has brought forward, what Bernstein has brought forward, what Arnott and Smithers and Shiller and others have brought forward?

If he sincerely believes that the 4 percent number can be justified, he should justify it in his study. He should address the questions in a serious way and tell readers his take on why he still believes that his methodology is analytically valid for purposes of determining SWRs.

If he did that, it would go a long way towards addressing the problem. It's unlikely that I would agree with what he said. It would take a lot to persuade me at this point. But there are others who do not feel so strongly about this matter as I do. Perhaps he would persuade them.

There is no possible justification for not addressing the questions in a reasoned manner. If he addresses the questions in the study, readers can make up their own minds as to which side to come down on. As things stand today, many of the readers of his study are not even aware of the claims re the shortcomings of the study that have been put forward.

This is unconsionable. People's retirements are at stake. If there is a reasoned justification for continued use of the conventional methodology, it should be added to the text of the study. If there is not, the terminology of the study should be changed. The 4 percent number should be referred to as the Historical Surviving Withdrawal Rate (that is what the methodology is designed to determine), not the safe withdrawal rate.
bpp
** Regular
Posts: 98
Joined: Tue Nov 26, 2002 6:46 am
Location: Japan

Post by bpp »

bpp wrote:As for "normalizing to a very high level of safety," that seems to be a way of saying that you are constantly seeking out new failure modes?
No. We are taking known factors including known failure modes into account. As we discover new failure modes, we will take them into account as well.
If you can't guarantee that you've found all the bad things that could happen, it also stands to reason that you might have missed some good things that might happen, no?
The conventional methodology claimed to provide safety under worst case conditions. It did not.

Have fun.

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

Post by JWR1945 »

We have shown that current valuations and dividend yields are well outside of the historical range used in those studies. The older 4% number is not safe.
and
That fact does not logically imply your conclusion in the second sentence.
It was not meant to. More precisely, numerous investigations on these boards show that 4% is unsafe. The remarks about dividends and valuations help us to understand why the earlier studies failed.

There are other ways to look at it as well. Here is a simple example. If P/E10 is higher than 25 (and remains higher), the earnings yield is 4% and the dividend yield cannot be any greater than 4%. The dividend growth rate cannot be positive. In fact, it is likely to be negative. This should give us pause about withdrawing 4% when valuations are above those from before the bubble.

There are offsetting factors that might allow you to withdraw 4% successfully. Multiples could expand even more. A super bubble may follow the bubble.

[More plausible would be arguments related to the return of capital since the final balance typically investigated is allowed to fall to zero.]

In terms of a rate to withdraw, 4% is not guaranteed to be safe.

Have fun.

John R.
bpp
** Regular
Posts: 98
Joined: Tue Nov 26, 2002 6:46 am
Location: Japan

Post by bpp »

In terms of a rate to withdraw, 4% is not guaranteed to be safe.
Neither is any number that you are going to come up with.

As far as I can see, the whole SWR methodology is fatally flawed. If your portfolio has any volatility or risk whatsoever, you can never guarantee a perfectly safe withdrawal rate if you insist on taking fixed or inflation-adjusted withdrawals at some rate that ignores how your portfolio is performing. The only way you can be completely safe is if you can find some risk-free, inflation-adjusted asset that returns more after inflation than you need to live on, like TIPS (if you can afford to buy enough of them). And even then, you will always face the risk that you won't be able to reinvest at the required minimum yield when your current holdings mature. (This seems to be an issue that Hocus has already encountered.)

If you want to be sure never to run out of money, you have to be able to limit yourself to some fraction of current portfolio value every year (or to yearly dividend earnings, or something like that). If you have some fixed, irreduceable level of expenses that must be maintained at all costs, then there is no such thing as a guaranteed safe withdrawal rate.

Do you disagree with this?

While you may find some tweak or set of tweaks that would have let you squeeze a little more, historically, out of the basic SWR methodology, in the end you are not going to be able to solve the fundamental problem with the whole approach.

[Snipped. -- Bpp]

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

Post by JWR1945 »

In terms of a rate to withdraw, 4% is not guaranteed to be safe.
Neither is any number that you are going to come up with.

As far as I can see, the whole SWR methodology is fatally flawed. If your portfolio has any volatility or risk whatsoever, you can never guarantee a perfectly safe withdrawal rate..
We have no real disagreement.

The 4% number refers to a very specific problem. A person withdraws a certain percentage of his portfolio's initial balance in the first year. He continues to withdraw the same amount plus inflation (i.e., adjusted to match inflation) for a total of 30 years. What is the greatest percentage that he can withdraw that will last (i.e., have a positive balance) for the entire 30-year period?

When set up properly in terms of probability and statistics, the only percentage that is truly 100% safe can only be zero when equity is included. We need to talk in terms of confidence levels and confidence limits. I generally speak in terms of 90% confidence levels because I think that claiming higher levels is only an illusion when equity is involved. [There are additional reasons having to do with the quality of the statistical estimates.]

When I stated that 4% is not guaranteed, I am talking in terms of a 90% confidence level.

We include all withdrawal strategies and criteria within the scope of our research. For example, we do not always demand that the final balance equals zero (as most other studies have done). By necessity, when we make a study, we examine specific problems with specific strategies and specific criteria.

Have fun.

John R.
bpp
** Regular
Posts: 98
Joined: Tue Nov 26, 2002 6:46 am
Location: Japan

Post by bpp »

The 4% number refers to a very specific problem.
With regards to this very specific problem, I agree completely with what salaryguru said.
When I stated that 4% is not guaranteed, I am talking in terms of a 90% confidence level.
I frankly don't see how you can assign meaningful confidence levels here. However, my opinion probably doesn't matter, because I won't be posting to this forum anymore unless salaryguru's post is restored to this thread.

Bpp
th
** Regular
Posts: 84
Joined: Thu Jun 10, 2004 7:35 am
Location: Northern CA

Post by th »

bpp wrote:
When I stated that 4% is not guaranteed, I am talking in terms of a 90% confidence level.
I frankly don't see how you can assign meaningful confidence levels here. However, my opinion probably doesn't matter, because I won't be posting to this forum anymore unless salaryguru's post is restored to this thread.

Bpp
You cant; perhaps in very long time periods, you might achieve a 90% rate. The "gordon equation" gives one a good bet, but thats only practical in 30+ year time periods.

I'm not sure restoring SG's post is even relevant. I think everything meaningful that can be said has been said already. At least twice.
He who fights with monsters might take care lest he thereby become a monster. And if you gaze for long into an abyss, the abyss gazes also into you. - Nietzsche
Post Reply