when to reset SWR basis

[REWahoo]

... or as REWahoo said, it's why FIRECalc says 4% under those inputs, and not 5%.

Thanks for the citation but the credit goes to pb4uski:

.... IOW, that is why it is 4% rather than 5% or more.

 

[SecondCor521]

That makes sense to me... while I haven't seen the data I suspect that the scenarios that drive the worst case scenarios are preceeded by a strong run... which implicitly addresses Running Bum's concern.... IOW, that is why it is 4% rather than 5% or more.

I wondered about this thought of yours.

I think it is based on a tendency to believe in reversion to the mean and that the longer a period of time we are above or below the mean, the "stronger" the tendency to revert to the mean.

But these are two different thoughts. I think the first is true and the second is not.

In other words, if I flip a fair coin 1000 times, it is true that I'm likely to end up with about (but not exactly) 500 heads and 500 tails. But the coin doesn't have a memory, so if I look at my last X flips and see:

HTHHTHHHTHHHHHHHHHH

it isn't any more likely that then next flip will be tails.

Similarly, I know of no data(*) that shows any correlation between the length or strength of bull markets and the length or depth of bear markets. They both happen in varying strengths and durations.

And of course it is fair to say that a fair coin and the US economy are not equivalent; I'm just using the coin to demonstrate my point that the two ideas are different.

So I did go back and look. The worst year in FIREcalc history using calculator defaults - I think - is 1966. The annual returns for the few years prior were good but not spectacular; the mid-1980's had better returns but resulted in a success scenario. (Note: I only looked at nominal returns for the 1960's; I'm too lazy to look at real returns for that period, so it's possible that high inflation then would defeat my conclusion.)

What we can say (and note that Dory36 did in the post someone else quoted a little bit above) is that failure years tend to start with a significant negative real returns in their first few years. Although you may be able to determine this in advance by looking at the overall economy, I don't think you can just look at returns - even real returns - alone.

(*) If there is some I'd like to see it.

 

[ERD50]

Thanks for the citation but the credit goes to pb4uski:

oops, sorry pb4uski. I did have time to edit the post though, so that's better. Well, REWahoo has offered plenty of quotable comments, so one more plus/minus isn't too big a deal!


-ERD50

 

[SecondCor521]

I'm curious about one other thing. When people talk about a 30 year horizon that the 4% guideline applies to, are you using life expectancy tables or something similar like that which give a typical life span, or are they doing more of a "worst" case, with worst meaning the longest your portfolio would probably have to survive?


For example, the SS calculator says on average I'll probably live another 27 years. I'm 56 now. But that means that anything above average, and I'll live longer. I've run my VPW table to last beyond 100. Not because I expect to live over 100, but because I don't want to hit 100 and have very little beyond SS to live on.

As a side comment, note that your probability of living to age 100 and the probability of your portfolio surviving to your age 100 (using a 4% WR as an example) are - relatively speaking - independent variables. The true probability of a "running out of money" scenario - IMHO IANAL YMMV - is the probability of both. Greaney did the math, and you can find his post on the subject, but it actually raises the probability of success for 4% to something like 99.9%.

 

[Huston55]

Then how do you explain all those runs in FIRECalc that never get fat yet survive?

I can't explain it. It's just my view of it, and, as are many things I believe, likely flawed. I'm not an expert here but few would contribute if we limited inputs to those who are.

My thinking is that some years one "could" WD 6% and other years 2% depending on portfolio growth/or not. The 4% allows one to build up a reserve of sorts in strong years to fill-in for lean years.

But that's just how I view it. As noted, the concept could be flawed.

Perhaps the answer is ‘low volatility’ for the non-fat, surviving FIRECalc runs. Most of the time, volatility in the market is low (=flat-ish FIRECalc line), and reversion to the mean as time periods are lengthened also means flat-ish lines. Just a thought but, makes sense...n’est-ce pas?

 

[Independent]

And I'm sure (but open to being shown I'm wrong), that is not the case.

Just think about it. If 64-65 was positive (it might have been flat or down, but we can assume positive for the discussion), then yes, the retiree could ratchet up using the same % success factor. And that success factor includes 1966, right? Then 1966 hits - so what? It's already included in their scenario.

I really think that by definition, ratcheting up must work, as it is the same as retiring in that year. If your portfolio is $X in a given year, it is $X in a given year, period (just restating the fact). FireCalc doesn't "know" and cannot "care" if your portfolio was a different value in the earlier years. Your portfolio may have gone up or down earlier, due to income, spending, market performance, draw down from retirement - makes no difference - $X is $X.

Am I wrong? How? I just can't see it.

edit/add: Isn't "back testing" it the same as just entering the figures normally? Testing and "back testing" seem the same to me.

-ERD50

I'm assuming that we know for sure that retiring in 1966 didn't work, I haven't checked.

Yes, Hypothetical people who retired in 1964 and took 4% then ratcheted up in 1966 put themselves in the same position as people who retired in 1966. But, people who retired in 1966 and took 4% ran out of money before 30 years.

Meanwhile, hypothetical people who retired in 1964, took 4%, and didn't ratchet up, did make it 30 years.

People retiring in 1964 and ratcheting up in 1966, move from a scenario where they don't run out of money into a scenario where they do run out of money.

(Of course I'm assuming that 1966 failed within 28 years, not sure if that's true.)

Maybe 3.8% is a rate which survived for retirements in 1966. In that case, someone who retired in 1964, took 3.8%, and ratcheted up in 1966 to 3.8% did get into a successful scenario.

The problem is that a "safe" withdrawal rate of 4.0% isn't successful 100% of the time.

Edit: Regarding the post by Dory. Note that she starts by saying 4% survives the worst historical scenario. I've been assuming that 4% failed in 5% of the historical scenarios.

I'm not sure if the difference is rounding, or if there were small changes in the FireCalc formulas since then.

I just ran FireCalc with the standard assumptions. A $30,000 withdrawal from a $750,000 portfolio failed in 6 out of 117 scenarios.

FWIW, retirement years 1962, 63, and 64 were all successful years.
1965, 66, and 67 all ran out of money.
It appears the the first year for retirements in 1965 was an up, even though that year ultimately failed.
So the first two years of 1964 were up, ratcheting would have turned success into failure.

 

[pb4uski]

....HTHHTHHHTHHHHHHHHHH...

Sorry, but the whole coin flip analogy does nothing for me at all.... especially since the likelihood of an up year exceed the likelihood of a down year.

I somewhat subscribe to reversion to the mean... but am not totally convinced.

What I am sure of is that 4% is based on the weaker historical series... we can see that in a 4% run in FIRECalc that has a 94.5% success rate but that MANY of the lines do very well and only a few fail.

I'm trying to avoid both the lines that fail AND the lines that are obscenely successful... the former for because I don't want to eat catfood and rely on my kids and the latter because I want to be like Robbie as much as I prudently can and also avoid huge inheritances for my kids.

Since there are so many scenarios/lines that end really high it seems to me that ratcheting is a legit way to prudently pare some of the higher ones.

Realistically, it is all a bit acedemic though.... the reality is that as the future emerges we will adjust our spending up and if necessary, down so catfood is never a possibility.... even if we ratcheted to 4% in 1966, within a few years we would have seen it wasn't working really well and tightened our belts rather than blindly continuing 4% withdrawals.

 

[SecondCor521]

@pb4uski, I agree with you on ratcheting and plan to do it myself. I also agree with your last paragraph about it being academic.

My only point in my long post where I thought I maybe disagreed with you is the notion that retiring at a point when there has been a long bull run (say, today, for example) increases the probability that one is retiring into one of the 5% failure cases. That's all.

 

[audreyh1]

BTW - 1966 is NOT the worst year in the FIRECALC historical data. There are several worse years from around 1921 and prior. It’s by far the worst year post 1921, but when I was doing my analysis I found many worse years. 1906 was usually the worst year, but 1899 was worse for some other scenarios.

 

[RunningBum]

You're not "wrong", I'm just not sure there is much to be gained by looking at it your way.

...

OK, so if we look at past market conditions, and have some good correlation that those 5% failures occurred near market peaks (I think that's probably the case, though inflation and other factors are in play), then sure, you could say with some confidence that an overall 95% success factor might really be assumed to be much lower if we currently look similar to those past 'fail' conditions. But like I said, that's baked into the cake, or as [correction] [-]REWahoo[/-] pb4uski said, it's why FIRECalc says 4% under those inputs, and not 5%. It is the bad years that drove the 4% value, it really has nothing at all to do with the good years.

I don't see how it's "baked in". The 5% failure rate comes from those. If you used a 5% WR rate, the failure rate would be greater than 5%. It's not just those bad failure years that drove the 4% value, it's that it kept the pretty bad years from causing failure.

So look at it this way - historically, any WR from ~ 3.5% - 4% is 'flirting' with a danger zone if you retire into a bad scenario. The only real option is to be more conservative - what else can you do (if were looking for actionable items here, and not just a debate on probability and statistics)? And if you're trying to be conservative, why even think about a 95% success rate? Go to 100%, and then reduce that WR until the lowest ending portfolio has some buffer that you can accept. Or work until you die.

The advantage of a conservative starting WR is that you can conservatively apply the "retire again & again" scenario going forward, as the odds are pretty good you will be on one of the 'good' paths.

I think I'd agree with you - if you are using 95% success, and do a full ratchet up at every opportunity, you are increasing your odds of failure if you do it blindly. But as I said, that doesn't match up with a conservative approach anyhow, so I don't think it's really relevant. If I decide to ratchet up, it will be based on a conservative initial WR, and probably only a partial ratchet up. Or I might decide my heirs/charity can use it more than I can, and not ratchet up spending at all (but maybe give it away instead).

-ERD50

Please define what you consider to be conservative. Is a starting rate of 4%, and adjusting for inflation each year conservative? And then resetting to 4% after a run up?

Because that's what the OP asked about, and the first post gave an unqualified Yes, it will work. And dory's post gives the same unqualified Yes, and (I think) you and a few others jumped on that as proof.

That's not the same thing as, well, don't start quite as high as 4%, and don't adjust all the way, and be ready to adjust back down. But that's not what most people are saying, they want to put a sticky in dory's statement that you can take 4% and reset at any time.

In most cases either way will work, but what I've been arguing against is that notion that you can reset to 4% anytime you want without additional risk. If nothing else, as I think Independent said, you started at 95%, the bull market took you to 100% or very close, but now you are choosing to expose yourself to that ~5% failure again by resetting.

 

[pb4uski]

I don't see how it's "baked in". The 5% failure rate comes from those. If you used a 5% WR rate, the failure rate would be greater than 5%. It's not just those bad failure years that drove the 4% value, it's that it kept the pretty bad years from causing failure. ...

Below is the table from the Trinity study showing success rates for withdrawals increased for inflation..... unfortunately it doesn't copy in very well.

For 50/50 AA, 30 years the success rate for 4% WR is 95%.... so 5% of the trials that failed and those 5% of trials are preventing the WR from being higher.... for easier discussion, let's assume that there were 100 scenarios and 5 of them failed. Logically, those 5 have to be the 5 worst sequence of return scenarios, meaning that 95 of the scenarios are better because they pass.

Would you agree that if those 5 worst scenarios never happened that the WR that allows 95% success would be higher? If so, then those 5 worst scenarios are baked into the 4% SWR.

Payout Period 3% 4% 5% 6% 7% 8% 9% 10% 11% 12%

100% Stocks

15 Years 100 100 100 91 79 70 63 55 43 34

20 Years 100 100 88 75 63 53 43 33 29 24

25 Years 100 100 87 70 59 46 35 30 26 20

30 Years 100 95 85 68 59 41 34 34 27 15

75% Stocks/25% Bonds

15 Years 100 100 100 95 82 68 64 46 36 27

20 Years 100 100 90 75 61 51 37 27 20 12

25 Years 100 100 85 65 50 37 30 22 7 2

30 Years 100 98 83 68 49 34 22 7 2 0

50% Stocks/50% Bonds

15 Years 100 100 100 93 79 64 50 32 23 13

20 Years 100 100 90 75 55 33 22 10 0 0

25 Years 100 100 80 57 37 20 7 0 0 0

30 Years 100 95 76 51 17 5 0 0 0 0

25% Stocks/75% Bonds

15 Years 100 100 100 89 70 50 32 18 13 7

20 Years 100 100 82 47 31 16 8 4 0 0

25 Years 100 93 48 24 15 4 2 0 0 0

30 Years 100 71 27 20 5 0 0 0 0 0

100% Bonds

15 Years 100 100 100 71 39 21 18 16 14 9

20 Years 100 90 47 20 14 12 10 2 0 0

25 Years 100 46 17 15 11 2 0 0 0 0

30 Years 80 20 17 12 0 0 0 0 0 0

 

[RunningBum]

Below is the table from the Trinity study showing success rates for withdrawals increased for inflation..... unfortunately it doesn't copy in very well.

For 50/50 AA, 30 years the success rate for 4% WR is 95%.... so 5% of the trials that failed and those 5% of trials are preventing the WR from being higher.... for easier discussion, let's assume that there were 100 scenarios and 5 of them failed. Logically, those 5 have to be the 5 worst sequence of return scenarios, meaning that 95 of the scenarios are better because they pass.

Would you agree that if those 5 worst scenarios never happened that the WR that allows 95% success would be higher? If so, then those 5 worst scenarios are baked into the 4% SWR.

OK, yes, the failures are baked into the Firecalc model that shows a 95% success rate. For whatever that means.

Firecalc applies all the historical runs, but assumes no history leading up to each run's starting point. I contend that there are some economic situations that are more likely to trigger one of the failure runs than others. After the nice bull run we've had, I believe today is one of those times. I don't know how much worse the success odds would be. 94% success? 90%? I don't know.

That's why I think saying the failures are "baked in" doesn't apply, because right now we aren't at a random time with no history, we are at a time where we know without doubt we've had a long market run up and PEs are high.

 

[ERD50]

.... Please define what you consider to be conservative. Is a starting rate of 4%, and adjusting for inflation each year conservative? And then resetting to 4% after a run up? ....

I personally don't consider 4% conservative, others are free to have their own definition. For me, since it failed 5 times out of a hundred based on history, and we all know that the future could be worse than the worst of the past, I like to use 100% as a starting point, and the longest age I think we as a couple might achieve. I use 100, one of us could live longer, but I draw the line there - and it isn't very sensitive to a few more years at that point anyhow.

We are dealing with unknowns, so it becomes somewhat arbitrary. My position was, that to have a "clear conscious" that I could retire early, and not end up needing support from someone, I need to plan for something somewhat worse than the worst in history. And if it comes to that, so be it - you can only plan so far, and that was my limit.

... a starting rate of 4%, and adjusting for inflation each year conservative? And then resetting to 4% after a run up?

Because that's what the OP asked about, and the first post gave an unqualified Yes, it will work. And dory's post gives the same unqualified Yes, and (I think) you and a few others jumped on that as proof. ...

I don't think either of them said it was "conservative" to ratchet up? But (and important here - strictly sticking to the numbers that FIRECalc has in its database) - it works. It has to.

Again, lets use a 100% historically safe WR - it's easier to discuss. So if that is say 3.2% for the time duration you've entered, well, it obviously was 100% for every period - that's tautology. So anytime you ratchet it up to 3.2%, you are doing nothing more than re-testing it at 3.2% - it was already tested and passed, right? So it works. Period.

I've experimented with this concept with other tools, and guess what - the 'risky' starting point years like 1966, never provide an opportunity to ratchet up (as expected). They don't get any 'riskier'. What it does is (if you want to put it this way), makes the other years 'riskier'. The useful effect of that is to preserve capital for the bad runs, and allow you to utilize the maximum capital from better runs, leaving less on the table in those cases. It's just arithmetic really.

That's a mechanical/mathematic view of it, with that data set. I can't see anyway for it to not work under those conditions. It's like if A > B and B > C, then A must be > C.

Now, it is 'riskier' (lowers the portfolio balance) to ratchet up? Of course it is! Again, it has to be! Taking more money out can only result in less money remaining. A>B>C.

We are going in circles. I think I mentioned a few posts back - what is "actionable" here? Here' a way I think you could approach it:

A) Find the 100% safe WR for your scenario. This WR% will get you through the worst in history - forget % success, you are covered. If you want to be more conservative, reduce that number to provide a buffer against a really bad future. Put in as much buffer as you feel comfortable with.

B) If you want, evaluate where you think we are in a cycle. But using your logic, that was already considered in A above. So if you think we aren't right a precipice, the only logical thing to do would be to RAISE your spending. But I don;t think that's your point, you are thinking the other way, and saying if we are in a danger area, we need to cut. But we already did that in "A".

OK, I guess I'm done, I can only repeat myself so many times, redundantly, and over and over. It simply boils down to how conservative you want to be. Just recognize that FC is already giving you the worst case in its history - go from there.

-ERD50

 

[RunningBum]

...

We are going in circles. I think I mentioned a few posts back - what is "actionable" here? Here' a way I think you could approach it:

A) Find the 100% safe WR for your scenario. This WR% will get you through the worst in history - forget % success, you are covered. If you want to be more conservative, reduce that number to provide a buffer against a really bad future. Put in as much buffer as you feel comfortable with.

B) If you want, evaluate where you think we are in a cycle. But using your logic, that was already considered in A above. So if you think we aren't right a precipice, the only logical thing to do would be to RAISE your spending. But I don;t think that's your point, you are thinking the other way, and saying if we are in a danger area, we need to cut. But we already did that in "A".

OK, I guess I'm done, I can only repeat myself so many times, redundantly, and over and over. It simply boils down to how conservative you want to be. Just recognize that FC is already giving you the worst case in its history - go from there.

-ERD50

Yes, I think this is good actionable advice for conservative people, like me. I just wanted to make sure I knew where you were coming from when you talked about being conservative, because that means different things to different people.

I don't think that starting at 4%, and resetting to 4% after a strong market run up, is good actionable advice, as some here are saying. Not without some caveats, and acknowledgement of the risks. I think it's a mistake to tack dory's post on the wall and treat it as gospel.

 

[ERD50]

...

I don't think that starting at 4%, and resetting to 4% after a strong market run up, is good actionable advice, as some here are saying. Not without some caveats, and acknowledgement of the risks. I think it's a mistake to tack dory's post on the wall and treat it as gospel.

I think you are reading that into it. I didn't see Dory and especially REWahoo saying that it was good, actionable advice.

What I saw in those posts, was that it works within the confines of FIRECalc's data set.

Theoretically, you should be able to reset your SWR at any time. ...

And as I've said, number wise, it must work, I can't how it cannot. I'm OK with calling that "gospel".

As far as what is actionable for you, I think you've got your answer (or at least a methodology to approach an answer).

Are we done now?

-ERD50

 

[RunningBum]

I think you are reading that into it. I didn't see Dory and especially REWahoo saying that it was good, actionable advice.

OK, they never used the words "actionable advice" specifically.

What I saw in those posts, was that it works within the confines of FIRECalc's data set.

but they never put that kind of disclaimer or caveat on it either. Firecalc wasn't even mentioned in Dory's post, nor was any notion that this was in the confines of any system. You apparently read that in.

Are we done now?

-ERD50

Sure.

 

[REWahoo]

Are we done now?

Sure.

Me, [-]too[/-] three (bull market!).

 

[USGrant1962]

I've experimented with this concept with other tools, and guess what - the 'risky' starting point years like 1966, never provide an opportunity to ratchet up (as expected).

BINGO - we have a winner! Any reasonable ratcheting strategy simply captures some of the excess wealth that accumulates >90% of the time using a SAFEMAX withdrawal rate.

 

[Independent]

I personally don't consider 4% conservative, others are free to have their own definition. For me, since it failed 5 times out of a hundred based on history, and we all know that the future could be worse than the worst of the past, I like to use 100% as a starting point, and the longest age I think we as a couple might achieve. I use 100, one of us could live longer, but I draw the line there - and it isn't very sensitive to a few more years at that point anyhow.

....
Again, lets use a 100% historically safe WR - it's easier to discuss. So if that is say 3.2% for the time duration you've entered, well, it obviously was 100% for every period - that's tautology. So anytime you ratchet it up to 3.2%, you are doing nothing more than re-testing it at 3.2% - it was already tested and passed, right? So it works. Period.

I agree with your statement that if I start with a WR which was successful in every one of the past 117 thirty year periods, I can ratchet up in response to good experience and say that I still have a withdrawal amount which was successful in every one of the past 117 thirty year periods.

But, the OP specified that he was withdrawing 4%, not 3.2%. FireCalc says that 4% is 95% successful.

Suppose I start with a 4% rate, and have good early results. This good news may have put me into one of those 100% successful rates, assuming I don't ratchet. If I do ratchet, I take myself out of a 100% successful WR and into a 95% successful WR.

IMO, that's a choice that requires some thought.

 

[RobLJ]

I seem to remember at some point over the last few years Bergen indicated that the SWR is more 4.5 than 4%. The most likely "risk" is that you'll die with a huge unspent estate at 85-90 than running out of money.

Like a few here, I'm more concerned with the portfolio being 4x-8x the starting level when I croak, because of the 4% SWR. So I'm fine with ratcheting up, although not too frequently.
We have a lot of slack in the budget, so that means I'm also willing to ratchet down under severe conditions, which is more like a Variable rate or really a hybrid.

But I would prefer to pass on money to the kids and to my grandbaby while it can do them good than pass on money only after I die 30 years from now off a huge estate that hasn't done them any good until they are greater than 60.
Just my view. We could cut back 30% without too much pain; both of us grew up without any thought of SWR amongst our parental units. If you told me I would be in this position (early retirement debating about the "right" SWR) I would have laughed at you.

 

[kcowan]

This. Just use 4% rule to make sure you have enough assets to retire in the first place. It would be very silly to literally use this rule once retired. It was never intended to be used that way.

There seems to be a slavish attachment to the 4% "rule" in this crowd!

Your "innumerate lawyer" is over-simplistic, but is doing the first basic step to the calculation.

1/30~3.33% (adjust longevity as needed).

I really think that people coming up with ultra-low SWRs have forgotten about this basic first step.

Right the 3.33% "rule"!

 

[Danmar]

There seems to be a slavish attachment to the 4% "rule" in this crowd!Right the 3.33% "rule"!

Could be. My biggest reason for not worrying too much about the “rule” is that my portfolio is quite different than that used in the studies. Firstly, I think the Canadian market supports a higher WR (based on historical results). Plus my portfolio has outperformed the Canadian market by about 4% per year over the last 20 years. Granted, I’m not as diversified as most people would want and that is my biggest risk, one I’m comfortable with.

In addition, once you have been retired for a while, rules of thumb seem less important. The actual sequence of return results would seem to me to be the important factor.

 

[USGrant1962]

There seems to be a slavish attachment to the 4% "rule" in this crowd!Right the 3.33% "rule"!

Interestingly, if you fiddle with FIRECalc you will find that 3.33% is exactly 100% success for a 40-year retirement at the defaults - 75/25, no SS, etc.

 

[kcowan]

Interestingly, if you fiddle with FIRECalc you will find that 3.33% is exactly 100% success for a 40-year retirement at the defaults - 75/25, no SS, etc.

Great so my innumerate lawyer who has never seen a spreadsheet came pretty close with his back of napkin calculation. It seems to reinforce his "Don't confuse me with numbers!" mantra...

 

[GravitySucks]

To the OP:
If you're slavishly following FIRECalc and the 4% rule notice that if you change the period from 30 to 25 years the historical failure rate is less than 2%, so you're odds of 'winning' are even better now that you're 8 years closer to the abyss.

Personally, I am hitting the 5 year mark this year and will have unfettered access to my main retirement account. The plan has always been to reevaluate this year considering both finances and health. Fortunately I'll be doing the evaluation from a position of both better health and way better finances than expected. I hope to be reevaluating every five years. We do still have free will and health and finances will change.
I expect I will be spending a bit more than the 4.5% I budgeted the last 5 years if I can figure what to spend it on. The only requirement I put on the new spending is that it is close to completely discretionary and can be cut out at will. (Not a second home, boat, or child in other words.)