SWR of 6.21% for 26 years
- Thread starter amt
- Start date
unclemick
Give me a museum and I'll fill it. (Picasso) Give me a forum ...
Cut expenses. Live off the SEC yield of your retirement portfolio(total taxable plus tax deferred). If history mean reverts and the yield goes north of 4% SWR - take the lower number. If the size (too small) of your portfolio prevents this - forget retirement calculators and look/read about value funds/individual stocks/MLPs/bonds/etc and construct the best portfolio you can.
Stocks above their 1871 trend channel and historic low interest have been a source of caution expressed by many posters to this forum.
Anyway - my two cents.
Stocks above their 1871 trend channel and historic low interest have been a source of caution expressed by many posters to this forum.
Anyway - my two cents.
Could you give me a capsule summary of what you think this message is, and what to do about it Rob?
I think that people with high stock allocations need to give some thought to the idea of lowering them until the PE10 number and dividend payout number return to levels closer to the historical norms.
Here's a post from the SWR Research Group board where JWR1945 puts forward an allocation strategy that the historical data indicates would support a 4 percent withdrawal at today's valuation levels.
http://nofeeboards.com/boards/viewtopic.php?t=2158
By no means do I think that the approach described in the link above is the only reasonable approach. Nor do I believe that everyone needs to be getting out of stocks altogether. There is lots of room for reasonable differences of opinion on lots of questions.
What alarms me are the claims that those with a 74 percent stock allocation can count on a 4 percent withdrawal for retirements beginning at the valuation levels that have applied since the late 1990s. I don't think those claims are reasonable. I think we all need to back off of that stuff a good bit before our promotion of that stuff ends up discrediting the entire Retire Early movement.
SWR analysis is not yet sufficiently developed to use it to support dogmatic pronouncements. We need to engage in years of study of what the historical data really says before we get to a point where we are telling people with confidence that particular take-out percentages are "100 percent safe."
I am more comfortable telling people what is not safe than I am telling them what is safe. If there is strong evidence in the data that a particular strategy may not work, that is enough to declare it something less than absolutely safe. If there is strong evidence that a particular strategy will work, the most you can reasonably do is say that it appears to be safe pending futher analysis.
I think that we have lost sight of the purpose of SWR analysis. The original purpose was to caution people as to possible risks. When Scott Burns used the Trinity study to show that Peter Lynch was wrong with his claims that a 7 percent withdrawal was safe, he was providing his readers an extremely valuable insight. SWR analysis is being used today to support investment strategies that may or may not work, but which in fairness can only be advertised as strategies entailing a good bit of risk. We all should do what we can to bring that sort of thing to a stop.
I think that people with high stock allocations need to give some thought to the idea of lowering them until the PE10 number and dividend payout number return to levels closer to the historical norms.
Here's a post from the SWR Research Group board where JWR1945 puts forward an allocation strategy that the historical data indicates would support a 4 percent withdrawal at today's valuation levels.
http://nofeeboards.com/boards/viewtopic.php?t=2158
By no means do I think that the approach described in the link above is the only reasonable approach. Nor do I believe that everyone needs to be getting out of stocks altogether. There is lots of room for reasonable differences of opinion on lots of questions.
What alarms me are the claims that those with a 74 percent stock allocation can count on a 4 percent withdrawal for retirements beginning at the valuation levels that have applied since the late 1990s. I don't think those claims are reasonable. I think we all need to back off of that stuff a good bit before our promotion of that stuff ends up discrediting the entire Retire Early movement.
SWR analysis is not yet sufficiently developed to use it to support dogmatic pronouncements. We need to engage in years of study of what the historical data really says before we get to a point where we are telling people with confidence that particular take-out percentages are "100 percent safe."
I am more comfortable telling people what is not safe than I am telling them what is safe. If there is strong evidence in the data that a particular strategy may not work, that is enough to declare it something less than absolutely safe. If there is strong evidence that a particular strategy will work, the most you can reasonably do is say that it appears to be safe pending futher analysis.
I think that we have lost sight of the purpose of SWR analysis. The original purpose was to caution people as to possible risks. When Scott Burns used the Trinity study to show that Peter Lynch was wrong with his claims that a 7 percent withdrawal was safe, he was providing his readers an extremely valuable insight. SWR analysis is being used today to support investment strategies that may or may not work, but which in fairness can only be advertised as strategies entailing a good bit of risk. We all should do what we can to bring that sort of thing to a stop.
J
John Galt
Guest
Hi *****! Good post. Since I don't really use SWR
planning much in my own situation, I guess I won't
get nervous until my net worth starts to decline.
In the meantime, I subscribe to Al Pacino's opinion regarding risk in general.............
"You can get killed walking your doggie!" You can also get
"killed" financially in spite of "walking" through all of the SWR formulas ever devised.
John Galt
planning much in my own situation, I guess I won't
get nervous until my net worth starts to decline.
In the meantime, I subscribe to Al Pacino's opinion regarding risk in general.............
"You can get killed walking your doggie!" You can also get
"killed" financially in spite of "walking" through all of the SWR formulas ever devised.
John Galt
BigMoneyJim
Thinks s/he gets paid by the post
By the way, personally I agree with unclemick and John Galt on the ultimate futility of predicting the future.
However, we frequently use the past to guide us for the future. We can measure past returns and apply historical statistical results of our future plans (FIREcalc). We can extrapolate based on market direction and performance (salaryguru), and we can extrapolate based on price versus value (*****).
I think any of the above given tools is valid on its own to give us different points of view to help plan, but if we start mixing the tools I want to be clear we are doing so. I don't want to say that I'm only relying on 105 samples of historical statistics when I'm applying a market or value bias on the results. That's when you end up mathematically "proving" Stevie Wonder is God.
***** has made a good point by comparing this visit's drunk man to two previous drunk visits by him (effectively eliminating all but 2 of the 100+ historical samples when the visitor isn't drunk and realising it's not enough to be sure), but he has added value bias even to his example's forward thinking by noting the man is twice as drunk now than before. If we were going to argue this point (I'm not), we should argue whether or not the man is really twice as drunk this time or even whether he's drunk at all; or if we use historical statistical analysis we should argue how many historical samples are valid and whether that's a sufficient sample pool. But to argue both at the same time is lots of trouble.
I think amt's example and salaryguru's support of it adds market movement analysis to the FIREcalc result and tries to claim FIREcalc's statistical method fully supports the judgement-modified conclusion.
I hope I'm not stepping on toes here; I'm just trying to understand what we're all saying and figure out why amt's 6.22% claim just doesn't sound right to me. I may be making a total fool of myself and not realizing it yet.
By the way, FIREcalc as a tool does not tell you what SWR is. It simply tells you what it was historically. After the result is given then you apply value judgements on whether that's useful going foreward, but let's not use FIREcalc's math to support the validity of our judgement.
However, we frequently use the past to guide us for the future. We can measure past returns and apply historical statistical results of our future plans (FIREcalc). We can extrapolate based on market direction and performance (salaryguru), and we can extrapolate based on price versus value (*****).
I think any of the above given tools is valid on its own to give us different points of view to help plan, but if we start mixing the tools I want to be clear we are doing so. I don't want to say that I'm only relying on 105 samples of historical statistics when I'm applying a market or value bias on the results. That's when you end up mathematically "proving" Stevie Wonder is God.
***** has made a good point by comparing this visit's drunk man to two previous drunk visits by him (effectively eliminating all but 2 of the 100+ historical samples when the visitor isn't drunk and realising it's not enough to be sure), but he has added value bias even to his example's forward thinking by noting the man is twice as drunk now than before. If we were going to argue this point (I'm not), we should argue whether or not the man is really twice as drunk this time or even whether he's drunk at all; or if we use historical statistical analysis we should argue how many historical samples are valid and whether that's a sufficient sample pool. But to argue both at the same time is lots of trouble.
I think amt's example and salaryguru's support of it adds market movement analysis to the FIREcalc result and tries to claim FIREcalc's statistical method fully supports the judgement-modified conclusion.
I hope I'm not stepping on toes here; I'm just trying to understand what we're all saying and figure out why amt's 6.22% claim just doesn't sound right to me. I may be making a total fool of myself and not realizing it yet.
By the way, FIREcalc as a tool does not tell you what SWR is. It simply tells you what it was historically. After the result is given then you apply value judgements on whether that's useful going foreward, but let's not use FIREcalc's math to support the validity of our judgement.
Well, in the past couple of days I did not check these boards, a lot had been discussed. It seems to me that most of the criticism against my initial proposal stems from the sentiment that the future is likely to be worse than the worst case of the past. I understand that sentiment and it is probably wise to asssume as such.
Salary, thanks for your well reasoned explanations in support of my hypothesis. I can't explain anywhere near what you have.
Personally, I am conservative. I plan on drawing about 3% when I ER. That's where people have to make their own plan and adjust the FIRECalc results up or down based on their comfort levels.
Regards,
amt
Salary, thanks for your well reasoned explanations in support of my hypothesis. I can't explain anywhere near what you have.
Personally, I am conservative. I plan on drawing about 3% when I ER. That's where people have to make their own plan and adjust the FIRECalc results up or down based on their comfort levels.
Regards,
amt
cute fuzzy bunny
Give me a museum and I'll fill it. (Picasso) Give me a forum ...
Hoo boy, we're down to semantics.
Well I think this is one of those discussions where we end up agreeing on the results but the path on which we get there is rife with disagreement.
Just to wrap up my thoughts on this:
- Correlation by its definition show a demonstrable connection between one or more things and another. Correlation cannot be proved without demonstrating the connection, ie, establishing that the presence, absence, or change in one thing has a measurable effect on the other. In the absence of a solid measurement, correlation becomes opinion.
- I cant agree with the weather example. Weather correlations and predictions include no social or psychological factors, which are the primary drivers of short term (<5-7 year) market movement.
- I think the principal disconnect here is that I perceive that you may be discounting the same social and psychological influences on the market and perceive that there is a weather-like pattern to markets that can be divined with some level of accuracy. If this were true, then actively managed funds should beat the stuffing out of index funds, except that historic data shows that they fare worse than if they were run by rules of chance.
- The argument that monte carlo simulation results fare worse is a failure of those calculators to follow some correlation seems off the mark to me. Monte carlo simulators produce scarier results because they produce more periods of calculation and can as a result of their nature string together more bad years in series than the shorter historic numbers produce.
- I'm still stuck on the short and long term expectations, and feel like we're dancing around it. I'll try to restate what I think you said again since i missed the first time. You feel that certain hard to measure correlations in market movements makes shorter term (5-7 year) movements more predictable than longer term ones (30+ years). I agree in principle that a lot may happen to disrupt markets over 30 years, but I also note that a lot HAS happened and during long periods of time the market has simply moved upwards. Which agreeably has nothing to do with where ours will end up. The same disruptions however make for far more substantial immediate and short term effects. I look at an 80 year log plot of any major long running index and except for the recent bump and dip I see a fairly smooth and straight line that over a 30 year period is higher at the end than at the beginning. Not so for ANY period of 1-20 years. Thus I can reasonably presume that short of major damage to or extinction of our society, the same is likely true for the next 30 years.
- I can reasonably presume that if stocks have run up a lot in a year or two, that at some point they will either head back down in a shot or a drizzle, or stay where they are for a long time until earnings catch up. My likely ability to predict that through any correlative action is very low.
- Is this tracking as random as a coin toss? Probably not. I think it would be unlikely to see 50 years of straight down or 50 years of straight up, both possibilities of a coin toss.
- I think there is reason to believe that there is SOME correlation between past and current actions in the market and future ones. I think it is unreasonable to believe that it is measurable, predictable, or actionable.
Is this last bit agreeable, makes sense, and sums up the whole thing?
As far as the original point and tying it in with *****'s points, I dont think the 6.21 works. Simply because stocks ran up so much prior to 2000 that even taking the run down into consideration, they never reached historical valuation levels, and this past years run up made it worse. Expecting that we'll get a fair and upward run up from here in line with historic results at historic valuation levels while excluding the plain observation that todays valuations are far higher than those historic ones may not be reasonable.
These long "bad times" calculations also presumes another big one: that someone would watch their portfolio decline steadily, as they did in the 1929 and 1965 scenario returns minus withdrawals, and do so for 20 years without changing anything or getting out. You can be brave running a calculator based on historic data. History shows that the reality of 5 or 10 years of this results in an exceptional lack of braveness.
All this having been said, I'll subscribe to any of John Galts posts in this thread. Do your planning, be careful, be patient, and in the end none of that really matters.
Fortunately the worst case end result is punching a time card. Not so bad.
Well I think this is one of those discussions where we end up agreeing on the results but the path on which we get there is rife with disagreement.
Just to wrap up my thoughts on this:
- Correlation by its definition show a demonstrable connection between one or more things and another. Correlation cannot be proved without demonstrating the connection, ie, establishing that the presence, absence, or change in one thing has a measurable effect on the other. In the absence of a solid measurement, correlation becomes opinion.
- I cant agree with the weather example. Weather correlations and predictions include no social or psychological factors, which are the primary drivers of short term (<5-7 year) market movement.
- I think the principal disconnect here is that I perceive that you may be discounting the same social and psychological influences on the market and perceive that there is a weather-like pattern to markets that can be divined with some level of accuracy. If this were true, then actively managed funds should beat the stuffing out of index funds, except that historic data shows that they fare worse than if they were run by rules of chance.
- The argument that monte carlo simulation results fare worse is a failure of those calculators to follow some correlation seems off the mark to me. Monte carlo simulators produce scarier results because they produce more periods of calculation and can as a result of their nature string together more bad years in series than the shorter historic numbers produce.
- I'm still stuck on the short and long term expectations, and feel like we're dancing around it. I'll try to restate what I think you said again since i missed the first time. You feel that certain hard to measure correlations in market movements makes shorter term (5-7 year) movements more predictable than longer term ones (30+ years). I agree in principle that a lot may happen to disrupt markets over 30 years, but I also note that a lot HAS happened and during long periods of time the market has simply moved upwards. Which agreeably has nothing to do with where ours will end up. The same disruptions however make for far more substantial immediate and short term effects. I look at an 80 year log plot of any major long running index and except for the recent bump and dip I see a fairly smooth and straight line that over a 30 year period is higher at the end than at the beginning. Not so for ANY period of 1-20 years. Thus I can reasonably presume that short of major damage to or extinction of our society, the same is likely true for the next 30 years.
- I can reasonably presume that if stocks have run up a lot in a year or two, that at some point they will either head back down in a shot or a drizzle, or stay where they are for a long time until earnings catch up. My likely ability to predict that through any correlative action is very low.
- Is this tracking as random as a coin toss? Probably not. I think it would be unlikely to see 50 years of straight down or 50 years of straight up, both possibilities of a coin toss.
- I think there is reason to believe that there is SOME correlation between past and current actions in the market and future ones. I think it is unreasonable to believe that it is measurable, predictable, or actionable.
Is this last bit agreeable, makes sense, and sums up the whole thing?
As far as the original point and tying it in with *****'s points, I dont think the 6.21 works. Simply because stocks ran up so much prior to 2000 that even taking the run down into consideration, they never reached historical valuation levels, and this past years run up made it worse. Expecting that we'll get a fair and upward run up from here in line with historic results at historic valuation levels while excluding the plain observation that todays valuations are far higher than those historic ones may not be reasonable.
These long "bad times" calculations also presumes another big one: that someone would watch their portfolio decline steadily, as they did in the 1929 and 1965 scenario returns minus withdrawals, and do so for 20 years without changing anything or getting out. You can be brave running a calculator based on historic data. History shows that the reality of 5 or 10 years of this results in an exceptional lack of braveness.
All this having been said, I'll subscribe to any of John Galts posts in this thread. Do your planning, be careful, be patient, and in the end none of that really matters.
Fortunately the worst case end result is punching a time card. Not so bad.
Th,
I mostly agree with your bold face summary except
for one minor point. Both Bernstiein and Bogle
suggest that it is probably smart to make small
adjustments to your asset allocation based on
current market fundamentals. I cited Bernstein
in an earlier post on this thread. Bogle expresses
the same sentiment on page 244 of his book
"Bogle on Mutual Funds". You yourself indicated
that you felt REITs were overvalued and were planning
to take some off the table. If you really believed
that short term returns are completely unpredictble
why would you do that? BTW, I value your comments
very much. You can type faster than I can think,
even while holding a glass of wine in one hand and
a 25 lb. lap cat in the other.
Cheers,
Charlie (aka Chuck-Lyn)
I mostly agree with your bold face summary except
for one minor point. Both Bernstiein and Bogle
suggest that it is probably smart to make small
adjustments to your asset allocation based on
current market fundamentals. I cited Bernstein
in an earlier post on this thread. Bogle expresses
the same sentiment on page 244 of his book
"Bogle on Mutual Funds". You yourself indicated
that you felt REITs were overvalued and were planning
to take some off the table. If you really believed
that short term returns are completely unpredictble
why would you do that? BTW, I value your comments
very much. You can type faster than I can think,
even while holding a glass of wine in one hand and
a 25 lb. lap cat in the other.
Cheers,
Charlie (aka Chuck-Lyn)
cute fuzzy bunny
Give me a museum and I'll fill it. (Picasso) Give me a forum ...
Charlie - thanks for the kind words. I dont feel like anything adversarial or discounting really happens here. Well not most of the time. Its just good clean expression of opinions based on wan facts and large supposition. And I'm only typing quickly because the big cat is sprawled on the couch defying the dogs to come over and try and smell his butt, and the coffee has gone cold. Wine is unavailable at this time and will remain so until a trip to the store is undertaken.
Why do I say the short term market movements are unpredictable and then with 30 breaths later say that I'm taking some reits off the table.
Because I'm an idiot. And also as per Bogle and Bernsteins allocation suggestions, I'm considering taking from one boat thats floated higher than the rest and donating to one boat that is 20 feet underwater with the idea that the overall rising tide will float all boats.
In this case I'm not predicting. I'm guessing. And probably wrong in the short term. Maybe just as wrong in the long term. Hell, more than half of my investments are still in (admittedly cheap and non-volatile) actively managed funds. So obviously I still dont get it.
Why do I say the short term market movements are unpredictable and then with 30 breaths later say that I'm taking some reits off the table.
Because I'm an idiot. And also as per Bogle and Bernsteins allocation suggestions, I'm considering taking from one boat thats floated higher than the rest and donating to one boat that is 20 feet underwater with the idea that the overall rising tide will float all boats.
In this case I'm not predicting. I'm guessing. And probably wrong in the short term. Maybe just as wrong in the long term. Hell, more than half of my investments are still in (admittedly cheap and non-volatile) actively managed funds. So obviously I still dont get it.
J
John Galt
Guest
Hey TH! Your last post makes me think maybe the wine was not as far away as you implied. It's okay though.
I've written some of my best stuff while under the
influence of "Old Stumpknocker". My advice is that it's
okay to pontificate during cocktail hour, but you're
better advised to wait until morning to hit the send button.
John Galt
I've written some of my best stuff while under the
influence of "Old Stumpknocker". My advice is that it's
okay to pontificate during cocktail hour, but you're
better advised to wait until morning to hit the send button.
John Galt
cute fuzzy bunny
Give me a museum and I'll fill it. (Picasso) Give me a forum ...
Well no trip to the store today, but I just discovered two bottles of beer hiding in the bottom of the fridge. I wouldnt expect anything good to come of two bottles of beer, but you never really know. The beer must have been there for a while because I dont recall buying it. It might get stronger over time.
dory36
Early-Retirement.org Founder, Developer of FIRECal
- Joined
- Jun 23, 2002
- Messages
- 1,841
Right. There is an overhelming tendency to reify the SWR concept.
The 6.21% versus 4.1% SWR paradox is interesting as a puzzler, but I can't imagine anyone locking into the concept as a 30+ year strategy. SWR isn't a strategy - it's a crosscheck for your own estimates of your ability to stretch your portfolio for a period longer than you've probably done anything else.
To me, Firecalc's primary, secondary, and tertiary purpose is to conduct a reality check before burning the bridge as you depart the working world. (I don't know the word for #4, so I stopped...)
Once you have burned that bridge, presumably the brains that allowed you to accumulate ~25 X your annual expenses haven't been left at the office, and you are able to adapt to the world as you encounter it.
wabmester
Thinks s/he gets paid by the post
- Joined
- Dec 6, 2003
- Messages
- 4,459
quaternary
No, I don't think so. You are trying to do something very different than find a SWR with the calculator when you are testing periods where a given safe withdrawal rate would fail.
To find a SWR, the calculator simply searches for the worst case. It finds only one case for the given allocation, expense ratio, nest egg, longevity, etc. Now, that worst case represents an individual's SWR only if you believe that the future will not be any worse than the worst case in the past. If you believe this, then you believe that the portfolio of Mr. ER 2000 will survive as long as he sticks with the historical SWR. And if you believe that his portfolio will survive, then clearly, someone who retires today with identical withdrawal rate and identical end dates to Mr ER 2000 will survive.
What you are trying to do is take the current withdrawal rate of Mr. ER 2000 and back test it against history. If that were valid, then you would conclude that Mr. ER 2000 was 100% safe when he started with is 4.1% inflation adjusted withdrawals, but is now less than 100% safe with that same withdrawal strategy. But if he's no longer 100% safe, then what did 100% safe mean 4 years ago?
I am simply pointing out that consistency requires that a 100% SWR is 100% safe. If 4.1% were 100% safe in 2000, then over 6% must be safe today. If 6% is not safe today, then 4.1% must not have been safe in 2000.
You decide which you believe. But you can't believe both without contradicting yourself. I know a lot of ER's don't like this result. But that's just the logical result.
If this result makes you question the validity of a 4% SWR, then maybe you should question it. If it makes you feel more comfortable about your current retirement situation, then maybe you are right to feel comforted.
Re: SWR of 6.21% for 26 yearsThe SWR simulation is
What you are saying is that you believe that the future may be worse than the worst case of the past. If you believe that, then the results of the historical simulator are of limited value. I would not try to argue against that point. I certainly don't know if tomorrow will be worse than the Great Depression or not.
In fact, I would not have felt comfortable retiring with a plan to spend at a 4% withdrawal rate. Even if I believed 100% in the SWR result, I would be worried about the accuracty of my budget plan, my exact longevity, etc. I think most of us would be wise to build some cushion into our plans.
There is an implication in your analogy, however, that I think is still unproven. My friend (let's call him, TH) is apparently drunker than I've ever seen him. For your SWR theory, this is analogous to what you refer to as overvaluation. But I have trouble with your equating PE10 to that overvaluation. PE10 is a metric that approximately quantifies valuation under certain assumptions, but it is not a direct measure of the valuation definition that really determines future returns. I might be able to give TH a breathalyzer test and get a fairly accurate determination of his "drunkenness", but the quantity that would determine future returns would be closer to "current price- to- future profits" with a lot of "future investor sentiment" thrown in. The problem is that I have no way to quantify those future profits or future investor sentiments. PE10 uses past (10 years) where investors really care about future. So it approximates what investors care about only provided that the future is not too different than the past 10 years.
What you are saying is that you believe that the future may be worse than the worst case of the past. If you believe that, then the results of the historical simulator are of limited value. I would not try to argue against that point. I certainly don't know if tomorrow will be worse than the Great Depression or not.
In fact, I would not have felt comfortable retiring with a plan to spend at a 4% withdrawal rate. Even if I believed 100% in the SWR result, I would be worried about the accuracty of my budget plan, my exact longevity, etc. I think most of us would be wise to build some cushion into our plans.
There is an implication in your analogy, however, that I think is still unproven. My friend (let's call him, TH)
is apparently drunker than I've ever seen him. For your SWR theory, this is analogous to what you refer to as overvaluation. But I have trouble with your equating PE10 to that overvaluation. PE10 is a metric that approximately quantifies valuation under certain assumptions, but it is not a direct measure of the valuation definition that really determines future returns. I might be able to give TH a breathalyzer test and get a fairly accurate determination of his "drunkenness", but the quantity that would determine future returns would be closer to "current price- to- future profits" with a lot of "future investor sentiment" thrown in. The problem is that I have no way to quantify those future profits or future investor sentiments. PE10 uses past (10 years) where investors really care about future. So it approximates what investors care about only provided that the future is not too different than the past 10 years.Correlation is determined as the result of a mathematical analysis of data. No physical or psychological understanding of the data is required. Typically, when correlation between data sets is analyzed it is because cause and effect is not completely understood or too complex to quantify.
I certainly never said anything and don't believe anything like that. I discuss "correlation" precisely because much of what affects stock prices is too complex to quantify in a cause-and-effect formula.
You are simply not correct about this. Monte carlo analysis must assume something about the distribution of stock returns, bond returns and inflation. The analysis must assume something about the correlation between these distributions. And the analysis must assume somthing about the correlation of the data from year to year. What most monte carlo simulators do is fit the distributions of stock returns, bond returns and inflation to the distributions found from historical data. They then assume zero correlation between variables and year-to-year. By changing any of the assumptions, monte carlo simulators can produce SWR predictions that are more or less optimistic than historical SWR predictions.
FYI: Recently, I discussed some results with one of the posters over on the nofeeboards who had modified his own monte carlo simulator to include a RTM forcing function (ie correlate future returns to immediate past returns). The deviation between his monte carlo results and historical results were dramatically reduced.
I think you can study factors and determine with high probability of accuracy that the market will move in a certain direction -- or that certain metrics will adjust upward or downward. But picking both the direction and the timing of those moves is a very low probability prediction. So I don't argue with people who say that the market is currently overvalued. I see the same indicators they do and believe that is probably true. I don't argue with those people if they say we are probably in for disapointing returns at some point till the market becomes less overvalued. But if you tell me those dissapointing returns are going to happen next year or you tell me they are going to be dispersed over the next 30 years, I am very skeptical. I have never seen any reason to believe such predictions and I've seen a lot of reasons to be skeptical.
So I don't neccesarily believe in short term or mid-term predictions more than I believe in 30 year predictions. I just don't generally believe in timed predictions.
I certainly agree with this.
Okay. I have never said that is an incorrect position. But if you believe that 4.1% (ie. the universal SWR predicted by FIRECALC) was correct for the 2000 retiree while you believe that 6.21% won't work for the current retiree, then you are contradicting yourself. And if you don't believe that 4.1% was correct for Mr. RE 2000, then you must question the validity of the historical result. By rejecting the 6.21% result, consistency requires that you reject the underlying hypothesis of the historical simulation result -- that the future will be no worse than the worst case of the past.
I want to point out that I did not retire in 2001 as my original RE plan had indicated precisely because I did not feel comfortable with a withdrawal rate of even less than 4% during the economic cycle we were in the middle of. I think that even if you believed with 100% certainty in the historical calculator results, you should try to provide significant cushion in your retirement portfolio. The accuracy of your post-retirement budget, the timing of your own longevity, the changes you may see in tax law, social security, etc. are all issues that are somewhat unknowable and yet have significant affect on your results.
But I do believe that it is important to understand the implications of the historical simulation results. If you don't believe in the 6.2% SWR of the current retiree, then you should be questioning the validity of the 4% SWR result. Not to do so is inconsistent logic.
And if you have begun to doubt the 4% result, then you might begin to ask, "What withdrawal rate is safe and how can I determine it if the historical simulator can't provide it?"
***** believes he has some of those answers. You can believe in the uncorrelated results from monte carlo. . .
My own reaction to understanding why a 4% universal SWR result implies a 6.2% rate for the current retiree is to believe that maybe the future could be a little worse than the past, but maybe times aren't quite as bad as many of the doom and gloom predictors are indicating. Comparing today to 1929 seems extreme, but maybe not that extreme. On the other hand, comparing today to 1965 doesn't seem quite so extreme, but a 4% withdrawal rate would have worked then. I'm thinking that waiting to retire till 2003 just before the markets made a nice recovery was probably a good thing to do and that it probably bought me a healthy bit of extra withdrawal margin. So even if the period from 2000 to 2030 is a little worse than 1929 to 1959 or 1965 to 1995, I've truncated the first 3 years of that period and I should be okay.
wabmester
Thinks s/he gets paid by the post
- Joined
- Dec 6, 2003
- Messages
- 4,459
SG, you seem like a smart guy, so I'm curious about something. How can you use FIREcalc with even a smidgen of confidence knowing that for your term of interest (presumably 30 years), there are only 4 independent (i.e., non-overlapping) data subsets in the data set?
In my case, I'd have to guess that either my wife or I will live another 50 years, in which case relying on historical data only two "epochs" deep seems like folly.
A lot can happen in 50 (or even 30) years, including events that might make the Great Depression look like a party. How can anybody, especially somebody with a background in archaeology, think that the worst is behind us?
In my case, I'd have to guess that either my wife or I will live another 50 years, in which case relying on historical data only two "epochs" deep seems like folly.
A lot can happen in 50 (or even 30) years, including events that might make the Great Depression look like a party. How can anybody, especially somebody with a background in archaeology, think that the worst is behind us?
First, I have not defended the 4% result of the FIRECALC simulator in a single one of my posts. Come on, guys, read them. I simply point out the logical and mathematical conclusions you have to come to in order to be consistent. If you don't like 6.2% today, then you better re-think 4.1% ever. There seem to be a lot of posters on this board who want to believe in a 4% SWR while not believing that amt came to a logical conclusion. People should re-think that position because it makes no sense.
Now . . . regarding other points in your post:
I don't have that much confidence that the worst is behind us. I can believe we'll see some period in the future that is as difficult to overcome as either the 1929 - 19xx period or as difficult as the 1965 - xxxx period. It could be the period that began in 2000, but I certainly haven't seen any quantifiable argument that leads me to believe that is a high probability. And if the period that began in 2000 is going to be worse than anything in the past, how much worse should we expect? 10%? 20%? 50%? I don't hear anyone providing quantifiable answers. In general, I have seen no analysis by anyone that would lead me to believe that they might be able to predict how much worse the next 40+ years is going to be than the two worst cases we've seen so far.
From a mathematical perspective, the number of independent data sets is of little concern to me. If I want to empirically answer the question, "What is the minimum number of heads I will see if I flip a coin 30 times in a row?" I don't need to start 100 different series of 30 tosses (ie. 3000 tosses) to answer the question. I can simply flip the coin 130 times and look at the 100 different series of 30 that are represented in those tosses. Of course the correct answer to the coin toss question is "zero" -- a result that you would have a very small probability of ever seeing in practice (forgive me if I don't compute the odds against seeing that event). And the same is true for SWR. No SWR is guaranteed.
But even more important, if I had more data, I question it's value. Based on all the law and regulation changes that took place after the Great Depression, I question how valid pre-1929 data is to my situation toady. Do you really want to trust your retirement on calculations dating to the Paleolithic Period?
So we all agree that you could retire using a result from a historical simulation and still run into trouble. Now what do you suggest we do? Here are some choices:
Should we comitt suicide? This insures that any withdrawal rate is safe for us.
Should we just arbitrarily decide that 6.2% today sounds unreasonable to us and use faulty logic to back up our decision? Meanwhile we can steadfastly believe in the 4.1% SWR and ignore the inconsistency in our logic.
Should we develop new models based on unproven hypothesis and shaky mathematical analysis and believe in them? If no one has bothered to prove the new models wrong yet, maybe they're right.
Should we throw out all of the complex correlations built into historical data sets and trust arbitarily calibrated monte carlo simulations?
My answer is that we should make every effort to understand the implications and limitations of all the models available to us. I think once you've done that, you decide that if you start with a generous post-retirement budget (one you can cut back on if you need it), plan for a long life, discount your social security benefits, choose a reasonable allocation plan, and keep your initial withdrawal rate a little below 4%, you are probably going to be okay.
wabmester
Thinks s/he gets paid by the post
- Joined
- Dec 6, 2003
- Messages
- 4,459
I believe the odds are 1 in 230, the same as seeing any specific sequence of 30 tosses.
I posted another interesting calculation elsewhere (second-hand math):
Probability of a Great Depression magnitude event (i.e. 3-sigma GDP decline lasting 3 years) this year: about 1 in 526.
Probability of same in your lifetime: about 1 in 3.
haha
Give me a museum and I'll fill it. (Picasso) Give me a forum ...
This SWR discussion remeinds me of the story about the fox and the hedgehog. The fox knows many things, but the hedgehog knows one big thing.
What this hedgehog knows about SWR is that it is nuts to take 6% from a retirement portfolio when the stock portion of the portfolio yields about 1.5 %, and the fixed proportion somewhat less than 4% nominal.
If anyone actually plans to do this, as opposed to just discuss it, IMO he has more b*lls than brains.
Mikey
What this hedgehog knows about SWR is that it is nuts to take 6% from a retirement portfolio when the stock portion of the portfolio yields about 1.5 %, and the fixed proportion somewhat less than 4% nominal.
If anyone actually plans to do this, as opposed to just discuss it, IMO he has more b*lls than brains.
Mikey
wabmester
Thinks s/he gets paid by the post
- Joined
- Dec 6, 2003
- Messages
- 4,459
I think SG has made the case eloquently: 6.21% is perfectly safe if we have the same 30-year sequence as the historically worse sequence (1 of 4 statistically independent 30-year sequences). And the odds of repeating that sequence (or any other specific sequence) are about 1 in a billion.
If we get a different sequence, even without a catastrophic event like the Great Depression, all bets are off. All it would take is a streak of bad luck to blow the SWR assumptions out of the water.
(Unfortunately, I see low probability streaks often when I play blackjack -- I expect the same will happen when I play the stock market).
If we get a different sequence, even without a catastrophic event like the Great Depression, all bets are off. All it would take is a streak of bad luck to blow the SWR assumptions out of the water.
(Unfortunately, I see low probability streaks often when I play blackjack -- I expect the same will happen when I play the stock market).
There is an overhelming tendency to reify the SWR concept....SWR isn't a strategy - it's a crosscheck.
That's a good statement, Dory36.
I had to look up the word "reify." It means "to treat an abstraction as if it had concrete existence." That's a good word to make us of in describing what has happened to us.
SWR analysis is a magical tool, in my view. Aspiring early retirees putting together investment strategies are faced with scores of factors that they need to make sense of in assessing what may happen over the course of 30 or 40 or 50 years. It is an imposing task. SWR analysis takes all of those factors and packages them into a nice little bundle of a number. It makes the imposing task manageable. It provides information that is actionable.
That said, common sense trumps numerical calculation every time. When a given methodology churns out numbers that your common sense tells you could not possibly be valid, you need to take a step back and search three times for flaws in the methodology. Then you need to act on what you find. All the better if you can redesign the tool to bring it back into comformance with what your common sense tells you must be so.
We are in the early years of development of the SWR concept. The people who designed the conventional methodology are heros in my eyes. They gave us a gift, and we owe them. But they would not want us to accept the conventional methodology as the last word in SWR analysis. They would want us to change the tool, to update it, to enhance it. That is what I was trying to do when I developed the data-based SWR tool back in the mid-1990s. I was trying to take something good and make it better.
There is going to come a day when some whippersnapper is going to come along and try to knock me down from the perch from which I have become comfortable doing my crowing in recent years. I hope that when the time comes I either invite the whippernsnapper to give it his or her best, or, if I am too old and tired for that, just fly away.
Whippersnapping make the discussion board world go round. In the process of me being knocked from my perch, there are going to be some juicy threads appearing on the Retire Early boards of the day. Those threads are going to rock. I hope that I enjoy them as much when they play out as I today look forward with anticipation to the day they begin showing up on computer screens everywhere.
That's a good statement, Dory36.
I had to look up the word "reify." It means "to treat an abstraction as if it had concrete existence." That's a good word to make us of in describing what has happened to us.
SWR analysis is a magical tool, in my view. Aspiring early retirees putting together investment strategies are faced with scores of factors that they need to make sense of in assessing what may happen over the course of 30 or 40 or 50 years. It is an imposing task. SWR analysis takes all of those factors and packages them into a nice little bundle of a number. It makes the imposing task manageable. It provides information that is actionable.
That said, common sense trumps numerical calculation every time. When a given methodology churns out numbers that your common sense tells you could not possibly be valid, you need to take a step back and search three times for flaws in the methodology. Then you need to act on what you find. All the better if you can redesign the tool to bring it back into comformance with what your common sense tells you must be so.
We are in the early years of development of the SWR concept. The people who designed the conventional methodology are heros in my eyes. They gave us a gift, and we owe them. But they would not want us to accept the conventional methodology as the last word in SWR analysis. They would want us to change the tool, to update it, to enhance it. That is what I was trying to do when I developed the data-based SWR tool back in the mid-1990s. I was trying to take something good and make it better.
There is going to come a day when some whippersnapper is going to come along and try to knock me down from the perch from which I have become comfortable doing my crowing in recent years. I hope that when the time comes I either invite the whippernsnapper to give it his or her best, or, if I am too old and tired for that, just fly away.
Whippersnapping make the discussion board world go round. In the process of me being knocked from my perch, there are going to be some juicy threads appearing on the Retire Early boards of the day. Those threads are going to rock. I hope that I enjoy them as much when they play out as I today look forward with anticipation to the day they begin showing up on computer screens everywhere.
Similar threads
