SWR: Rule of Thumb or Something Else

Financial Independence/Retire Early -- Learn How!

Do you think:

Poll ended at Fri Aug 13, 2004 4:30 am

There is a definitive right answer?
3
20%
Rule of Thumb?
12
80%
 
Total votes: 15

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

SWR: Rule of Thumb or Something Else

Post by ataloss »

Hocus:
With SWR analysis, there is a right and wrong answer..........................................
There is no right or wrong answer when someone is just picking a withdrawal according to a nice rule of thumb. You could pick three percent and that might be "adequate"for you, someone else could pick four percent and that might be "adequate" for him, and yet someone else might pick five percent and that might be "adequate" for her. What of it?


I guess I am not sure that there is no 'wrong" answer for a rule of thumb (e.g. 10% appears to be wrong for a 30 year retirement) but I think the idea of a rule of thumb suggests some flexibility as Hocus has indicated.
Have fun.

Ataloss
therealchips
*** Veteran
Posts: 174
Joined: Sat Jan 04, 2003 4:00 am
Location: Henderson, Nevada, USA

Post by therealchips »

I come down firmly in the middle, and respond "Both". Looking at sites such as http://www.capn-bill.com/fire and http://www.i-orp.com I think it highly likely that, in those site, the calculations are correct and the quoted data are correct so the results are mathematically correct and not merely a rule of thumb. All this calculation becomes only a rule of thumb, though, in making a decision to act on these results now, for the key reason that we have no way of knowing today how closely the future will resemble the past.

(Was that too obvious to be worth saying? Sorry.)

By now I have given a couple reasons to depart from a plan of equal inflation-adjusted withdrawals, equal in purchasing power that is, not equal as a percentage of the retirement capital (or "stash" as I usually call it). One reason is that it does not suit my psychology to barrel on through with a preplanned standard of living, regardless of what my investments do. I find it easy and natural to let that standard float up or down a bit with the market. Maybe if someone's stash is barely adequate to support a minimal standard of living in retirement, then such flexibility would not work for him. A second reason has to do with life expectancy, my own that is, not the portfolios. I might actually plan on a slightly declining standard of living over the course of a retirement lasting decades on the grounds that I cannot tell ahead of time how long I will live. The idea is that if, in my nineties, I find I have less to spend than I did in my sixties, my attitude could be that this is fine and represents the natural outcome of a plan that acknowledges human mortality and its unpredictability in any one person's case.

As I posted before, a plan to let the standard of living decline over the course of a retirement fell out mathematically when I used a logarithmic utility function for the amount withdrawn and asked the optimizer to pick thirty annual withdrawals that maximize the total expected utility of those withdrawals, using a standard mortality table. I haven't had any reaction to these ideas, either here or at some other place, so maybe I'll let the subject rest for another long while. Hocus tells us, and I don't quarrel with the idea, that planning for withdrawals is incomplete if current valuations are ignored. I suggest that planning may be incomplete also if the plan ignores the mortality of the planner. I don't in fact follow this approach, though, because it calls for such large withdrawals in the early years of retirement. Those are the years that I am most likely to live through.
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 »

I was thinking of it as a sophisticated rule of thumb. I found this on the web.
The OED2 defines "rule of thumb" thusly: "A method or procedure derived entirely from practice or experience, without any basis in scientific knowledge; a roughly practical method. Also, a particular stated rule that is based on practice or experience."
Have fun.

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

Post by JWR1945 »

I just cast the first vote to assert that there is a right answer.

It is just that there are a lot of questions and each one has its own right answer. There are lots of Safe Withdrawal Rates. Each one is unique. Each one applies to a slightly different circumstance.

That puts me pretty much into agreement with therealchips in this sense: I would prefer to have said both.

I think that the phrase Safe Withdrawal Rate should be expressed only in terms of its own context. The reason is simple. If I am going to make a calculation, I need to limit the scope of the problem so that there is such a thing as a right answer. Later, I can look at a lot of similar calculations along with the context behind each. From that, I can decide what to do.

Have fun.

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

Post by hocus »

The OED2 defines "rule of thumb" thusly: "A method or procedure derived entirely from practice or experience, without any basis in scientific knowledge

This hits the nail on the head, so far as I am concerned. I am saying that the calculation of an SWR should to a large extent follow scientific procedures. There are rules that must be followed for a scientific analysis to be valid.

I am not saying that the SWR should predict the future accurately. This is not the purpose. It tells you what annual withdrawal you can begin making on the date of your retirement that the data in existence on that date says will provide you with a strong assurance of a successful retirement (assuming the future is like the past).

If I am going to make a calculation, I need to limit the scope of the problem so that there is such a thing as a right answer. Later, I can look at a lot of similar calculations along with the context behind each. From that, I can decide what to do.

JWR1945 comes the closest to expressing my viewpoint, but we are not in precise agreement. I agree that stating the context in which a calculation is performed goes a long ways toward solving the problem. If Intercst had said, "4 percent is the SWR for 2000 in the event that for the first time in the history of the stock market changes in valuation levels end up having no effect on returns," that warning would have let people know that they could not have much confidence in the number, and that would have been a good thing.

I go one step further, however. I say that some elements of the calculation are so essential that the exercise being performed is not a valid SWR calculation if they are not included. I put the valuation item in this category. I have no objection to people doing the calculation without factoring in the effects of changes in valuation levels, but I think they should come up with some other name for what they are doing than SWR analysis.

I am not dogmatic about how changes in valuation levels should be factored in, however. Bernstein did it one way and he came up with an SWR of 2 perdent for the year 2000. JWR1945 did it another way, and he came up with 2.3 percent. Both of those numbers seem reasonable to me. If these two analyses were formal studies, the proper thing to do would be to describe how the numbers were calculated and why that procedure was elected over other possibilities. That would provide the "context" that JWR1945 is seeking.

Our difference is that I believe that some effort to incorporate the effects of changes in valuation levels is needed now that we know with "mathematical certainty" that changes in valuation levels affect the result. I am saying that it is an essential step to performing a valid analysis to include this factor in some way, but that there is room for reasonable differences of opinion as to exactly how the factor should be incorporated into the analysis.

In the year 2000, the number "2" is a reasonable SWR assessment, and the number "2.3" is reasonable too. The number "1.7" might be reasonable as well. But not the number "4." Permit that much variance in results and you are taking a chance of reducing the SWR concept to insignificance. To define the concept in such a way as to provide that both 2 percent and 4 percent are "correct" answers is to provide for too much flexibility for a concept that I would like to see retain at least some semblance of objective science to it.

If both of those numbers are correct, then SWR analysis really does provide nothing more than a rule of thumb. If that is all it provides, I question whether it serves much purpose. There are lots of ways to come up with rules of thumb. Rules of thumb are partly informed guesses, and I do not like the idea of making the survival of my Retire Early plan dependent on my ability to make partly informed guesses. I prefer to make as fully informed a guess as it is possible to make, given the data available to me. A fully informed "guess" as to the withdrawal rate that will make your plan work absolutely requires consideration of the valuation question, in my view.
User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

Post by ataloss »

This hits the nail on the head, so far as I am concerned. I am saying that the calculation of an SWR should to a large extent follow scientific procedures. There are rules that must be followed for a scientific analysis to be valid.

I guess I see Monte Carlo analysis as being more like baking than science. In baking you add the ingredients run a mixer and get a result (ok you have to heat in a oven.) I see the MC as a tool. What you get still depends critically on what you put in. Is this wrong?
Have fun.

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

Post by JWR1945 »

ataloss
What you get still depends critically on what you put in. Is this wrong?


No. This is not wrong. It just isn't confined to the Monte Carlo approach.

Lurking in the background is an assumption that a historical sequence approach is inherently better than a Monte Carlo approach. It is inherently better only in the sense that it is easier to discern what the assumptions are. Even then, a Monte Carlo approach requires you to look into your assumptions. In that sense and from a model creator's standpoint, a Monte Carlo model is better.

In general, there are times when a historical sequence approach is better. There are other times when a Monte Carlo approach is better.

Have fun.

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

Post by bpp »

Chips,
As I posted before, a plan to let the standard of living decline over the course of a retirement fell out mathematically when I used a logarithmic utility function for the amount withdrawn and asked the optimizer to pick thirty annual withdrawals that maximize the total expected utility of those withdrawals, using a standard mortality table. I haven't had any reaction to these ideas, either here or at some other place, so maybe I'll let the subject rest for another long while.
I think your approach sounds quite interesting. Would you mind posting some details on the mathematics?

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

Post by wanderer »

SWR = tar baby :shock:
regards,

wanderer

The field has eyes / the wood has ears / I will see / be silent and hear
wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer »

maximize the total expected utility of those withdrawals, using a standard mortality table. I haven't had any reaction to these ideas, either here or at some other place, so maybe I'll let the subject rest for another long while.

sorr trc, i didn't catch this. a) how do you define "utility", other than as a place that generates power or tremendous financial losses, and b) did you see a post from dagrims(?) re: spending more earlier than later.

I will regrettably die with a certain amt of cash because I have no ability to predict my demise with great accuracy. i.e. i thought i would die yesterday, at the graduation. the excruciating torture delivered by the entire ceremony, together with the hacking SARS carrier on my left, convinced me of the high probability of this "negative" outcome (at least, for our hero). in fact i secretly (okay, openly and loudly), prayed for sweet release. still, here i am today...
regards,

wanderer

The field has eyes / the wood has ears / I will see / be silent and hear
User avatar
BenSolar
*** Veteran
Posts: 242
Joined: Mon Nov 25, 2002 5:46 am
Location: Western NC

Post by BenSolar »

therealchips wrote:As I posted before, a plan to let the standard of living decline over the course of a retirement fell out mathematically when I used a logarithmic utility function for the amount withdrawn and asked the optimizer to pick thirty annual withdrawals that maximize the total expected utility of those withdrawals, using a standard mortality table. I haven't had any reaction to these ideas, either here or at some other place, so maybe I'll let the subject rest for another long while.
I think this sounds fascinating :D

I've been trying to catch up on the posts, and it's hard to keep track of which ones to respond to there are so many interesting ones. :shock:

I am with bpp. Could you start a new thread with some details on your approach? Please? :D

Thanks, Chips,

Ben
"Do not spoil what you have by desiring what you have not; remember that what you now have was once among the things only hoped for." - Epicurus
[KenM]
*** Veteran
Posts: 133
Joined: Tue Mar 04, 2003 12:54 am

Post by [KenM] »

I voted "there is a definitive right answer" - which doesn't mean blindly adopting a precisely calculated mathematical result but it's not just following a rule of thumb based on experience either.
A bit like ataloss's friend Bob - there are now lots of lovely computer models to carry out precise calculations to design a bridge - and Bob might say, OK lets run the model; apply a factor of safety of, say, 1.5 and that gives a beam 57.35mm deep and concrete's cheap so we'll round that off to 600mm. And then he might say, I've been designing bridges for 30 years and from my experience, as a rule of thumb, 700mm is perhaps about right but that was using my trusty old slide rule and the young whizzkids tell me that computer models are very good. But maybe at the next millenium a couple of thousand people might stand on the bridge to watch the fireworks. So to be safe I'll round it up a bit more to 650mm and that's the definitive right answer for this particular bridge.

BTW, SWR = tar baby
wanderer,
a very cryptic statement - looks American to me - I don't really understand the "Chicken Little" thread either.
Last edited by [KenM] on Thu May 29, 2003 8:52 pm, edited 2 times in total.
KenM
Never try to teach a pig to sing. It wastes your time and annoys the pig.
User avatar
BenSolar
*** Veteran
Posts: 242
Joined: Mon Nov 25, 2002 5:46 am
Location: Western NC

Post by BenSolar »

KenM wrote:that gives a beam 54.5mm deep and concrete's cheap so we'll round that up to 600mm. And then he might say, I've been designing bridges for 30 years and from my experience, as a rule of thumb, 700mm is perhaps about right but that was using my trusty old slide rule and there appear to be no unusual factors likely to affect this brigde and computer models now allow a more comprehensive analysis so I'll go with 600mm as the definitive right answer for this particular bridge.
:lol: LOL :lol:
SWR = tar baby
...
a very cryptic statement
'tar baby' is a reference to the 'Uncle Remus' stories. http://xroads.virginia.edu/~UG97/remus/tar-baby.html

The tar baby was a trap laid by Brer Fox to catch Brer Rabbit ...

Or am I missing your humor, KenM? I'm bad for that sometimes. Please don't throw me in the briar patch Brer Ken.

BenStuckInTheTarBaby
"Do not spoil what you have by desiring what you have not; remember that what you now have was once among the things only hoped for." - Epicurus
[KenM]
*** Veteran
Posts: 133
Joined: Tue Mar 04, 2003 12:54 am

Post by [KenM] »

Or am I missing your humor, KenM
Ben,
It's interesting to see that often humour (note the spelling) can depend on assuming that everybody grew up with the same cultural background. I think that the Uncle Remus and Chicken Little stories are essentially American - at least I didn't come across them as a kid. So I didn't really understand the SWR=tar baby reference but I guessed it's American humor. So you can relax, I'm not trying to catch you out with my own obscure Brit humour. :D
KenM
Never try to teach a pig to sing. It wastes your time and annoys the pig.
wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer »

KenM:
I don't really understand the "Chicken Little" thread either
Having spent several years during the 1970s in Germany without access to peanut butter (a distinctly American treat), I understand your frustration with culture assumed by the majority of board participants.

Chicken Little was a story about a chicken who had an acorn fall on her head and ran around crying "the sky is falling" as a prediction of the end of the world. She then proceeded to act as if the world truly were ending. Moral=make sure the sky really is falling as opposed to an acorn falling on your head before you make permanent decisions assuming that's the case.

Fellow Americans, did I summarize correctly? Been a long time since I heard the Chicken Little story.
regards,

wanderer

The field has eyes / the wood has ears / I will see / be silent and hear
wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer »

a tar baby is another word for a "sticky" problem. You think you'll be done with it quickly, but each time you touch it, you get further encumbered and can't shake it off. Think "quagmire". Think "iraq is a 'tar baby'". IIRC, it is also a particularly ugly slang term, as well.

The SWR discussion is too "thick" for me, right now. Too many items not tied down. No one complete, manageable list in one place. Terms used loosely with no resolution.

I appreciate folks efforts - nice charts, gummy .

I would appreciate it if tmf references were left out - I don't see it as useful to revisit that place and I don't see how it helps learning take place.

my $.02.
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 »

A tar baby is another word for a "sticky" problem. You think you'll be done with it quickly, but each time you touch it, you get further encumbered and can't shake it off. Think "quagmire".

If the SWR Debate is a quagmire, it is the most wonderful sort of quagmire there ever could be. There are two reasons why this discussion never dies out: (1) It is of tremendous consequence; and (2) no one yet has a complete understanding of all the angles.

Each time we think we have something settled, someone puts up a post noting an angle we had not considered before and we have to go back to the drawing board. Good! That's what you want on a discussion board. Lots of returns to the drawing board means lots of illuminating posts.

The longer this goes on, the better it gets. The goal should not be to force the discussion to an unnatural close. The goal should be to expand on every tentative insight we have already uncovered. There have been scores of them! One day I am going to put up a post summarizing the powerful insights we have developed in just one year's time. It is going to be a long post. We have made incredible progress, despite the occasional bumpiness of the road we have been traveling.

I guess the tone of frustration you sometimes hear comes from the growing suspicion that this is not going to end anytime soon. I believe that those suspicions are well-placed. I plan later this year to open a new board called "SWR Research Group." Why? Because there are too many sub-topics that need to be explored for the topic to be done justice at a general FIRE board. Our explorations are going to just keep expanding and expanding, heading off in dozens of different directions at once. Again, Good! That's what we want.

SWR analysis is key to planning for FIRE. Most people have not yet grasped the amazing power of this tool to help in investment planning. There is no reason to limit SWR analysis to deciding on your take-out number. It helps you do that, but the more important use in my view is in assessing the merit of various allocation alternatives. SWR analysis, done properly, allows you to customize the investment decision-making process so that your investments are being employed in the most efficient manner possible to achieve what you in particular want them to achieve.

We have only scratched the surface of this topic. There is a lot more to come. One reason why I want a separate SWR Research Group board is that we need a repository of the best posts. We need to start organizing the insights we have developed and laying out plans for channeling our efforts in the most constructive directions.

People don't feel drawn to discuss things they already understand. It's not worth the effort because there is not much to learn by repeating generally appreciated insights. No one truly understands SWRs yet. So this is a rich vein of profitable discussion we have discovered,. People also don't feel drawn to discuss things of little consequence. And most of us possess at least a foggy sense that developing a clearer understanding of the SWR concept can lead us to some exxciting and enriching places.

If you take a broad perspective, most of what has happened should not be so surprising. Change is hard, and what we are doing here is changing how we approach the question of how to make effective investment decisions.

It is rare in any field of learning for all of the answers to be discovered the first time someone ventures forth with a rough draft effort to analyze a question. The pioneers of SWR research were our friends. They got the ball rolling, and we all owe them a debt of gratitude. But we shouldn't fall into the trap of thinking that, because these pioneers were good and smart people, they they were able to anticipate every issue that would ever arise. That's simply not a fair or reasonable thing to expect.

The pioneers did a good job. But their explorations are not even close to being finished products. We have lots more to do and it's all very exciting. We need to let go of the expectation that this is going to be resolved in a week or a month or a year. We are only getting started.

The last sentence of this post may come across braggish. I don't intend it that way. I have made some contributions, but there are a good number of others who have made equally valuable contributions and an even larger number who have made at least one or two highly significant contributions. This has been a group effort from the start. I put the comment forward despite the risk of it being interpreted wrongly because I think it is true and I think it puts what has been going on for the past 12 months in a more optimistic and more accurate perspective than the one that has been the more frequently voiced one.

Here goes.

I believe that the May 13, 2002, post will go down in the history of the FIRE movement as being the End of the SWR World As We Know It--and I'm Glad.
User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

Post by ataloss »

ataloss
What you get still depends critically on what you put in. Is this wrong?
jwr1945
No. This is not wrong. It just isn't confined to the Monte Carlo approach.

Lurking in the background is an assumption that a historical sequence approach is inherently better than a Monte Carlo approach. It is inherently better only in the sense that it is easier to discern what the assumptions are.
I think the historical approach is a perfect guide to the past but it remains to be seen if itis a good guide to the future. MC has the benefit of allowing straightforward generation of what ifs. (although this can be done using historical sequences as well)

Hocus
5/13/02 may be an important day for you but my friend Bob says that it was always rather obvious that valuation affected returns and swr. Historically there is a distribution of maximal withdrawal rates ranging from 4% to greater than 10%. For the sake of conservaqtism you would chose the lowest maximal withdrawal rate for the study period as the swr ie 100% safe for all time periods studied. Perhaps 5/13/02 was important in that you questioned your teacher for the first time?
Have fun.

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

Post by hocus »

5/13/02 may be an important day for you but my friend Bob says that it was always rather obvious that valuation affected returns and swr.

If it has always been so obvious, why do the existing SWR studies not take the effect of changes in valuation levels into account in the results they produce?

Intercst was saying on the REHP board in the year 2000 that the SWR for a retirement beginning that year was 4 percent. Today he is saying that the SWR for a retirement beginning this year is 4 percent.

Does that make a lick of sense to you? If there any possibility that the SWR is the same for both years?

There is not. A methodology that produces the same SWR for both years is fundamentally flawed. There is something missing in such a methodology. The results just do not make sense.
User avatar
ataloss
**** Heavy Hitter
Posts: 559
Joined: Mon Nov 25, 2002 3:00 am

Post by ataloss »

Intercst was saying on the REHP board in the year 2000 that the SWR for a retirement beginning that year was 4 percent. Today he is saying that the SWR for a retirement beginning this year is 4 percent.
In post 7 of his board he was saying 4% was a "rule of thumb" and on 5/13/03 he was clearly saying that :
That's correct. "100% safe" in terms of the safe withdrawal studies means that the selected withdrawal rate survived all the pay out periods examined from 1871 to 2000. If you are predicting that something worse ..... then the safe withdrawal rates are less than 100% safe. That's all it is saying.
Now the maximum amount that could be withdrawn will vary each year. Sometimes it will be 10% (for 30 years) and sometimes it will 4%. $% is not the average it is the minimum. It will not change unless the future is worse than the past

intercst:
If I was smart enough to be able to tell when stock prices were high or low, I would adjust my initial withdrawal rate accordingly. For example, the 100% safe withdrawal rate for a 77% stock/23% fixed income portfolio and a 20-year pay out period is 4.56% (The years 1929-1949). If I was bright enough to fortell the 20-year bull market in 1980, I could have raised my 20-year safe withdrawal rate to 16.00%.
Have fun.

Ataloss
Post Reply