The 4% rule DOES fail but still makes sense!

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 9:23 am

Here is a case written by sambb in one of my other threads. Similar examples have appeared in other places. I don't know if anyone had good answers.
*****************************
What about the following:
Need 100k/year, so desire to have 2.5M (4%) withdrawal.
Lets say right now in 2017, i hit 2.5M,so I could retire. Great.
Instead, lets say that i didn't check my balance. And we hit a bear market. And my 2.5M is now 2.0M
at 2.0 M in 2018 (after bear market) - my 4% won't give me 100k/year
So therefore, in one case I could retire in 2017, and in another, i couldn't.
******************************

First, many MMM people's claim is: the 4% rule never fails. They did not understand the statistics. The original 4% rule calculation had a success rate of about 96%, so it did fail in about 4% of the tested cases.

Let's come back to the sambb's example. Since FIRECalc is an approximation of the 4% rule, I use it to illustrate my points. Use 2.5M and 100K/year for 30 years, FIRECalc calculated 117 cycles with 6 failures and a success rate of 94.9%. (This is Test A.)
Use 2.0M and 100K/year for 29 years, FIRECalc calculated 118 cycles with 32 failures and a success rate of 72.9%. (This is Test B.)

It seems that we can view Test A as a success, and Test B as a failure. However, things are not so simple.

Note that in Test A, with the known situation of Test B, in the first year, the portfolio dropped from 2.5M to 2.0M, a 20% drop, which makes many tested cases (with first year return better than -20%) in Test A invalid. I guess if one could fix the first year return to be -20% in Test A, both Test A and Test B would have similar success rates. In this case, we say Test A suffered a Sequence of Return Risk.

Since I could not do FIRECalc with the first year return fixed, I just say without proof but with huge confidence that the 4% rules DOES fail but still makes sense.

To answer sambb's original question: Either you retire in 2017 (with 2.5M) or 2018 (with 2.0M), your success rate may be only around 73%.

Comments?

smitcat
Posts: 674
Joined: Mon Nov 07, 2016 10:51 am

Re: The 4% rule DOES fail but still makes sense!

Post by smitcat » Tue Nov 14, 2017 9:30 am

All rules fail at some larger or smaller %.
When the market is up the chances of decline are higher , when it is lower the chance of increases is higher.
There is no solution that will be 100% correct and attempting to prove that with math will be a long exercise in futility.

I believe the correct answer lies in having a balanced plan for funds and also having flexibility with withdrawals.

flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

Re: The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 9:34 am

smitcat wrote:
Tue Nov 14, 2017 9:30 am
All rules fail at some larger or smaller %.
When the market is up the chances of decline are higher , when it is lower the chance of increases is higher.
There is no solution that will be 100% correct and attempting to prove that with math will be a long exercise in futility.

I believe the correct answer lies in having a balanced plan for funds and also having flexibility with withdrawals.
I believe that the 4% rule is a good planning tool, i.e., to help you determine how much you may need to accumulate before retirement. Nobody should blindly follow the 4% rule in withdrawing.

User avatar
dwickenh
Posts: 790
Joined: Sun Jan 04, 2015 9:45 pm
Location: Illinois

Re: The 4% rule DOES fail but still makes sense!

Post by dwickenh » Tue Nov 14, 2017 9:55 am

smitcat wrote:
Tue Nov 14, 2017 9:30 am
All rules fail at some larger or smaller %.
When the market is up the chances of decline are higher , when it is lower the chance of increases is higher.
There is no solution that will be 100% correct and attempting to prove that with math will be a long exercise in futility.

I believe the correct answer lies in having a balanced plan for funds and also having flexibility with withdrawals.
+1
The 4% rule(guideline) is based on the results of a backward looking study. The future is less predictable than the past. Your own
use of the guideline is more important than the guideline. Adjustments are necessary. Allocation is important in the withdrawal rate.

Best wishes,

Dan
The market is the most efficient mechanism anywhere in the world for transferring wealth from impatient people to patient people.” | — Warren Buffett

kalrocmk
Posts: 21
Joined: Wed Apr 08, 2015 10:51 am

Re: The 4% rule DOES fail but still makes sense!

Post by kalrocmk » Tue Nov 14, 2017 10:03 am

You have to take the context into account. The original Bengen study based it on actual market returns. The worst starting year was in 1967 IIRC. If your portfolio drops by 20% in year 1 of 30 the subsequent 29 years will likely be better because the markets have better valuations after the drop. When you restart the clock after the first year, you are sort of double dipping on the losses. IMO all the study tells you is that you have better than 90% chance of not running out of money if you withdraw 4% of your initial portfolio adjusted for inflation for 30 years, even if you start at a market top. your odds are better if you do not start at a market top.
Just trying to make a living

flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

Re: The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 10:08 am

kalrocmk wrote:
Tue Nov 14, 2017 10:03 am
You have to take the context into account. The original Bengen study based it on actual market returns. The worst starting year was in 1967 IIRC. If your portfolio drops by 20% in year 1 of 30 the subsequent 29 years will likely be better because the markets have better valuations after the drop. When you restart the clock after the first year, you are sort of double dipping on the losses. IMO all the study tells you is that you have better than 90% chance of not running out of money if you withdraw 4% of your initial portfolio adjusted for inflation for 30 years, even if you start at a market top. your odds are better if you do not start at a market top.
The odds does not change, since you now have a smaller portfolio by 20%.

smitcat
Posts: 674
Joined: Mon Nov 07, 2016 10:51 am

Re: The 4% rule DOES fail but still makes sense!

Post by smitcat » Tue Nov 14, 2017 10:41 am

flyingaway wrote:
Tue Nov 14, 2017 10:08 am
kalrocmk wrote:
Tue Nov 14, 2017 10:03 am
You have to take the context into account. The original Bengen study based it on actual market returns. The worst starting year was in 1967 IIRC. If your portfolio drops by 20% in year 1 of 30 the subsequent 29 years will likely be better because the markets have better valuations after the drop. When you restart the clock after the first year, you are sort of double dipping on the losses. IMO all the study tells you is that you have better than 90% chance of not running out of money if you withdraw 4% of your initial portfolio adjusted for inflation for 30 years, even if you start at a market top. your odds are better if you do not start at a market top.
The odds does not change, since you now have a smaller portfolio by 20%.
I believe that you are saying that the 4% rule now starts with the lower portfolio and the same level of probability that the market will go up or down.
I believe that kalrocnk is saying that the probability that the market will go up improves after a decline.

When I look at any back testing charts they tell me that lower starting points will yield higher mid term yields and that higher starting points will yield lower mid term yields.
I remain faithful to the fact that you need tp have a balanced approach and a flexible withdrawal plan for maximum certainty/safety.

And FWIW- thanks to this forum for teaching me this stuff as I was mostly ignorant of these facts prior to visiting here.

caffeperfavore
Posts: 18
Joined: Thu Aug 06, 2015 11:45 am

Re: The 4% rule DOES fail but still makes sense!

Post by caffeperfavore » Tue Nov 14, 2017 10:42 am

flyingaway wrote:
Tue Nov 14, 2017 9:23 am
First, many MMM people's claim is: the 4% rule never fails. They did not understand the statistics. The original 4% rule calculation had a success rate of about 96%, so it did fail in about 4% of the tested cases.
As someone who's spent a fair amount of time lurking on the MMM blog and boards, I'm not sure this is true. Most people there are aware of the failure rate, but find it to be an acceptable degree of risk, especially if you are able to supplement your income by picking up a little side work if needed. Personally, I'll happily take odds of 96% or even 90%, but I'll have a Plan B in case the 4 or 10% or whatever % ends up being the case or if the future does not look like the past.

dbr
Posts: 23755
Joined: Sun Mar 04, 2007 9:50 am

Re: The 4% rule DOES fail but still makes sense!

Post by dbr » Tue Nov 14, 2017 10:47 am

smitcat wrote:
Tue Nov 14, 2017 10:41 am
flyingaway wrote:
Tue Nov 14, 2017 10:08 am
kalrocmk wrote:
Tue Nov 14, 2017 10:03 am
You have to take the context into account. The original Bengen study based it on actual market returns. The worst starting year was in 1967 IIRC. If your portfolio drops by 20% in year 1 of 30 the subsequent 29 years will likely be better because the markets have better valuations after the drop. When you restart the clock after the first year, you are sort of double dipping on the losses. IMO all the study tells you is that you have better than 90% chance of not running out of money if you withdraw 4% of your initial portfolio adjusted for inflation for 30 years, even if you start at a market top. your odds are better if you do not start at a market top.
The odds does not change, since you now have a smaller portfolio by 20%.
I believe that you are saying that the 4% rule now starts with the lower portfolio and the same level of probability that the market will go up or down.
I believe that kalrocnk is saying that the probability that the market will go up improves after a decline.

When I look at any back testing charts they tell me that lower starting points will yield higher mid term yields and that higher starting points will yield lower mid term yields.
I remain faithful to the fact that you need tp have a balanced approach and a flexible withdrawal plan for maximum certainty/safety.

And FWIW- thanks to this forum for teaching me this stuff as I was mostly ignorant of these facts prior to visiting here.
The issue here is whether or not one can effectively estimate retirement outcomes using probabilities conditional on knowledge of the current situation, such as valuation, that there has been a recent decline, etc. I think the theoretical answer to that is that you can but the practical answer to that is that you can't because there isn't enough data to get enough certainty to make distinctions. The results using the entire set of historical information on average are already just estimates with significant uncertainty. Kitces has written a paper arguing that you maintain your withdrawal after a market decline because the future return should be greater than average. One could also examine what might be learned from the Variable Percentage Withdrawal method.

smitcat
Posts: 674
Joined: Mon Nov 07, 2016 10:51 am

Re: The 4% rule DOES fail but still makes sense!

Post by smitcat » Tue Nov 14, 2017 10:50 am

dbr wrote:
Tue Nov 14, 2017 10:47 am
smitcat wrote:
Tue Nov 14, 2017 10:41 am
flyingaway wrote:
Tue Nov 14, 2017 10:08 am
kalrocmk wrote:
Tue Nov 14, 2017 10:03 am
You have to take the context into account. The original Bengen study based it on actual market returns. The worst starting year was in 1967 IIRC. If your portfolio drops by 20% in year 1 of 30 the subsequent 29 years will likely be better because the markets have better valuations after the drop. When you restart the clock after the first year, you are sort of double dipping on the losses. IMO all the study tells you is that you have better than 90% chance of not running out of money if you withdraw 4% of your initial portfolio adjusted for inflation for 30 years, even if you start at a market top. your odds are better if you do not start at a market top.
The odds does not change, since you now have a smaller portfolio by 20%.
I believe that you are saying that the 4% rule now starts with the lower portfolio and the same level of probability that the market will go up or down.
I believe that kalrocnk is saying that the probability that the market will go up improves after a decline.

When I look at any back testing charts they tell me that lower starting points will yield higher mid term yields and that higher starting points will yield lower mid term yields.
I remain faithful to the fact that you need tp have a balanced approach and a flexible withdrawal plan for maximum certainty/safety.

And FWIW- thanks to this forum for teaching me this stuff as I was mostly ignorant of these facts prior to visiting here.
The issue here is whether or not one can effectively estimate retirement outcomes using probabilities conditional on knowledge of the current situation, such as valuation, that there has been a recent decline, etc. I think the theoretical answer to that is that you can but the practical answer to that is that you can't because there isn't enough data to get enough certainty to make distinctions. The results using the entire set of historical information on average are already just estimates with significant uncertainty. Kitces has written a paper arguing that you maintain your withdrawal after a market decline because the future return should be greater than average. One could also examine what might be learned from the Variable Percentage Withdrawal method.
You can always measure the past with certainty - you cannot measure or predict the future with certainty.

dbr
Posts: 23755
Joined: Sun Mar 04, 2007 9:50 am

Re: The 4% rule DOES fail but still makes sense!

Post by dbr » Tue Nov 14, 2017 10:51 am

caffeperfavore wrote:
Tue Nov 14, 2017 10:42 am
flyingaway wrote:
Tue Nov 14, 2017 9:23 am
First, many MMM people's claim is: the 4% rule never fails. They did not understand the statistics. The original 4% rule calculation had a success rate of about 96%, so it did fail in about 4% of the tested cases.
As someone who's spent a fair amount of time lurking on the MMM blog and boards, I'm not sure this is true. Most people there are aware of the failure rate, but find it to be an acceptable degree of risk, especially if you are able to supplement your income by picking up a little side work if needed. Personally, I'll happily take odds of 96% or even 90%, but I'll have a Plan B in case the 4 or 10% or whatever % ends up being the case or if the future does not look like the past.
In a historical periods model there are perhaps a hundred periods so plus or minus one means nothing,including that 100% success is meaningless. A couple of failures out of a hundred is not of concern. In a Monte-Carlo simulation it might be conventional to decree 95% as "successful." It has been pointed out by Bernstein in particular that life itself is uncertain enough that past 85% success enough other things intervene that security is illusory. Swedroe suggests that everyone have a Plan B. Finally, actual investment, income, and spending are variable enough that 4% just can't be an actual plan at all.

flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

Re: The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 10:53 am

caffeperfavore wrote:
Tue Nov 14, 2017 10:42 am
flyingaway wrote:
Tue Nov 14, 2017 9:23 am
First, many MMM people's claim is: the 4% rule never fails. They did not understand the statistics. The original 4% rule calculation had a success rate of about 96%, so it did fail in about 4% of the tested cases.
As someone who's spent a fair amount of time lurking on the MMM blog and boards, I'm not sure this is true. Most people there are aware of the failure rate, but find it to be an acceptable degree of risk, especially if you are able to supplement your income by picking up a little side work if needed. Personally, I'll happily take odds of 96% or even 90%, but I'll have a Plan B in case the 4 or 10% or whatever % ends up being the case or if the future does not look like the past.
I also play on the MMM forum and use the same screen name.

kaudrey
Posts: 823
Joined: Fri Nov 22, 2013 2:40 pm

Re: The 4% rule DOES fail but still makes sense!

Post by kaudrey » Tue Nov 14, 2017 11:46 am

OP - that is not how the 4% rule/guideline works. What you are describing is a variable withdrawal plan.

The 4% rule states that you withdraw 4% the first year, and increase that amount for inflation each year. You don't take 4% of the balance each year.

So, for example, if you have $2.5M, the first year you withdraw $100K. The second year you withdraw, say $102K, assuming 2% inflation.

Now, of course, in practice, if the market tanks, people are likely to be more conservative in their spending; but, that is not what the outcomes you are talking about are reflecting.

TonyDAntonio
Posts: 315
Joined: Thu Mar 03, 2016 8:32 pm

Re: The 4% rule DOES fail but still makes sense!

Post by TonyDAntonio » Tue Nov 14, 2017 11:48 am

I plan on the 4% never failing. I take 4% of whatever is left every year. 😀

wrongfunds
Posts: 837
Joined: Tue Dec 21, 2010 3:55 pm

Re: The 4% rule DOES fail but still makes sense!

Post by wrongfunds » Tue Nov 14, 2017 12:29 pm

In a historical periods model there are perhaps a hundred periods so plus or minus one means nothing,including that 100% success is meaningless.
Can you explain why there are only 100 periods? Is it because one only has data for a single point in a year? What if one had data for every quarter or better, data for every single trading day or for every single trading hour?

Why do you need to take 4% at the beginning of the year? Why not take 0.333% every month instead? 4/26 % every other week aka like your paycheck?Would you not get different results? Now that we have the computers, this type study should be trivial. Do the results converge? This is how real world scientific (e.g. physical or chemical) data would be analyzed.

If the 100% is not a guarantee then 95% is not a failure either.

dcabler
Posts: 293
Joined: Wed Feb 19, 2014 11:30 am

Re: The 4% rule DOES fail but still makes sense!

Post by dcabler » Tue Nov 14, 2017 12:37 pm

kaudrey wrote:
Tue Nov 14, 2017 11:46 am
OP - that is not how the 4% rule/guideline works. What you are describing is a variable withdrawal plan.

The 4% rule states that you withdraw 4% the first year, and increase that amount for inflation each year. You don't take 4% of the balance each year.

So, for example, if you have $2.5M, the first year you withdraw $100K. The second year you withdraw, say $102K, assuming 2% inflation.

Now, of course, in practice, if the market tanks, people are likely to be more conservative in their spending; but, that is not what the outcomes you are talking about are reflecting.
+1 I was going to say the same thing. The OP is mixing dollar amounts and %. It still work in his example, it's just that it's now 4% of $2M instead of 4% of $2.5M and the likelihood of failure doesn't change in either case. If the OP, for example, needs to have $100K per year, then he will need at least $2.5M.

User avatar
David Jay
Posts: 3995
Joined: Mon Mar 30, 2015 5:54 am
Location: Michigan

Re: The 4% rule DOES fail but still makes sense!

Post by David Jay » Tue Nov 14, 2017 12:38 pm

TonyDAntonio wrote:
Tue Nov 14, 2017 11:48 am
I plan on the 4% never failing. I take 4% of whatever is left every year. 😀
Merriman's work suggests that 6% is a realistic withdrawal rate for a variable withdrawal strategy. I am planning on a 5% variable withdrawal strategy.
Prediction is very difficult, especially about the future - Niels Bohr | To get the "risk premium", you really do have to take the risk - nisiprius

dspencer
Posts: 122
Joined: Wed Jul 06, 2016 11:29 am

Re: The 4% rule DOES fail but still makes sense!

Post by dspencer » Tue Nov 14, 2017 12:54 pm

flyingaway wrote:
Tue Nov 14, 2017 9:23 am
Here is a case written by sambb in one of my other threads. Similar examples have appeared in other places. I don't know if anyone had good answers.
*****************************
What about the following:
Need 100k/year, so desire to have 2.5M (4%) withdrawal.
Lets say right now in 2017, i hit 2.5M,so I could retire. Great.
Instead, lets say that i didn't check my balance. And we hit a bear market. And my 2.5M is now 2.0M
at 2.0 M in 2018 (after bear market) - my 4% won't give me 100k/year
So therefore, in one case I could retire in 2017, and in another, i couldn't.
******************************

First, many MMM people's claim is: the 4% rule never fails. They did not understand the statistics. The original 4% rule calculation had a success rate of about 96%, so it did fail in about 4% of the tested cases.

Let's come back to the sambb's example. Since FIRECalc is an approximation of the 4% rule, I use it to illustrate my points. Use 2.5M and 100K/year for 30 years, FIRECalc calculated 117 cycles with 6 failures and a success rate of 94.9%. (This is Test A.)
Use 2.0M and 100K/year for 29 years, FIRECalc calculated 118 cycles with 32 failures and a success rate of 72.9%. (This is Test B.)

It seems that we can view Test A as a success, and Test B as a failure. However, things are not so simple.

Note that in Test A, with the known situation of Test B, in the first year, the portfolio dropped from 2.5M to 2.0M, a 20% drop, which makes many tested cases (with first year return better than -20%) in Test A invalid. I guess if one could fix the first year return to be -20% in Test A, both Test A and Test B would have similar success rates. In this case, we say Test A suffered a Sequence of Return Risk.

Since I could not do FIRECalc with the first year return fixed, I just say without proof but with huge confidence that the 4% rules DOES fail but still makes sense.

To answer sambb's original question: Either you retire in 2017 (with 2.5M) or 2018 (with 2.0M), your success rate may be only around 73%.

Comments?
I have thought about the same question. If tools like FIRECalc were ironclad statistical facts about the future, then it would be easy to understand the situation. Once you have more information (ie returns for year 1) then the current information is more accurate and represents the statistical "truth". So if rerun the calculations after each year that should be the best prediction about the future.

The problem is that it's not really that simple and FIRECalc is simply looking at the past and assuming the future will be similar. So the concept of valuations might be valid; meaning that starting a 29 year period with 2.0 MM and 100K/year withdrawal might have better than 73% success rate if that year starts right after a large drop in the market.

My guess is the "truth" or "likely reality" is somewhere in between. Once you know that year 1 is a big drop, you can't assume your odds of success are still the same as they were when starting out. This becomes obvious when looking at an extreme situation: for example after 25 years you run out of money. It's pretty clear you can't keep banking on the original stats being relevant when you've already found yourself in a failure case. On the other hand, there's probably some validity that your odds are better if you start after a big market drop than starting off in a random time period.

For me, the answer is to be a bit more conservative if you are doing projections at a time with high valuations and low interest rates. Maybe 3.5% is more appropriate at a time like today whereas if you still have your number after the great crash of 20?? then 4% may be fine.

TonyDAntonio
Posts: 315
Joined: Thu Mar 03, 2016 8:32 pm

Re: The 4% rule DOES fail but still makes sense!

Post by TonyDAntonio » Tue Nov 14, 2017 1:12 pm

David Jay wrote:
Tue Nov 14, 2017 12:38 pm
TonyDAntonio wrote:
Tue Nov 14, 2017 11:48 am
I plan on the 4% never failing. I take 4% of whatever is left every year. 😀
Merriman's work suggests that 6% is a realistic withdrawal rate for a variable withdrawal strategy. I am planning on a 5% variable withdrawal strategy.
99% works too. All kidding aside I think most folks on this board will use some sort of variable spending percentage and error on the side of spending too little.

Admiral
Posts: 863
Joined: Mon Oct 27, 2014 12:35 pm

Re: The 4% rule DOES fail but still makes sense!

Post by Admiral » Tue Nov 14, 2017 1:19 pm

If you cannot, in any way, adjust your living expenses downward such that you don't HAVE to withdraw 4% per year, then you don't have enough money. It's as simple as that. You need some flexibility when on a fixed income if that income is based on investments.

Variable Withdrawal Rate is a better "rule" if one needs such a rule. 4% has been shown to be safe most of the time, not all of the time, and particularly because the future may not be like the past.

User avatar
David Jay
Posts: 3995
Joined: Mon Mar 30, 2015 5:54 am
Location: Michigan

Re: The 4% rule DOES fail but still makes sense!

Post by David Jay » Tue Nov 14, 2017 1:20 pm

TonyDAntonio wrote:
Tue Nov 14, 2017 1:12 pm
David Jay wrote:
Tue Nov 14, 2017 12:38 pm
TonyDAntonio wrote:
Tue Nov 14, 2017 11:48 am
I plan on the 4% never failing. I take 4% of whatever is left every year. 😀
Merriman's work suggests that 6% is a realistic withdrawal rate for a variable withdrawal strategy. I am planning on a 5% variable withdrawal strategy.
99% works too. All kidding aside I think most folks on this board will use some sort of variable spending percentage and error on the side of spending too little.
Yea, that would provide serious variability for the first 2 years. :(
Prediction is very difficult, especially about the future - Niels Bohr | To get the "risk premium", you really do have to take the risk - nisiprius

flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

Re: The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 1:23 pm

dcabler wrote:
Tue Nov 14, 2017 12:37 pm
kaudrey wrote:
Tue Nov 14, 2017 11:46 am
OP - that is not how the 4% rule/guideline works. What you are describing is a variable withdrawal plan.

The 4% rule states that you withdraw 4% the first year, and increase that amount for inflation each year. You don't take 4% of the balance each year.

So, for example, if you have $2.5M, the first year you withdraw $100K. The second year you withdraw, say $102K, assuming 2% inflation.

Now, of course, in practice, if the market tanks, people are likely to be more conservative in their spending; but, that is not what the outcomes you are talking about are reflecting.
+1 I was going to say the same thing. The OP is mixing dollar amounts and %. It still work in his example, it's just that it's now 4% of $2M instead of 4% of $2.5M and the likelihood of failure doesn't change in either case. If the OP, for example, needs to have $100K per year, then he will need at least $2.5M.
I did not mix anything. I was trying to explain the examples that I saw in several places, which tried to discredit the 4% rule.

flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

Re: The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 1:26 pm

kaudrey wrote:
Tue Nov 14, 2017 11:46 am
OP - that is not how the 4% rule/guideline works. What you are describing is a variable withdrawal plan.

The 4% rule states that you withdraw 4% the first year, and increase that amount for inflation each year. You don't take 4% of the balance each year.

So, for example, if you have $2.5M, the first year you withdraw $100K. The second year you withdraw, say $102K, assuming 2% inflation.

Now, of course, in practice, if the market tanks, people are likely to be more conservative in their spending; but, that is not what the outcomes you are talking about are reflecting.
I think I fully understand the 4% rule and what you said. I was trying to explain the example used in my OP. If one is not careful, one might be fooled by the example, which has been used in a few places to show that the 4% rule is a joke.

Da5id
Posts: 1519
Joined: Fri Feb 26, 2016 8:20 am

Re: The 4% rule DOES fail but still makes sense!

Post by Da5id » Tue Nov 14, 2017 1:26 pm

wrongfunds wrote:
Tue Nov 14, 2017 12:29 pm
In a historical periods model there are perhaps a hundred periods so plus or minus one means nothing,including that 100% success is meaningless.
Can you explain why there are only 100 periods? Is it because one only has data for a single point in a year? What if one had data for every quarter or better, data for every single trading day or for every single trading hour?

Why do you need to take 4% at the beginning of the year? Why not take 0.333% every month instead? 4/26 % every other week aka like your paycheck?Would you not get different results? Now that we have the computers, this type study should be trivial. Do the results converge? This is how real world scientific (e.g. physical or chemical) data would be analyzed.

If the 100% is not a guarantee then 95% is not a failure either.
There are less than 100 independent periods. Unless you consider two periods, where 29 out of the 30 years are the same, are "independent" which seems crazy. That means really there is less data than you think. If you break down by hour it gets sillier.

These threads always are fun. 3% is the new 4% :)

itstoomuch
Posts: 4732
Joined: Mon Dec 15, 2014 12:17 pm
Location: midValley OR

Re: The 4% rule DOES fail but still makes sense!

Post by itstoomuch » Tue Nov 14, 2017 1:27 pm

Don't want to ever test the theory :annoyed :greedy
Ymmv
Rev90517; 4 Incm stream buckets: SS+pension; dfr'd GLWB VA & FI anntys, by time & $$ laddered; Discretionary; Rentals. LTCi. Own, not asset. Tax 25%. Early SS. FundRatio (FR) >1.1 67/70yo

flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

Re: The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 1:33 pm

To ALL:

In the OP example, which was used in several places, some people wanted to show that when person A and person B (Test A and Test B), with the same portfolio and same spending level, which person B, retired in 2018, will fail according to the 4% rule, while person A, retired in 2017, will succeed.

Note that person A spends money in 2017. Person B does not spend money in 2017. So in 2018, person B has more money than person A, but the FIRECalc says that person B will fail, however, predicted one year ago person A will succeed.

My "analysis" says that they are comparing different things.
Last edited by flyingaway on Tue Nov 14, 2017 3:48 pm, edited 1 time in total.

marcopolo
Posts: 267
Joined: Sat Dec 03, 2016 10:22 am

Re: The 4% rule DOES fail but still makes sense!

Post by marcopolo » Tue Nov 14, 2017 1:35 pm

flyingaway wrote:
Tue Nov 14, 2017 9:23 am
Here is a case written by sambb in one of my other threads. Similar examples have appeared in other places. I don't know if anyone had good answers.
*****************************
What about the following:
Need 100k/year, so desire to have 2.5M (4%) withdrawal.
Lets say right now in 2017, i hit 2.5M,so I could retire. Great.
Instead, lets say that i didn't check my balance. And we hit a bear market. And my 2.5M is now 2.0M
at 2.0 M in 2018 (after bear market) - my 4% won't give me 100k/year
So therefore, in one case I could retire in 2017, and in another, i couldn't.
******************************

First, many MMM people's claim is: the 4% rule never fails. They did not understand the statistics. The original 4% rule calculation had a success rate of about 96%, so it did fail in about 4% of the tested cases.

Let's come back to the sambb's example. Since FIRECalc is an approximation of the 4% rule, I use it to illustrate my points. Use 2.5M and 100K/year for 30 years, FIRECalc calculated 117 cycles with 6 failures and a success rate of 94.9%. (This is Test A.)
Use 2.0M and 100K/year for 29 years, FIRECalc calculated 118 cycles with 32 failures and a success rate of 72.9%. (This is Test B.)

It seems that we can view Test A as a success, and Test B as a failure. However, things are not so simple.

Note that in Test A, with the known situation of Test B, in the first year, the portfolio dropped from 2.5M to 2.0M, a 20% drop, which makes many tested cases (with first year return better than -20%) in Test A invalid. I guess if one could fix the first year return to be -20% in Test A, both Test A and Test B would have similar success rates. In this case, we say Test A suffered a Sequence of Return Risk.

Since I could not do FIRECalc with the first year return fixed, I just say without proof but with huge confidence that the 4% rules DOES fail but still makes sense.

To answer sambb's original question: Either you retire in 2017 (with 2.5M) or 2018 (with 2.0M), your success rate may be only around 73%.

Comments?

I am not sure if there are enough periods to test this, but i think a more useful way of looking at this might be to consider the survival of the $100k out of $2M in context.

For example, portfolio is $2.5M, $100k Withdrawal (4% WR). Assume 60/40 portfolio, so a drop to $2M is a ~33% drop in the stock portion (assume flat fixed income to simplify), and $100k withdraw is a 5% WR. The question is how often did a 5% WR survive for 29 years when the start point is immediately AFTER a 33% drop in equity markets? I would suspect that the survival rate for that scenario is quite high.
Once in a while you get shown the light, in the strangest of places if you look at it right.

itstoomuch
Posts: 4732
Joined: Mon Dec 15, 2014 12:17 pm
Location: midValley OR

Re: The 4% rule DOES fail but still makes sense!

Post by itstoomuch » Tue Nov 14, 2017 1:39 pm

Older Bro at one time worked in risk management at Big Bank, still does in another capacity. He and his like managers thought that they had enough safety built into the algorithms. Unknown to them was that very senior management had ignored/changed the stops/rules. Thus the algorithms spewed out meaningless data to the true condition and risk that this Big Bank possessed. :? :annoyed :oops:
Ymmv :idea:
Rev90517; 4 Incm stream buckets: SS+pension; dfr'd GLWB VA & FI anntys, by time & $$ laddered; Discretionary; Rentals. LTCi. Own, not asset. Tax 25%. Early SS. FundRatio (FR) >1.1 67/70yo

Da5id
Posts: 1519
Joined: Fri Feb 26, 2016 8:20 am

Re: The 4% rule DOES fail but still makes sense!

Post by Da5id » Tue Nov 14, 2017 1:39 pm

flyingaway wrote:
Tue Nov 14, 2017 1:33 pm
To ALL:

In the OP example, which was used in several places, some people wanted to show that when person A and person B (Test A and Test B), with the same portfolio and same spending level, which person B, retired in 2018, will fail according to the 4% rule, while person A, retired in 2017, will succeed.

Note that person A spends money in 2017. Person B does not spend money in 2017. So in 2018, person B has more money than person A, but the 4% rule says that person B will fail, however, predicted one year ago person A will succeed.

My "analysis" says that they are comparing different things.
The 4% rule has always been simplistic and has absurd seeming results. That is one of the criticisms. It is just a "rule of thumb", not a retirement spending plan. The inconsistency is often seen in the following

Person A retires in year X with 800K. They can spend 32K/year, adjusting up for inflation, per the 4% rule.
Person B retires in year X+1. They have 1000K, they can spend 40K/year, adjusting for inflation. But the stock market boomed, and person A now has 1000K too. But they can only spend 32K (inflation adjusted) if following the 4% "rule", despite having 1 fewer year to fund than person B.

It is perverse if followed literally as a spending plan, almost all would agree. But as an estimate of "can I retire" 25-33x expenditures seems like an OK guestimate.

marcopolo
Posts: 267
Joined: Sat Dec 03, 2016 10:22 am

Re: The 4% rule DOES fail but still makes sense!

Post by marcopolo » Tue Nov 14, 2017 1:43 pm

flyingaway wrote:
Tue Nov 14, 2017 1:33 pm
To ALL:

In the OP example, which was used in several places, some people wanted to show that when person A and person B (Test A and Test B), with the same portfolio and same spending level, which person B, retired in 2018, will fail according to the 4% rule, while person A, retired in 2017, will succeed.

Note that person A spends money in 2017. Person B does not spend money in 2017. So in 2018, person B has more money than person A, but the 4% rule says that person B will fail, however, predicted one year ago person A will succeed.

My "analysis" says that they are comparing different things.
I don't think the 4% rule says that at all. Person A and B are essentially the same person, and will have the same probability of success (based on historical data). Person B is actually a little better off since they did not take withdrawals in year one.

Based on historical data, if Person B starts at 4% WR absed on current portfolio value ($80k/yr, instead of $100k/yr like person A), they will likely end up as more than half the scenarios where the ending portfolio size is quite a bit bigger than then starting value.

Now whether the 4% rule will hold true in the future is anyone's guess...
Once in a while you get shown the light, in the strangest of places if you look at it right.

User avatar
Artsdoctor
Posts: 2958
Joined: Thu Jun 28, 2012 3:09 pm
Location: Los Angeles, CA

Re: The 4% rule DOES fail but still makes sense!

Post by Artsdoctor » Tue Nov 14, 2017 2:11 pm

flyingaway wrote:
Tue Nov 14, 2017 9:23 am
Here is a case written by sambb in one of my other threads. Similar examples have appeared in other places. I don't know if anyone had good answers.
*****************************
What about the following:
Need 100k/year, so desire to have 2.5M (4%) withdrawal.
Lets say right now in 2017, i hit 2.5M,so I could retire. Great.
Instead, lets say that i didn't check my balance. And we hit a bear market. And my 2.5M is now 2.0M
at 2.0 M in 2018 (after bear market) - my 4% won't give me 100k/year
So therefore, in one case I could retire in 2017, and in another, i couldn't.
******************************

First, many MMM people's claim is: the 4% rule never fails. They did not understand the statistics. The original 4% rule calculation had a success rate of about 96%, so it did fail in about 4% of the tested cases.

Let's come back to the sambb's example. Since FIRECalc is an approximation of the 4% rule, I use it to illustrate my points. Use 2.5M and 100K/year for 30 years, FIRECalc calculated 117 cycles with 6 failures and a success rate of 94.9%. (This is Test A.)
Use 2.0M and 100K/year for 29 years, FIRECalc calculated 118 cycles with 32 failures and a success rate of 72.9%. (This is Test B.)

It seems that we can view Test A as a success, and Test B as a failure. However, things are not so simple.

Note that in Test A, with the known situation of Test B, in the first year, the portfolio dropped from 2.5M to 2.0M, a 20% drop, which makes many tested cases (with first year return better than -20%) in Test A invalid. I guess if one could fix the first year return to be -20% in Test A, both Test A and Test B would have similar success rates. In this case, we say Test A suffered a Sequence of Return Risk.

Since I could not do FIRECalc with the first year return fixed, I just say without proof but with huge confidence that the 4% rules DOES fail but still makes sense.

To answer sambb's original question: Either you retire in 2017 (with 2.5M) or 2018 (with 2.0M), your success rate may be only around 73%.

Comments?
In order to answer your own questions, you'll need to understand the 1994 study on which the 4% SWR was originally calculated. Once you've actually looked into that study and subsequent studies, you'll see that there are many caveats. You can start with a variety of synopses but Michael Kitces does a reasonable job:

https://www.aicpa.org/interestareas/per ... ch2012.pdf

User avatar
Kalo
Posts: 429
Joined: Sat May 25, 2013 1:01 pm

Re: The 4% rule DOES fail but still makes sense!

Post by Kalo » Tue Nov 14, 2017 2:34 pm

Another problem is behavioral. The studies assume rebalancing and/or not changing to a safer AA midstream. I can't be sure I would not succumb to this as my earning power and willingness to return to the work force are declining.

My solution is LMP to cover basic needs, and risk portfolio to cover luxuries and legacy. If the markets don't cooperate I can still live comfortably. Luxuries and legacy can suffer and are less likely to drive me to behavioral errors.

Kalo
"When people say they have a high risk tolerance, what they really mean is that they are willing to make a lot of money." -- Ben Stein/Phil DeMuth - The Little Book of Bullet Proof Investing.

avalpert
Posts: 6197
Joined: Sat Mar 22, 2008 4:58 pm

Re: The 4% rule DOES fail but still makes sense!

Post by avalpert » Tue Nov 14, 2017 2:39 pm

flyingaway wrote:
Tue Nov 14, 2017 9:23 am
Here is a case written by sambb in one of my other threads. Similar examples have appeared in other places. I don't know if anyone had good answers.
*****************************
What about the following:
Need 100k/year, so desire to have 2.5M (4%) withdrawal.
Lets say right now in 2017, i hit 2.5M,so I could retire. Great.
Instead, lets say that i didn't check my balance. And we hit a bear market. And my 2.5M is now 2.0M
at 2.0 M in 2018 (after bear market) - my 4% won't give me 100k/year
So therefore, in one case I could retire in 2017, and in another, i couldn't.
******************************

First, many MMM people's claim is: the 4% rule never fails. They did not understand the statistics. The original 4% rule calculation had a success rate of about 96%, so it did fail in about 4% of the tested cases.

Let's come back to the sambb's example. Since FIRECalc is an approximation of the 4% rule, I use it to illustrate my points. Use 2.5M and 100K/year for 30 years, FIRECalc calculated 117 cycles with 6 failures and a success rate of 94.9%. (This is Test A.)
Use 2.0M and 100K/year for 29 years, FIRECalc calculated 118 cycles with 32 failures and a success rate of 72.9%. (This is Test B.)

It seems that we can view Test A as a success, and Test B as a failure. However, things are not so simple.

Note that in Test A, with the known situation of Test B, in the first year, the portfolio dropped from 2.5M to 2.0M, a 20% drop, which makes many tested cases (with first year return better than -20%) in Test A invalid. I guess if one could fix the first year return to be -20% in Test A, both Test A and Test B would have similar success rates. In this case, we say Test A suffered a Sequence of Return Risk.

Since I could not do FIRECalc with the first year return fixed, I just say without proof but with huge confidence that the 4% rules DOES fail but still makes sense.

To answer sambb's original question: Either you retire in 2017 (with 2.5M) or 2018 (with 2.0M), your success rate may be only around 73%.

Comments?
So let's say you are on a game show with three doors. Behind one door is a car and behind the other two are goats. You are asked to pick the door with a car and if you are right you win it - you pick door #1. Monty, who knows what’s behind all three doors, opens door #2 showing you a goat. Before opening the door you chose, Monty asks if you want to switch doors - what do you do?

marcopolo
Posts: 267
Joined: Sat Dec 03, 2016 10:22 am

Re: The 4% rule DOES fail but still makes sense!

Post by marcopolo » Tue Nov 14, 2017 2:50 pm

avalpert wrote:
Tue Nov 14, 2017 2:39 pm


So let's say you are on a game show with three doors. Behind one door is a car and behind the other two are goats. You are asked to pick the door with a car and if you are right you win it - you pick door #1. Monty, who knows what’s behind all three doors, opens door #2 showing you a goat. Before opening the door you chose, Monty asks if you want to switch doors - what do you do?
You switch to the other remaining door. It doubles your chance of winning.
I guess you are trying to illustrate the value of prior information? I think that is kind of what i was trying to point out above, but did so less elegantly.
Once in a while you get shown the light, in the strangest of places if you look at it right.

avalpert
Posts: 6197
Joined: Sat Mar 22, 2008 4:58 pm

Re: The 4% rule DOES fail but still makes sense!

Post by avalpert » Tue Nov 14, 2017 2:57 pm

marcopolo wrote:
Tue Nov 14, 2017 2:50 pm
avalpert wrote:
Tue Nov 14, 2017 2:39 pm


So let's say you are on a game show with three doors. Behind one door is a car and behind the other two are goats. You are asked to pick the door with a car and if you are right you win it - you pick door #1. Monty, who knows what’s behind all three doors, opens door #2 showing you a goat. Before opening the door you chose, Monty asks if you want to switch doors - what do you do?
You switch to the other remaining door. It doubles your chance of winning.
I guess you are trying to illustrate the value of prior information? I think that is kind of what i was trying to point out above, but did so less elegantly.
Exactly, and to be fair I stole the elegance from others.

randomguy
Posts: 4890
Joined: Wed Sep 17, 2014 9:00 am

Re: The 4% rule DOES fail but still makes sense!

Post by randomguy » Tue Nov 14, 2017 3:00 pm

Da5id wrote:
Tue Nov 14, 2017 1:39 pm
flyingaway wrote:
Tue Nov 14, 2017 1:33 pm
To ALL:

In the OP example, which was used in several places, some people wanted to show that when person A and person B (Test A and Test B), with the same portfolio and same spending level, which person B, retired in 2018, will fail according to the 4% rule, while person A, retired in 2017, will succeed.

Note that person A spends money in 2017. Person B does not spend money in 2017. So in 2018, person B has more money than person A, but the 4% rule says that person B will fail, however, predicted one year ago person A will succeed.

My "analysis" says that they are comparing different things.
The 4% rule has always been simplistic and has absurd seeming results. That is one of the criticisms. It is just a "rule of thumb", not a retirement spending plan. The inconsistency is often seen in the following

Person A retires in year X with 800K. They can spend 32K/year, adjusting up for inflation, per the 4% rule.
Person B retires in year X+1. They have 1000K, they can spend 40K/year, adjusting for inflation. But the stock market boomed, and person A now has 1000K too. But they can only spend 32K (inflation adjusted) if following the 4% "rule", despite having 1 fewer year to fund than person B.

It is perverse if followed literally as a spending plan, almost all would agree. But as an estimate of "can I retire" 25-33x expenditures seems like an OK guestimate.
The 4% rule doesn't say some WILL fail. It says they MAY fail. As you add info, you can get more accurate results. The 4% rule math says that the SWR for the first person is in the 32k/yr (1966) to 80k/yr (1982) range. As you add more info you can start discarding various outcomes and be more accurate. There are numerous papers out there that talk about various schemes for doing that.

People have tried throwing things in like valuations to get better estimates. I am not convinced that the results we get out are much outside the margins of error.

FactualFran
Posts: 447
Joined: Sat Feb 21, 2015 2:29 pm

Re: The 4% rule DOES fail but still makes sense!

Post by FactualFran » Tue Nov 14, 2017 3:10 pm

flyingaway wrote:
Tue Nov 14, 2017 9:23 am
First, many MMM people's claim is: the 4% rule never fails. They did not understand the statistics. The original 4% rule calculation had a success rate of about 96%, so it did fail in about 4% of the tested cases.
I don't know who the MMM people are, but the 4% inital withdrawal rate idea originated in the 1994 article Determining Withdrawal Fates Using Historical Data by William P. Bengen. According to it
Assuming a minimum requirement of 30 years of portfolio longevity, a first year withdrawal of 4 percent [Figure l(b)], followed by inflation-adjusted withdrawals in subsequent years, should be safe. In no past case has it caused a portfolio to be exhausted before 33 years, and in most cases it will lead to portfolio lives of 50 years or longer.
Note that the success rate was 100%, not 96%. Those who state otherwise do not understand what Bengen wrote.

User avatar
TheTimeLord
Posts: 4715
Joined: Fri Jul 26, 2013 2:05 pm

Re: The 4% rule DOES fail but still makes sense!

Post by TheTimeLord » Tue Nov 14, 2017 3:25 pm

Still waiting for the first Boglehead thread to claim an negative withdrawal rate is unsafe because it causes an increase in taxes.

Distracting animated gif removed by Mod Mel. (They're not allowed.).
Run, You Clever Boy!

dcabler
Posts: 293
Joined: Wed Feb 19, 2014 11:30 am

Re: The 4% rule DOES fail but still makes sense!

Post by dcabler » Tue Nov 14, 2017 3:35 pm

TheTimeLord wrote:
Tue Nov 14, 2017 3:25 pm
Still waiting for the first Boglehead thread to claim an negative withdrawal rate is unsafe because it causes an increase in taxes.

Distracting animated gif removed by Mod Mel. (They're not allowed.)
Bwahahaha! Now can we talk about tilting again?

Da5id
Posts: 1519
Joined: Fri Feb 26, 2016 8:20 am

Re: The 4% rule DOES fail but still makes sense!

Post by Da5id » Tue Nov 14, 2017 3:39 pm

TheTimeLord wrote:
Tue Nov 14, 2017 3:25 pm
Still waiting for the first Boglehead thread to claim an negative withdrawal rate is unsafe because it causes an increase in taxes.

Distracting animated gif removed by Mod Mel. (They're not allowed.)
If it weren't for beating dead horses (4% rule, international or not, home ownership, etc) what would there be to talk about :) Truly nothing is new under the sun.

User avatar
CaliJim
Posts: 2727
Joined: Sun Feb 28, 2010 8:47 pm
Location: San Fran Bay Area

Re: The 4% rule DOES fail but still makes sense!

Post by CaliJim » Tue Nov 14, 2017 3:42 pm

The reason the probabilities for success change in the two scenarios is that in the first 30 year simulation, a 20% drop in year one has only a fractional probability of happening.

In the second 29 year scenario, that 20% chance of a drop in year one has now gone to 100%.
-calijim- | | For more info, click this

flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

Re: The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 3:43 pm

marcopolo wrote:
Tue Nov 14, 2017 1:43 pm
flyingaway wrote:
Tue Nov 14, 2017 1:33 pm
To ALL:

In the OP example, which was used in several places, some people wanted to show that when person A and person B (Test A and Test B), with the same portfolio and same spending level, which person B, retired in 2018, will fail according to the 4% rule, while person A, retired in 2017, will succeed.

Note that person A spends money in 2017. Person B does not spend money in 2017. So in 2018, person B has more money than person A, but the 4% rule says that person B will fail, however, predicted one year ago person A will succeed.

My "analysis" says that they are comparing different things.
I don't think the 4% rule says that at all. Person A and B are essentially the same person, and will have the same probability of success (based on historical data). Person B is actually a little better off since they did not take withdrawals in year one.

Based on historical data, if Person B starts at 4% WR absed on current portfolio value ($80k/yr, instead of $100k/yr like person A), they will likely end up as more than half the scenarios where the ending portfolio size is quite a bit bigger than then starting value.

Now whether the 4% rule will hold true in the future is anyone's guess...
If person B withdraws 80k/year, this person will have a worse life than person A, who withdraws 100k/year. Not a fair comparison.

flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

Re: The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 3:46 pm

CaliJim wrote:
Tue Nov 14, 2017 3:42 pm
The reason the probabilities for success change in the two scenarios is that in the first 30 year simulation, a 20% drop in year one has only a fractional probability of happening.

In the second 29 year scenario, that 20% chance of a drop in year one has now gone to 100%.
In fact, for the portfolio of 25X to support 30 years of retirement, the average yearly return must be positive, possibly around 2.5% without inflation adjustment.

User avatar
TheTimeLord
Posts: 4715
Joined: Fri Jul 26, 2013 2:05 pm

Re: The 4% rule DOES fail but still makes sense!

Post by TheTimeLord » Tue Nov 14, 2017 3:58 pm

Da5id wrote:
Tue Nov 14, 2017 3:39 pm
TheTimeLord wrote:
Tue Nov 14, 2017 3:25 pm
Still waiting for the first Boglehead thread to claim an negative withdrawal rate is unsafe because it causes an increase in taxes.

Distracting animated gif removed by Mod Mel. (They're not allowed.)

If it weren't for beating dead horses (4% rule, international or not, home ownership, etc) what would there be to talk about :) Truly nothing is new under the sun.
Of course you are correct. It ain't rocket science or time travel.
Run, You Clever Boy!

randomguy
Posts: 4890
Joined: Wed Sep 17, 2014 9:00 am

Re: The 4% rule DOES fail but still makes sense!

Post by randomguy » Tue Nov 14, 2017 4:02 pm

flyingaway wrote:
Tue Nov 14, 2017 3:46 pm
CaliJim wrote:
Tue Nov 14, 2017 3:42 pm
The reason the probabilities for success change in the two scenarios is that in the first 30 year simulation, a 20% drop in year one has only a fractional probability of happening.

In the second 29 year scenario, that 20% chance of a drop in year one has now gone to 100%.
In fact, for the portfolio of 25X to support 30 years of retirement, the average yearly return must be positive, possibly around 2.5% without inflation adjustment.
What about if we have 2% deflation for 30 years. Should be able to make thing work with a negative nominal return:)

Note the portfolio is expected to survive the 20% drop so the fact the odds go to 100% isn't the big issue. What the 4% rule and the like don't do is factor in the changes of odds of having a 20% drop followed by another 20% drop and so on. The valuation argument simplifies to something like after a 20% drop, the SWR goes from 4% to 5% because of higher expected returns. But these type of estimates are always pretty bad. Go back and look at the 10 year stock market predications for the past 20 years and see how accurate they have been.

flyingaway
Posts: 1385
Joined: Fri Jan 17, 2014 10:19 am

Re: The 4% rule DOES fail but still makes sense!

Post by flyingaway » Tue Nov 14, 2017 8:34 pm

randomguy wrote:
Tue Nov 14, 2017 4:02 pm
flyingaway wrote:
Tue Nov 14, 2017 3:46 pm
CaliJim wrote:
Tue Nov 14, 2017 3:42 pm
The reason the probabilities for success change in the two scenarios is that in the first 30 year simulation, a 20% drop in year one has only a fractional probability of happening.

In the second 29 year scenario, that 20% chance of a drop in year one has now gone to 100%.
In fact, for the portfolio of 25X to support 30 years of retirement, the average yearly return must be positive, possibly around 2.5% without inflation adjustment.
What about if we have 2% deflation for 30 years. Should be able to make thing work with a negative nominal return:)

Note the portfolio is expected to survive the 20% drop so the fact the odds go to 100% isn't the big issue. What the 4% rule and the like don't do is factor in the changes of odds of having a 20% drop followed by another 20% drop and so on. The valuation argument simplifies to something like after a 20% drop, the SWR goes from 4% to 5% because of higher expected returns. But these type of estimates are always pretty bad. Go back and look at the 10 year stock market predications for the past 20 years and see how accurate they have been.
I don't understand why the SWR goes from 4% to 5% after a 20% drop, does the portfolio becomes smaller and thus SWR becomes smaller also?

avalpert
Posts: 6197
Joined: Sat Mar 22, 2008 4:58 pm

Re: The 4% rule DOES fail but still makes sense!

Post by avalpert » Tue Nov 14, 2017 8:39 pm

flyingaway wrote:
Tue Nov 14, 2017 8:34 pm
randomguy wrote:
Tue Nov 14, 2017 4:02 pm
flyingaway wrote:
Tue Nov 14, 2017 3:46 pm
CaliJim wrote:
Tue Nov 14, 2017 3:42 pm
The reason the probabilities for success change in the two scenarios is that in the first 30 year simulation, a 20% drop in year one has only a fractional probability of happening.

In the second 29 year scenario, that 20% chance of a drop in year one has now gone to 100%.
In fact, for the portfolio of 25X to support 30 years of retirement, the average yearly return must be positive, possibly around 2.5% without inflation adjustment.
What about if we have 2% deflation for 30 years. Should be able to make thing work with a negative nominal return:)

Note the portfolio is expected to survive the 20% drop so the fact the odds go to 100% isn't the big issue. What the 4% rule and the like don't do is factor in the changes of odds of having a 20% drop followed by another 20% drop and so on. The valuation argument simplifies to something like after a 20% drop, the SWR goes from 4% to 5% because of higher expected returns. But these type of estimates are always pretty bad. Go back and look at the 10 year stock market predications for the past 20 years and see how accurate they have been.
I don't understand why the SWR goes from 4% to 5% after a 20% drop, does the portfolio becomes smaller and thus SWR becomes smaller also?
The amount of a withdrawal that was safe for 30 years if taken last year is presumable safe for 29 years if taken this year - since the portfolio declined 20%, whereas last year the withdrawal amount was 4 out $100 this year it is the same amount, 4 but out of only $80 - or 5%.

User avatar
Raybo
Posts: 1541
Joined: Tue Feb 20, 2007 11:02 am
Location: San Francisco
Contact:

Re: The 4% rule DOES fail but still makes sense!

Post by Raybo » Tue Nov 14, 2017 9:25 pm

There are several periods in US market history where there was a decline of 20% or more. Instead of using made up examples, why not use the real returns from real markets to see what happens after a 20% drop.

Here is a list of drops: https://www.yardeni.com/pub/sp500corrbear.pdf
No matter how long the hill, if you keep pedaling you'll eventually get up to the top.

Atgard
Posts: 260
Joined: Wed Apr 09, 2014 2:02 pm

Re: The 4% rule DOES fail but still makes sense!

Post by Atgard » Tue Nov 14, 2017 10:03 pm

It sounds like people are really arguing (with the 2 examples) that if you have a 20% drop in Year 1, you may be in trouble. Yeah, that is really the big risk, a big drop right after retirement, although in most or all cases in the past, markets have rebounded enough that your original 4% withdrawal rate was still sustainable! (But of course, this time it could be different! So nothing forward-looking can be 100%.)

But I think for most of us, if we face a 20% drop in Year 1, we may re-calculate 4% based on the new figure or go back to work part time or cut back on spending for a bit. 4% is still a very very solid plan, just don't put all your spending on irrevocable auto-pilot the day after you retire or shut yourself in a cabin and throw your statements in the fire. But working an extra 2, 5, or 10 years because you're scared of that chance that (a) the market crashes 20% immediately after you retire, and (b) unlike in all past cases, it doesn't rebound enough to support 4%... those don't seem like good odds to me. Odds are better you die of a work-stress-related ailment in those extra working years.

wrongfunds
Posts: 837
Joined: Tue Dec 21, 2010 3:55 pm

Re: The 4% rule DOES fail but still makes sense!

Post by wrongfunds » Wed Nov 15, 2017 9:32 am

I still don't get why you guys are stuck up on annual numbers. Given the available data and the unlimited compute power, run the various scenario with monthly or twice weekly or quarterly payout. I do see an issue that the real inflation numbers are only computed annually but we can probably make some changes to the computation to account for that.

What you will most likely find is that the calculations do NOT converge. This is only my hunch and I have not done the necessary calculations to prove it. However, since the underlying numbers are not based upon natural aka physical or chemical world, I see no reason one should expect convergence. It is a mathematical concept and applies to continuous functions.

My claim is that if you were to run exact same study on Apr 1 annual date vs Jan 1 as annual date, the resultant number will be way different than the 4%. I expect the number would be off by 0.5% or more.

If you don't understand that, you will be always get dazzled by some academics who will spin bunch of numerical analysis and take them as gospel vs taking them as just a rough guideline.

Post Reply