Author Topic: What is a reasonable WR to retire on ?  (Read 7319 times)

steveo

  • Handlebar Stache
  • *****
  • Posts: 1928
What is a reasonable WR to retire on ?
« on: March 09, 2015, 12:35:20 AM »
There was a thread on here recently regarding a conservative WR to retire on. I'd like to know the flip side - i.e is a 4% WR with a high success rate too safe or is it more realistic to aim for a 5% withdrawal rate.

Edited due to updated figures.
« Last Edit: March 09, 2015, 09:15:54 PM by steveo »

dividendman

  • Handlebar Stache
  • *****
  • Posts: 1901
Re: What is a reasonable WR to retire on ?
« Reply #1 on: March 09, 2015, 12:46:39 AM »
It all depends on your flexibility, in my opinion. If you have a bunch of ways to buffer (cut back spending or earn income) when bad shit goes down you can probably get away with > 4%. If you have no flexibility then 4% may be too aggressive.

deborah

  • Senior Mustachian
  • ********
  • Posts: 15967
  • Age: 14
  • Location: Australia or another awesome area
Re: What is a reasonable WR to retire on ?
« Reply #2 on: March 09, 2015, 12:47:00 AM »
Depends on your options. I was on a good salary (that I would be unlikely to get again after a few years off), had health issues and so didn't want to count on being able to work in retirement. I went for a lower WR.

It would be quite reasonable for someone who is really young, and would be likely to get just as much as they are currently making later on to have a much higher WR.

It is all dependent upon your perceived risks and how you plan to mitigate them.

steveo

  • Handlebar Stache
  • *****
  • Posts: 1928
Re: What is a reasonable WR to retire on ?
« Reply #3 on: March 09, 2015, 01:06:09 AM »
http://www.mrmoneymustache.com/2012/05/29/how-much-do-i-need-for-retirement/

I get from this article the rate from MMM's perspective is 4%. I think I'd go with that based on my current thinking however I'd be interested in why anyone would deviate from this figure. Are you for instance happy with a 5% WR or a 3% WR ?

deborah

  • Senior Mustachian
  • ********
  • Posts: 15967
  • Age: 14
  • Location: Australia or another awesome area
Re: What is a reasonable WR to retire on ?
« Reply #4 on: March 09, 2015, 01:30:15 AM »
I would be much happier looking at the returns I was currently getting from my investments before retiring. It's all very well to have $1.6 million in investments as someone here may have, but if they were getting very little for it (or nothing if it was hidden in pillowcases under the bed), or it was not available for 10 years... it would not be much use for them to have a 3% WR.

Retire-Canada

  • Walrus Stache
  • *******
  • Posts: 8685
Re: What is a reasonable WR to retire on ?
« Reply #5 on: March 09, 2015, 08:31:07 AM »
It all depends on your flexibility, in my opinion. If you have a bunch of ways to buffer (cut back spending or earn income) when bad shit goes down you can probably get away with > 4%. If you have no flexibility then 4% may be too aggressive.

^^^^^ a FIRE plan is too complex to reduce to a single number.

Save/invest towards 4%.

Come up with a bunch of risk mitigation plans.

Adjust what you do based on annual investment returns.

-- Vik

skyrefuge

  • Handlebar Stache
  • *****
  • Posts: 1015
  • Location: Suburban Chicago, IL
Re: What is a reasonable WR to retire on ?
« Reply #6 on: March 09, 2015, 10:05:35 AM »
or is it more realistic to aim for a 5% withdrawal rate which has a success rate greater than 80%.

Er...Citation Desperately Needed! What math shows a success rate greater than 80% for a 5% WR?

cFIREsim shows only a 70% success rate, using its default parameters (75/25 portfolio, 30 years, 0.18 expenses). FIRECalc, which is inferior to cFIREsim (or any other SWR analysis) because it uses a naive calculation for bond returns, shows 72%.

Additionally, no one I'm aware of shows 100% success for 4%, even for only a 30 year period. cFIREsim shows 93%. And MMM's hand-waving that "it's all basically the same for periods longer than 30 years" isn't very true either. Success drops to 79% for a 50 year period.

None of this is to say that I think a 4% WR is particularly risky. I think an 80% success rate is plenty high, and I think that flexibility and other sorts of income mean that the real odds of success will be even better than those theoretical odds.

But it's tough to combat the doom-and-gloom pessimists if we wreck our credibility by starting with bullshit numbers that aren't actually supported by the math.

sol

  • Walrus Stache
  • *******
  • Posts: 8433
  • Age: 47
  • Location: Pacific Northwest
Re: What is a reasonable WR to retire on ?
« Reply #7 on: March 09, 2015, 10:13:55 AM »
Your chosen WR should also depend on your time horizon.  If you expect to have all of your expenses covered by a pension in 15 years, then you can "safely" increase your WR to 6% and still have a 94% success rate.

ZiziPB

  • Magnum Stache
  • ******
  • Posts: 3417
  • Location: The Other Side
Re: What is a reasonable WR to retire on ?
« Reply #8 on: March 09, 2015, 11:18:54 AM »
http://www.mrmoneymustache.com/2012/05/29/how-much-do-i-need-for-retirement/

I get from this article the rate from MMM's perspective is 4%. I think I'd go with that based on my current thinking however I'd be interested in why anyone would deviate from this figure. Are you for instance happy with a 5% WR or a 3% WR ?

I am planning on 3% because once I'm retired I want to be done with work - I do not want to have a backup plan that involves getting another job.

steveo

  • Handlebar Stache
  • *****
  • Posts: 1928
Re: What is a reasonable WR to retire on ?
« Reply #9 on: March 09, 2015, 02:50:58 PM »
or is it more realistic to aim for a 5% withdrawal rate which has a success rate greater than 80%.

Er...Citation Desperately Needed! What math shows a success rate greater than 80% for a 5% WR?

cFIREsim shows only a 70% success rate, using its default parameters (75/25 portfolio, 30 years, 0.18 expenses). FIRECalc, which is inferior to cFIREsim (or any other SWR analysis) because it uses a naive calculation for bond returns, shows 72%.

Additionally, no one I'm aware of shows 100% success for 4%, even for only a 30 year period. cFIREsim shows 93%. And MMM's hand-waving that "it's all basically the same for periods longer than 30 years" isn't very true either. Success drops to 79% for a 50 year period.

None of this is to say that I think a 4% WR is particularly risky. I think an 80% success rate is plenty high, and I think that flexibility and other sorts of income mean that the real odds of success will be even better than those theoretical odds.

But it's tough to combat the doom-and-gloom pessimists if we wreck our credibility by starting with bullshit numbers that aren't actually supported by the math.

I just put a 4% withdrawal rate into cfiresim with a 60 year retirement and the success rate was 100%. A 5% withdrawal rate had a success rate > 80%.

Thinking this through my personal take is that the 4% rate is pretty safe especially when I expect to receive some inheritance, I will probably decrease my spending and I am not including my house within my asset portfolio.

skyrefuge

  • Handlebar Stache
  • *****
  • Posts: 1015
  • Location: Suburban Chicago, IL
Re: What is a reasonable WR to retire on ?
« Reply #10 on: March 09, 2015, 03:25:19 PM »
I just put a 4% withdrawal rate into cfiresim with a 60 year retirement and the success rate was 100%. A 5% withdrawal rate had a success rate > 80%.

cFIREsim doesn't give randomized results. If we each enter the same settings, we should get the same result. With a 4% WR ($40k Yearly Spending vs. $1M portfolio), 75/25 allocation, 0.18% expenses, Inflation Adjusted Withdrawal Method, and a 60-year period (2015-2075), I get 82.35% success. If I move it to a 100/0 allocation with 0%, it only rises to 88.24% success.

So one of us is doing something wrong. I'm pretty sure it's not me. Help convince me that it is.
« Last Edit: March 09, 2015, 03:36:11 PM by skyrefuge »

Retire-Canada

  • Walrus Stache
  • *******
  • Posts: 8685
Re: What is a reasonable WR to retire on ?
« Reply #11 on: March 09, 2015, 03:29:44 PM »

I just put a 4% withdrawal rate into cfiresim with a 60 year retirement and the success rate was 100%. A 5% withdrawal rate had a success rate > 80%.


Using the default settings and 4% WR + 60yr retirement period you get 82% success with cFIREsim.

Same settings, but 5% WR = 47% success rate.

-- Vik

brooklynguy

  • Handlebar Stache
  • *****
  • Posts: 2204
  • Age: 43
Re: What is a reasonable WR to retire on ?
« Reply #12 on: March 09, 2015, 03:37:25 PM »
The question as stated in the thread title is asking for an objective answer to what should be a subjective determination.  We usually define the "safety" of a withdrawal rate by reference to its historical likelihood of success.  How high do those historical odds need to be in order to consider a WR to be "safe"?  Depends who you ask.

I would say that even at the more aggressive end of the spectrum, most of us are not willing to pull the FIRE trigger until we have a historical success rate somewhere north of 90%, ignoring the usual levels of safety margin.  "Quitting my high-paying job when there's a one-in-ten chance of DEPLETEING MY PORTFOLIO DOWN TO ZERO unless I fall back on one of my back-up plans, and that's only if we make a generous assumption that the (unknowable) future will to some extent resemble the past -- that would be nuts!"

Conversely, as sol likes to point out, using any success rate greater than 50% guarantees that you are most likely working longer than necessary (assuming the future is no worse than the past).  So if it is already more-likely-than-not that I will never again have to return to work if I retire today, based on a history that includes various financial and geopolitical calamities, then why in the world would I voluntarily opt to experience the 100% certainty of continuing to work in exchange for a mere reduction in the odds of maybe having to work in the future (which are already weighted in my favor), especially when I can exercise flexibility in my retirement plan to improve my already-favorable odds of success even further?!

Everyone needs to figure out their own risk tolerance.  I prefer the more optomistic of the two above perspectives, but I don't quite have the cojones to retire on anything close to a 50% success rate myself.

deborah

  • Senior Mustachian
  • ********
  • Posts: 15967
  • Age: 14
  • Location: Australia or another awesome area
Re: What is a reasonable WR to retire on ?
« Reply #13 on: March 09, 2015, 03:39:16 PM »
cFIREsim doesn't give randomized results. If we each enter the same settings, we should get the same result. With a 4% WR ($40k Yearly Spending vs. $1M portfolio), 75/25 allocation, 0.18% expenses, Inflation Adjusted Withdrawal Method, and a 60-year period (2015-2075), I get 82.35% success. If I move it to a 100/0 allocation with 0%, it only rises to 88.24% success.

So one of us is doing something wrong. I'm pretty sure it's not me. Help convince me that it is.
I agree with you but I have a few caveats:

First off, the monte carlo gives random. Secondly, 60 years gives only 15 variations without monte carlo, and I suspect that is too few samples for the length of time you are after to be of any use. Thirdly, many people have something they get at 62 or whatever age, so this is possibly unrealistic.

steveo

  • Handlebar Stache
  • *****
  • Posts: 1928
Re: What is a reasonable WR to retire on ?
« Reply #14 on: March 09, 2015, 08:24:16 PM »
The figures are interesting. I don't know what is going on but I've checked and this is what I get.

I put in retire in 2022 for 60 years and spend 25k per year with 625k asset base with a 72/25 asset allocation. There are no failures out of 78 cycles.

I checked with the figures listed above and changed it to 1 million asset with 40k spending and it was still a 100% success ratio.

desk_jockey

  • CM*MW 2023 Attendees
  • Bristles
  • *
  • Posts: 326
Re: What is a reasonable WR to retire on ?
« Reply #15 on: March 09, 2015, 08:46:05 PM »
Stevo, I think you found a bug.  I re-ran the same numbers that you put in, and got a 0% failure rate as well.   I then ran 2015 retire for 30 years using the same 25K spend / 625K asset / 75&25 split, and got the 93% success rate that I would have roughly estimated.   

If it's failing 7% of the time at 30 years, you can bet it should be failing at a higher rate at 60 years.

Try running it again retiring this year for 60 years.   Maybe we are overlooking a calculation when projecting out years.


skyrefuge

  • Handlebar Stache
  • *****
  • Posts: 1015
  • Location: Suburban Chicago, IL
Re: What is a reasonable WR to retire on ?
« Reply #16 on: March 09, 2015, 08:57:19 PM »
I put in retire in 2022 for 60 years

Ah, there's the problem. You have to use 2015 through 2075. By putting the start date in the future, you're giving that money 7 years to grow before you start spending it. From the documentation: "If this is set into the future, the simulation will not deduct your spending amount from your portfolio until the retirement date".

If you're feeling helpful, you might want to edit your original post, to prevent skimming newbies from getting the idea that 4% has a 100% success rate.

Also, as deborah notes, using a 60 year term can be a bit misleading. Here are the 4%, default, success rates for various terms:

30 years: 93.04%
40 years: 81.90%
50 years: 77.89%
60 years: 82.35%

What? Living 60 more years works better than living 40 or 50 years? No, the problem is that there are only 85 sixty-year periods to analyze, compared to 115 30-year periods. The most-recent 60-year cycle begins in 1955, so it misses the ~10 cycles that started in the 1960s that we already know would have failed. So the real 60-year success rate is probably more like 70% or less.

steveo

  • Handlebar Stache
  • *****
  • Posts: 1928
Re: What is a reasonable WR to retire on ?
« Reply #17 on: March 09, 2015, 09:14:46 PM »
I put in retire in 2022 for 60 years

Ah, there's the problem. You have to use 2015 through 2075. By putting the start date in the future, you're giving that money 7 years to grow before you start spending it. From the documentation: "If this is set into the future, the simulation will not deduct your spending amount from your portfolio until the retirement date".

If you're feeling helpful, you might want to edit your original post, to prevent skimming newbies from getting the idea that 4% has a 100% success rate.

Also, as deborah notes, using a 60 year term can be a bit misleading. Here are the 4%, default, success rates for various terms:

30 years: 93.04%
40 years: 81.90%
50 years: 77.89%
60 years: 82.35%

What? Living 60 more years works better than living 40 or 50 years? No, the problem is that there are only 85 sixty-year periods to analyze, compared to 115 30-year periods. The most-recent 60-year cycle begins in 1955, so it misses the ~10 cycles that started in the 1960s that we already know would have failed. So the real 60-year success rate is probably more like 70% or less.

That makes sense. I can try and amend the previous posts.

Doulos

  • Stubble
  • **
  • Posts: 116
  • Age: 45
Re: What is a reasonable WR to retire on ?
« Reply #18 on: March 09, 2015, 09:21:17 PM »
After looking over the discussion and the FIRE calculator, I have to side with view that we have a calculator failure when trying to do long retirements.
Like Sky here is saying, the calculator just cannot handle long retirements due to lack of historical data. 
Living longer is not making you less and less likely to succeed.  The calculator just cannot handle it.

I want to say that anything over about 30 years looks broken.

Also; My opinion is on the 5% side.  Maybe more.
There is just so much wiggle room in there.  Like the MMM article on it.
  • You can spend less - ie, make yourself a real emergency budget and what Broke spending looks like and run that number.
  • You can get free money - ie, there is the possibility that social security might actual not go broke and send you checks every month.
  • You can get a job - ie, maybe just maybe, you find out you like doing something that makes cash; like writing a blog, or streaming video games.
« Last Edit: March 09, 2015, 09:25:42 PM by Doulos »

skyrefuge

  • Handlebar Stache
  • *****
  • Posts: 1015
  • Location: Suburban Chicago, IL
Re: What is a reasonable WR to retire on ?
« Reply #19 on: March 09, 2015, 09:43:01 PM »
Living longer is not making you less and less likely to succeed.  The calculator just cannot handle it.

I want to say that anything over about 30 years looks broken.

And I want to say that Scarlett Johansson comes to my house every week and does my vacuuming for free because she thinks I'm so awesome.

But unfortunately wanting to say that does not actually make it true!

Living longer can certainly make a given portfolio at a given withdrawal rate less and less likely to succeed.

The calculator is not "broken", you just have to understand how the calculation is designed. While the analysis has its flaws, it's better than anything else we have, and it's performing exactly as intended.

deborah

  • Senior Mustachian
  • ********
  • Posts: 15967
  • Age: 14
  • Location: Australia or another awesome area
Re: What is a reasonable WR to retire on ?
« Reply #20 on: March 09, 2015, 09:51:17 PM »
To get more scenarios you can always use the monte carlo simulation - which uses all the data and generates random scenarios with it.

Doulos

  • Stubble
  • **
  • Posts: 116
  • Age: 45
Re: What is a reasonable WR to retire on ?
« Reply #21 on: March 09, 2015, 11:31:20 PM »
I want to say a better option than trying 60 year scenarios is to run 10 years with additional failure considerations.  The 60 years just starts hurting your sample size instead of giving you better information.  Less years with a minimum balance is going to give you much better "chance of success" information.

A lot of you seem to be looking for a worse case.  You already know the worse case. 
The stock market crashes the year after you quit your job. 
The entire industry you worked in also become obsolete.

There is just no pleasing worse case.

If you want some good pessimistic numbers run with
  • 10 year windows 2015-2025
  • Stash size = ?
  • Spending plan = % of Portfolio (no floor, no ceiling)
  • "Criteria for marking a cycle as "failed" Yearly Withdrawal falls below $". set to whatever your minimum yearly spending is.

That will give you a much better idea of what retired forever looks like than the 60 year windows.  And some much lower success rates too.

Success being based on "not hitting account = zero" is really more of an optimistic view, in which case you should be perfectly happy just using 5 or 6%.

Some will be perfectly happy being optimistic with $600k and 5%.
Others will be sad being pessimistic with $1.2M and 2.5%.

steveo

  • Handlebar Stache
  • *****
  • Posts: 1928
Re: What is a reasonable WR to retire on ?
« Reply #22 on: March 10, 2015, 03:46:14 AM »
After looking over the discussion and the FIRE calculator, I have to side with view that we have a calculator failure when trying to do long retirements.
Like Sky here is saying, the calculator just cannot handle long retirements due to lack of historical data. 
Living longer is not making you less and less likely to succeed.  The calculator just cannot handle it.

I want to say that anything over about 30 years looks broken.

Also; My opinion is on the 5% side.  Maybe more.
There is just so much wiggle room in there.  Like the MMM article on it.
  • You can spend less - ie, make yourself a real emergency budget and what Broke spending looks like and run that number.
  • You can get free money - ie, there is the possibility that social security might actual not go broke and send you checks every month.
  • You can get a job - ie, maybe just maybe, you find out you like doing something that makes cash; like writing a blog, or streaming video games.

There are lots of options I think to cut back on spending however there is also the chance that you have to spend more for health reasons. I'm starting to like the 4% number right now. Its not foolproof but with buffer in your spending its a pretty good figure.