A "Ratcheting" SWR - Kitces

[rayvt]

Right, but as you say (with G-K), you have to be ready to cut spending. I personally would rather approach things with my conservative WR% with the hope that I never have to cut the buying power amount.

Well, as they say, "hope is not a plan".
Something they taught us in one of the classes I took was "The person with the best Plan B is the one most likely to be the winner."

This is one of those interminable topics. We are trying to balance two conflicting things, to maximize the annual withdrawals, to not run out of money, to not have to reduce any annual withdrawal. The problem is we do not know what will happen in the future.
When the portfolio goes down, you don't know if this is just a momentary pothole, or if it is the start of a long downward slope.

If you want to never run out of money, you go with a constant percentage draw --- but the draws will fluctutate wildly. Somebody recently posted a spreadsheet showing a draw of $59K one year and a draw of $48K the next year, in the 2008/2009 period.

As G-K, Kitces, Burns, and others have said, *most* of the large market jumps are smoothed out over the next few years, so abruptly changing your draw is not neccessary. But changing it *will* be neccessary if the jump does not revert. That's the case no matter how you manage the changes--the ultimate reduction is if your account goes to zero, and you have $0 withdrawls forever.

These strategies have a Plan B for the case of portfolio losses: slowly reduce withdrawals.
The "take 3%, period" strategy has no Plan B. The real world imposes its own Plan B -- "...and then you die."

The alternative to possibly making periodic reductions is to take a low SWR. But that is essentially taking a permanent preventative reduction.
Take 3% with no reductions, or take 5% with the possibility of having to reduce to 4% or 3%. Or a low probability of 2%. It sounds to me like a boxer taking a dive to avoid getting punched. Preemptively declaring defeat, and starting off with the initial presumption that you will lose anyway.

If you are okay with 3%, then why not take 5% and then reduce down to 3% if and when it ever becomes neccessary? Because looking at past history, there is less that a 1-in-20 chance that it will be neccessary. But if those odds catch up with you -- well, you have a plan that will let you survive.

We all know that when push comes to shove we *will* reduce our spending in adverse curcumstances. So why not make that an explicit part of the plan? Whether you are taking 5% or 3%, if the market crashes 50% you'll hunker down and avoid all unneccessary spending. So a plan that has you cutting your draw is just explicitly stating what you'll do.

 

[pb4uski]

I seem to recall that we had a similar thread recently about using Firecalc to determine spending consistent with a target success rate and then refreshing it if your portfolio increased as if one was retiring anew which I suspect is similar to the retire again approach being referred to.

 

[ERD50]

Well, as they say, "hope is not a plan"....

Agreed. I only included 'hope' in there, as I'm a 'never say never' guy. And while a 100% HSWR and a long LE plan is very unlikely to require a cut in buying power... never say never. If the future is worse than the worst of the past, even that plan could require an adjustment.

This is one of those interminable topics. We are trying to balance two conflicting things, to maximize the annual withdrawals, to not run out of money, to not have to reduce any annual withdrawal. The problem is we do not know what will happen in the future.

Absolutely.

If you want to never run out of money, you go with a constant percentage draw

Right, but that's just shifting the definition of 'failure', as your WR can possibly drop very low

The alternative to possibly making periodic reductions is to take a low SWR. But that is essentially taking a permanent preventative reduction.

Agree to a point, but if you use some form of the 'retire again and again' plan, you re-evaluate, and spending can likely go up. Sure, people will say they'd rather spend it when they are young, but that danged unknown future is the fly in the ointment.

Take 3% with no reductions, or take 5% with the possibility of having to reduce to 4% or 3%. Or a low probability of 2%. ...

If you are okay with 3%, then why not take 5% and then reduce down to 3% if and when it ever becomes neccessary? Because looking at past history, there is less that a 1-in-20 chance that it will be neccessary. But if those odds catch up with you -- well, you have a plan that will let you survive.

I agree with your analysis. But in my personal case, I'm actually OK with a conservative WR at this point in my life, I'm not sacrificing or scrimping (just normal LBYM I would pretty much do anyhow). So I'd rather increase my confidence factor that I can go for a very long time @ ~ 3%, rather than taking more now, and risking, even a small risk, the future lifestyle.

We all know that when push comes to shove we *will* reduce our spending in adverse curcumstances. So why not make that an explicit part of the plan? Whether you are taking 5% or 3%, if the market crashes 50% you'll hunker down and avoid all unneccessary spending.

I know for a fact (remember 2008?) that I *won't* (matching your emphasis!) do that. I went on spending like I always did. In my mind, if I had to tell my wife that we need to cut back a few years after I retired, I would have 'failed', and should go back to work. A low WR provides that comfort level.

Again, that's not the right answer for everyone, but it is my thinking, and I've been-there-done-that at this point.

I seem to recall that we had a similar thread recently about using Firecalc to determine spending consistent with a target success rate and then refreshing it if your portfolio increased as if one was retiring anew which I suspect is similar to the retire again approach being referred to.

Right, that's all it is. Some calculators have the function built in, and will show the historical spending increases (if any).

-ERD50

 

[rayvt]

Hmmm. Thinking about it a little more, it struck me that it probably doesn't matter much which strategy you use for making variable withdrawls.

First off, there is a 95% chance that you will never need to make a reduction. According to Firecalc.

Only a 5% chance that you'll even need to make a reduction. For that 5% case, you just need to decide how fast to rip the bandaid off--fast or slow. Most of the strategies do it slow, just different details of how to do it slowly.
From that standpoint, it doesn't much matter which of these strategies you use, so you might as well use whichever one is th easiest for you to figure.

83% chance that you'll be able to increase your withdrawals.
12% chance that you'll neither need to reduce nor increase your withdrawals.
The most likely case, by far, is that you'll be able to increase your withdrawals. Doesn't it make sense to have a plan for that as well as the disaster plan?

For other SWR's:
3% SWR: 0% must reduce, 0% flat, 100% to increase. (ERD50)
6% SWR: 50% must reduce, 11% flat, 34% to increase. (Scott Burns)
5% SWR: 28% must reduce, 12% flat, 60% to increase. (Guyton)
4% SWR: 5% must reduce, 12% flat, 83% to increase. (Bengen/Trinity standard)

Clearly, the Burns strategy would require you to have a reduction strategy in place.

=============
My method for figuring the breakdown of the probabilities:
Put your parameters into firecalc, then hit the Investigate tab. Check the "success rate" button and "provide date" box, then click Submit. Download the spreadsheet, then sort by the last column (final portfolio value). Count how many are negative. These are the failures. Count how many are positive and less that the starting amount. These are the ones with level withdrawal. The rest are the where the inital withdrawal could have been larger. The counts I got with the default parameters were 6, 14, and 95, for 115 possible periods.

 

[mathjak107]

while income may stay constant the wild card is the legacy money bucket which can be all over the place depending on sequence risk and your actions. that has ranged from zero at 30 years to many times what you started with . so the adjustments relate more to the legacy bucket then spending changes

 

[Retch The Grate]

The thing I'd worry about with doing the reset your SWR every time you have a gain method is that you are continuously selecting for being at the beginning of a N year withdrawal window and as we know, the scenarios that fail are where the market goes down at the early in the withdrawal window. Doing continuous reset on gains with 3% seems fairly safe since the 3% is historical 100% survival, but personally I'd want to adjust down when a 2008 occurred regardless. Overly paranoid for sure, but...

 

[REWahoo]

I agree, a good article.

One thing I especially liked was his chart illustrating the 30 year starting SWR from 1871 to 1985. This really brings home the fact the 4% rule of thumb is very, very conservative and that a WR of 5+% is usually safe.

Of course that "usually" qualifier is what we're all concerned about...

 

[RockyMtn]

Looks like I'm due for a raise! Good article.

 

[Options]

I agree, a good article.

One thing I especially liked was his chart illustrating the 30 year starting SWR from 1871 to 1985. This really brings home the fact the 4% rule of thumb is very, very conservative and that a WR of 5+% is usually safe.

Of course that "usually" qualifier is what we're all concerned about...

Yes, but I just read Bernstein say that future returns have never looked so bleak in recorded U.S. history, so that qualifier calls for a conservative stance, IMO.

 

[REWahoo]

The nice thing about choosing your SWR is you can create your own definition of "S" - and modify that definition at any time.

At age 68, ten years into retirement, I'm comfortable with loosening the purse strings a bit, even if the "I'll keep making predictions - one of these days I'll be right" crowd says the end is nigh.

 

[Andre1969]

I agree, a good article.

One thing I especially liked was his chart illustrating the 30 year starting SWR from 1871 to 1985. This really brings home the fact the 4% rule of thumb is very, very conservative and that a WR of 5+% is usually safe.

Of course that "usually" qualifier is what we're all concerned about...

A couple weeks ago I ran some scenarios using 12/31/07 as a retirement date. I tried both 3% and 4%. Well, I tried adding 5%, just to compare. This is assuming a starting portfolio of $1M, just to make things simple.

3% wd: 1,163,575 as of 12/31/14
4% wd: 1,045,566
5% wd: 927,557
Simply adjusting for inflation, $1M in 2007 would be roughly $1,145,000 today.

I think I read in another thread in the ER forums that one predictor of a failure cycle in FireCalc is that the first 10 years have a return of less than 2%? I might have some of the details mixed up, though.

Hopefully, 2007 ultimately pans out to be one of those anomaly cycles, like 1929, ~1965, 1973, etc. Moot point I guess, since I didn't retire in 2007. However, 2007 was the end of a good, long run that showed me early retirement really was within my reach.

 

[Options]

And for those of us with income streams coming online at various points in retirement the SWR is a moot, if not virtually useless, point as well.

 

[RunningBum]

And for those of us with income streams coming online at various points in retirement the SWR is a moot, if not virtually useless, point as well.

You could always annuitize the value of those future income streams to give an approximation of a "net worth" to base a SWR off of. There may be better ways to do it, but this is a way to make a SWR strategy useful even in those cases.

 

[REWahoo]

And for those of us with income streams coming online at various points in retirement the SWR is a moot, if not virtually useless, point as well.

Is that why you chose 2.5%?

 

[Options]

No I didn't choose anything. Early on, in an attempt to arrive at an SWR, I did exactly what RunningBum suggested and came up with an overall WR rate. I should delete that from my profile beause I really don't place much credence in that figure. It's even more meaningless to me personally as I spend the way I've always spent in life. Meaning, if things get bad I'll simply reduce spending during down years. I've gone several years without an inflation increase and suffered no loss in quality of life. My spending habits simply changed.

IMO, SWR is overrated anyway, what with Guyten spending rules, possible annuitization in later years, and other strategies to avoid PF depletion.

 

[Independent]

How about, "The greater of 4% of your initial fund or 4% of your current fund"?
That's an easier rule to remember and apply.

As Kitches says, the issue is to come up with some rule-of-the-thumb that tells you how soon and how much you can increase spending in good scenarios.

I think he is being more conservative than he needs to be. The problem is in the word "ratchet", which is one way only.

Suppose I start with $1 million and I feel I really "need" $40k from my portfolio. I keep my first year withdrawal to $40k since I have no downside flexibility.

Now I have some good, early years and I'd like to increase to $45k. Kitce assumes I should only go up if I want to put a new floor on my withdrawals.

I don't think that's necessary. If I originally thought I could live on $40k, it seems the extra $5k is "optional" spending that I can readily turn off if my portfolio goes back down.

That is, if I'm willing to live with variable withdrawals only on the excess over 4%, I can be quicker to increase.

I tried back-testing this rule using the standard FireCalc assumptions. It does not add any new failure years. The only "close" year that it impacts is 1964, when this rule converts an ending balance of 5.9% of original portfolio into an ending balance of 4.6% of original portfolio.

I'm sure if I ran 10,000 randomly chosen scenarios, I'd find a few that this rule impacts. But this backtesting says they would be rare.

 

[big-papa]

How about, "The greater of 4% of your initial fund or 4% of your current fund"?
That's an easier rule to remember and apply.

As Kitches says, the issue is to come up with some rule-of-the-thumb that tells you how soon and how much you can increase spending in good scenarios.

I think he is being more conservative than he needs to be. The problem is in the word "ratchet", which is one way only.

Suppose I start with $1 million and I feel I really "need" $40k from my portfolio. I keep my first year withdrawal to $40k since I have no downside flexibility.

Now I have some good, early years and I'd like to increase to $45k. Kitce assumes I should only go up if I want to put a new floor on my withdrawals.

I don't think that's necessary. If I originally thought I could live on $40k, it seems the extra $5k is "optional" spending that I can readily turn off if my portfolio goes back down.

That is, if I'm willing to live with variable withdrawals only on the excess over 4%, I can be quicker to increase.

I tried back-testing this rule using the standard FireCalc assumptions. It does not add any new failure years. The only "close" year that it impacts is 1964, when this rule converts an ending balance of 5.9% of original portfolio into an ending balance of 4.6% of original portfolio.

I'm sure if I ran 10,000 randomly chosen scenarios, I'd find a few that this rule impacts. But this backtesting says they would be rare.

This seems similar to retire again and again found at: New withdrawal method: Retire Again & Again

 

[Independent]

This seems similar to retire again and again found at: New withdrawal method: Retire Again & Again

I read the link, but I'm not 100% sure I know the rule he wants to use.

It appears to me he has a ratchet system - once you go up, you never come down. That introduces additional risk, so he compensates by lowering the initial SWR.

Similarly, Kitces has a ratchet system. He compensates for the additional risk by being very cautious about when and how much to increase.

In my suggestion, the additional amount over the original SWR is variable. If you do well and later poorly, you drop back, potentially all the way to the original SWR. I claim you can do this because your initial SWR covered all your "basic" spending, and the additional is only bonus money that gets spent on fun-to-have stuff.

It looks like my "4% of greater of original portfolio or current portfolio" lets you start at a higher number than RA&A, and increases your spending more quickly than Kitces, if you're open to going back to the original SWR if necessary.

 

[walkinwood]

....

In my suggestion, the additional amount over the original SWR is variable. If you do well and later poorly, you drop back, potentially all the way to the original SWR. I claim you can do this because your initial SWR covered all your "basic" spending, and the additional is only bonus money that gets spent on fun-to-have stuff.

Inflation would eat into the purchasing power of your original 4%. Unless you meant an inflation adjusted value.

 

[Independent]

Inflation would eat into the purchasing power of your original 4%. Unless you meant an inflation adjusted value.

Yes, my comments assume the regular, inflation adjusted SWR.

I should have said "The greater of 4% of your inflation-adjusted initial fund or 4% of your current fund"

 

[audreyh1]

You could always annuitize the value of those future income streams to give an approximation of a "net worth" to base a SWR off of. There may be better ways to do it, but this is a way to make a SWR strategy useful even in those cases.

Generally - SWR is applied to whatever retirement investments exist above pensions, annuities, SS, etc.

No need to model income streams to derive a "net worth". You simply take income needed minus income streams, and compare that to your additional investments to see what the withdrawal rate is, and whether it might be "safe".

If all inflation-adjusted annual expenses are covered by income streams, then the SWR is indeed moot. No point in modeling anything. You can do whatever you want with any additional retirement investments without worrying about safety.

 

[big-papa]

I read the link, but I'm not 100% sure I know the rule he wants to use.

It appears to me he has a ratchet system - once you go up, you never come down. That introduces additional risk, so he compensates by lowering the initial SWR.

Similarly, Kitces has a ratchet system. He compensates for the additional risk by being very cautious about when and how much to increase.

In my suggestion, the additional amount over the original SWR is variable. If you do well and later poorly, you drop back, potentially all the way to the original SWR. I claim you can do this because your initial SWR covered all your "basic" spending, and the additional is only bonus money that gets spent on fun-to-have stuff.

It looks like my "4% of greater of original portfolio or current portfolio" lets you start at a higher number than RA&A, and increases your spending more quickly than Kitces, if you're open to going back to the original SWR if necessary.

Another one that's similar to this is Gummy's "Sensible Withdrawals". You use whatever SWR makes sense, but in good years you get a bonus. A good year is defined this way: if, after the year's withdrawal, your portfolio is greater than it was a year ago (after inflation) then take a % of that growth as a bonus. He suggests 3% SWR, but I think it would work with 4% as well. This one has been on my "top 5" list as a variable method as it tends to produce more bonuses early in retirement than later.

 

[big-papa]

Oops forgot to post the link: sensible withdrawals

 

[Options]

Generally - SWR is applied to whatever retirement investments exist above pensions, annuities, SS, etc.

No need to model income streams to derive a "net worth". You simply take income needed minus income streams, and compare that to your additional investments to see what the withdrawal rate is, and whether it might be "safe".

If all inflation-adjusted annual expenses are covered by income streams, then the SWR is indeed moot. No point in modeling anything. You can do whatever you want with any additional retirement investments without worrying about safety.

Except when income streams come online at various points, then the whole SWR exercise is useless.

 

[clifp]

Yes, it is interesting.

It reminds me of a plan discussed from time to time, the "Retire Again and Again" plan.

https://www.bogleheads.org/forum/viewtopic.php?t=121138#p1771077


In its simplest form, it is more extreme, and ratchets up spending anytime your portfolio buying power increases. Mathematically, sticking to the historical data-set, it has to work (easier to see if you start with a historically 100% safe withdraw rate). And it answers that paradox about two people retiring on different dates - why can't the earlier retiree increase their withdraws to match the new retiree (if the market has gone up enough to cover withdraws and inflation)? The answer is "they can" (historically).

I haven't looked in a while, but I'm pretty sure it provides the most 'efficient' withdraws - minimizing the portfolio end value, and maximizing withdraws, across all historical paths.

Of course, when taken literally, this is getting into data-mining territory, but the concept is sound and worth thinking about.

-ERD50

Reading this interesting paper did remind me a lot of the twins paradox. Especially because if your assets have increased year over year, your life expectancy has also decreased. So if 62 year with 1 million dollars in assets and 4% (40K) withdrawal who find herself with 1.1 million at age 63 logically should be more likely to be able to withdraw 4% (44k) without running out of money in her life than a 62 year old taking 4%.

I suspect that 50% figure before adjusting the withdrawal is almost certainly much too conservative.

The more I think about withdrawals and observe how real world retirees react, the less important I think your initial starting portfolio should be on your current withdrawal.

I am curious how many people who have been retired lets 5+ actually based their current withdrawal on their initial portfolio? Heck I don't even have records of what my portfolio was until 2 years after I retired.