Bill Bengen's 4% WR is not safe in a low yield world

[StuckinCT]

The 4% rules still works in a 6.5% return and 2% inflation world. That's is the return on a 60/40 portfolio assuming Blackrock's and Vanguard's expected returns for the next decade. The key is to remain invested through business cylcles with appropriate exposure to the various asset classes, rebalancing etc, but mostly to stay invested.

 

[Lawrence of Suburbia]

I'm planning on a 3% constant withdrawal rate; whaddya think? Too risky given what we're likely facing going forward?

 

[mathjak107]

I'm wondering about this failed in the first 15 years statistic. What is the definition of failure? Because in FIRECALC for 4% from a 50/50 portfolio starting in 1966, the portfolio did not run out of money until around 1989 which was after 24 years.

Maybe the definition for failure was a different one - withdrawal exceeding some high percent of current portfolio?

Or are you running some Montecarlo simulations?

failure means the money ran out by 30 years but the fate was sealed in the first 15 .to much had to be spent down the first 15 in every case so even the greatest bull markets later on could not save the group

 

[mathjak107]

The 4% rules still works in a 6.5% return and 2% inflation world. That's is the return on a 60/40 portfolio assuming Blackrock's and Vanguard's expected returns for the next decade. The key is to remain invested through business cylcles with appropriate exposure to the various asset classes, rebalancing etc, but mostly to stay invested.

sequence of those returns is what matters not the average of the returns .

the order that average comes in at determines if you survive or not .

there can be as much as a 15 year difference in how long the money lasts just taking the same average and playing with the sequence that average comes in .

moishe milevsky in his now famous article retirement ruin and the sequence of returns demonstrated how average returns mean nothing and showed how the money ran out 15 years sooner or later with the same draw and same average return .


these simple reverse amortization calculators many folks use to determine how much they can draw are awful .

amortization calculators never have you spending down in a down year. every year is subject to the same positive average return and negative real return years don't exist . .

the 17-year period (1987-2003), the S&P 500's average return was 13.47%. It doesn't make any difference if we look at the returns from 1987 to 2003 or from 2003 to 1987.

But when taking withdrawals, the sequence of returns makes all the difference. The same initial capital, the same withdrawal amount, the same returns - but a different sequence produces dramatically different results.

if we leave the rate of inflation growth with the principal and spend everything else .

For a $100,000 portfolio adjusted for inflation over those 17 years, the difference is a remaining balance of $76,629 to a deficit of $187,606. depending on the order those gains and losses came in .

never ever use a calculator that simply asks you to fill in a projected average return and inflation rate .

 

[audreyh1]

failure means the money ran out by 30 years but the fate was sealed in the first 15 .to much had to be spent down the first 15 in every case so even the greatest bull markets later on could not save the group

But what was the criteria that showed a "point of no return" had been reached? How does the retiree know by 15 years?

 

[mathjak107]

less than a 2% real return as an average the first `15 years is dangerous . all the failures were less than 2% the first 15 years .

if you see 5-10 years in you are running below that a red flag should go up that you may have to cut the draw

 

[audreyh1]

I thought 1% real return was still enough to make it 30 years.

 

[StuckinCT]

sequence of those returns is what matters not the average of the returns .

the order that average comes in at determines if you survive or not .

there can be as much as a 15 year difference in how long the money lasts just taking the same average and playing with the sequence that average comes in .

moishe milevsky in his now famous article retirement ruin and the sequence of returns demonstrated how average returns mean nothing and showed how the money ran out 15 years sooner or later with the same draw and same average return .


these simple reverse amortization calculators many folks use to determine how much they can draw are awful .

amortization calculators never have you spending down in a down year. every year is subject to the same positive average return and negative real return years don't exist . .

the 17-year period (1987-2003), the S&P 500's average return was 13.47%. It doesn't make any difference if we look at the returns from 1987 to 2003 or from 2003 to 1987.

But when taking withdrawals, the sequence of returns makes all the difference. The same initial capital, the same withdrawal amount, the same returns - but a different sequence produces dramatically different results.

if we leave the rate of inflation growth with the principal and spend everything else .

For a $100,000 portfolio adjusted for inflation over those 17 years, the difference is a remaining balance of $76,629 to a deficit of $187,606. depending on the order those gains and losses came in .

never ever use a calculator that simply asks you to fill in a projected average return and inflation rate .

The sequence of returns matters only when you are withdrawing principal rather than income from a portfolio. I actually disagree on calculators with return/inflation expectations. Today we should be modeling a lower than average return, say 6.5% versus 7.5% and 2% inflation versus 3% inflation. Equity risk premiums and other measures of future return expectations matter and modeling what happened in boom and bust cycles a 100 years ago though valid, may be less applicable.

 

[mathjak107]

average returns mean little when spending down . there are times you will always be drawing principal because of negative real returns .

high valuations can mean lower returns going forwrard for a while but at the end of the day if spending down it is the sequence of those returns that rule .


don't forget sequence of returns even holds for bonds and cash too. as inflation can make even what appears decent returns in to negative real returns which do the same damage.

your sequence of returns depends on markets ,rates and inflation outcomes combined .

even cd's have had sequence risk since almost 40% of the time they had negative real returns after inflation and taxes so spending at a loss is always a possibility .

in fact 1965/1966 failed because inflation set the failure up for the group. in the first 15 years . .

 

[mathjak107]

I thought 1% real return was still enough to make it 30 years.

kitces said it can but you may have a buck left in year 31 . so to provide a real world safety net you really should strive for 2% real returns. 1% is only verified to 30 years and no longer nor does it provide for some excessive over budget emergency and unexpected spending in a year .

cutting things to close may not be the best planning even though the math would hold for a 30 year time frame .

life expectancy statistics play a part too .

the fact that statistically many of us will not last 30 years takes even a 90% success rate and boosts it way up , and that is what blanchett left off the equation .

if you are playing with statistical probability you have to combine in the statistical probability for life expectancy's role .


that makes 4% draws even more conservative and improves success rate even more than the results from something like firecalc or fidelity since statistically less years will be needed to be supported .

the problem is we just don't know who of us will live and die . but if you are looking at statistical success rates who does not matter .

 

[marko]

90% of every rolling 30 year period since 1926 left you with more than you started with 30 years earlier . it left you with more than 2x what you started with 67% of the time and 1/2 the time tripled ----kitces

Ok. Good data.
But I suspect a 5% W/D might prove disastrous. Any data on that scenario?

 

[exnavynuke]

Ok. Good data.
But I suspect a 5% W/D might prove disastrous. Any data on that scenario?

If I put in a 50K spend rate, a $1M portfolio (5%), no SS or other income, an age of 56 using Bernicke's Reality Retirement Plan, and a 40 year retirement, Firecals shows a 100% chance of success.

Here is how your portfolio would have fared in each of the 106 cycles. The lowest and highest portfolio balance at the end of your retirement was $189,066 to $13,550,680, with an average at the end of $3,331,093. (Note: this is looking at all the possible periods; values are in terms of the dollars as of the beginning of the retirement period for each cycle.)

If I drop the age to 50 (to increase overall spending under Bernicke's plan), Firecalc still gives it a 90.6% chance of success.

 

[marko]

If I put in a 50K spend rate, a $1M portfolio (5%), no SS or other income, an age of 56 using Bernicke's Reality Retirement Plan, and a 40 year retirement, Firecals shows a 100% chance of success.



If I drop the age to 50 (to increase overall spending under Bernicke's plan), Firecalc still gives it a 90.6% chance of success.

This is all good stuff but now I'm conflicted with the 'experts' saying that we all need to drop our W/D to 3% and 2.5% is even better.

Doesn't matter to me personally, but as an academic exercise it gives me more questions than answers.

OTOH I ran $1M with a 5% spend rate, no other income for 30 years, Bernicke Plan and got this:
FIRECalc looked at the 116 possible 30 year periods in the available data, starting with a portfolio of $1,000,000 and spending your specified amounts each year thereafter.
Here is how your portfolio would have fared in each of the 116 cycles. The lowest and highest portfolio balance at the end of your retirement was $-715,635 to $5,382,755, with an average at the end of $1,643,383. (Note: this is looking at all the possible periods; values are in terms of the dollars as of the beginning of the retirement period for each cycle.)
For our purposes, failure means the portfolio was depleted before the end of the 30 years. FIRECalc found that 8 cycles failed, for a success rate of 93.1%.

 

[exnavynuke]

This is all good stuff but now I'm conflicted with the 'experts' saying that we all need to drop our W/D to 3% and 2.5% is even better.

Doesn't matter to me personally, but as an academic exercise it gives me more questions than answers.

OTOH I ran $1M with a 5% spend rate, no other income for 30 years, Bernicke Plan and got this:
FIRECalc looked at the 116 possible 30 year periods in the available data, starting with a portfolio of $1,000,000 and spending your specified amounts each year thereafter.
Here is how your portfolio would have fared in each of the 116 cycles. The lowest and highest portfolio balance at the end of your retirement was $-715,635 to $5,382,755, with an average at the end of $1,643,383. (Note: this is looking at all the possible periods; values are in terms of the dollars as of the beginning of the retirement period for each cycle.)
For our purposes, failure means the portfolio was depleted before the end of the 30 years. FIRECalc found that 8 cycles failed, for a success rate of 93.1%.

Looks like the results for a 48 year old on a 30 year retirement. At 54 the FIRECalc results go to 100% for a 5% withdrawal of a $1M portfolio under Bernicke's plan (99% at 53) for a 30 year plan. The older you start, the better the chance of success generally.

If you use the "constant spending" then the numbers are a lot worse than they are under the Bernicke plan. For constant spending on a 30 year retirement, you need to get under 3.6% withdrawal rate before FIRECalc gives you a 100% success rate. So changing spending assumption, years in retirement, future return assumptions (i.e. FIRECalc doesn't calculate for any scenario worse than history has shown) etc can all lead to more conservative recommendations for withdrawal rates.

 

[GravitySucks]

90% of every rolling 30 year period since 1926 left you with more than you started with 30 years earlier . it left you with more than 2x what you started with 67% of the time and 1/2 the time tripled ----kitces
Thanks Mathjack.
Ive been looking for this in a text form.
I'll assume this is a terrible period and that in 30 years the nut will only be 150% of it's starting value.

 

[marko]

Looks like the results for a 48 year old on a 30 year retirement.

Ahhhh! Yes, I missed that little box for 'age'. Thanks.

Mods: Not sure who does this, but when I clicked on the link: " See his research for more details." re: Bernicke model, the page did not have anything relevant.

 

[samclem]

Just the normal reminder: Taking our historical returns too seriously will lead to a higher expected "precision" in our estimates than is warranted. Yes, history is all we have, but if we are using the modern US data set, let's not forget we're already selecting the very best results out of about 146 countries--we're already cherry-picking. Maybe the horse will continue to perform in the same relative way--I hope it does. But maybe it will not.

 

[ziggy29]

May be true in a prolonged now interest rate environment. That said, we've been discussing this idea since about 2009 when the Fed basically started slashing interest rates to zero...

 

[Ed_The_Gypsy]

I'm planning on a 3% constant withdrawal rate; whaddya think? Too risky given what we're likely facing going forward?

I have not started systematic w/d yet, but targeting 3% of assets each Dec w/o adjustments (maybe zip; depends on how deep the coming dip is)--flexible withdrawals, I think the term is--for the first few years. I am looking at 25 years max. Nothing for de kinder planned. All planning for surviving frau.

At these valuations, I fear rough seas in the near future ("In zee near" as my old Kraut engineering comrade used to say).

 

[Ed_The_Gypsy]

I should mention that after wasting an amazing amount of time and $, I have had a revelation that I cannot do any better than VWELX, and it doesn't take any work.

 

[Midpack]

I should mention that after wasting an amazing amount of time and $, I have had a revelation that I cannot do any better than VWELX, and it doesn't take any work.

I noted the same a while back. The only benefit I'm getting from my broader fund selection is placement for taxes, but that's not as important now as it was when I was accumulating. And if I tried to switch now, the CG losses would be painful. But my portfolio will slowly get simpler in the decades ahead.