Stop worrying about the 4% rule

[brooklynguy]

So in 2008 assuming 2% inflation, Retiree A would pull out $40,800 of the $605,000 but based on 4% SWR they should still have a 96% chance of success for the money lasting them 29 more years.

Retiree B can only take out $25,200 in 2008 but still only has a 96% success probability. 

This is something I can never seem to reconcile. If the odds are the same then Retiree B should be able at least take out as much as Retiree A. 

The answer is that in reality there is no inconsistency that needs to be reconciled--in reality, Retiree B's withdrawal plan clearly has better odds of succeeding than does Retiree A's.  The issue you point out arises only because of our tendency to use SWR terminology too loosely.  A given withdrawal rate (under given portfolio parameters) has a known rate of historical success, but not a known probability of future success.  You can't use the former as a proxy for the latter unless you make both of the following assumptions (neither of which is true in reality):  (i) the entire universe of possible outcomes in the future is perfectly reflected in the distribution of outcomes that have actually occurred in the past and (ii) at any given time, each of these possible outcomes has an equal likelihood of occurring.  We tend to use shorthand language (such as "success probability" in your post quoted above) that on its face appears to use historical success rates as a proxy for likelihood of future success without explicitly spelling out this caveat, but it always applies.

[oysters]

Agreed - good debate should be encouraged.  The pushback I see a lot with early retiring is really not about the 4% rule, but rather its about whether a person has estimated/anticipated their expenses correctly. 

And I think that's valid.  It's far more likely that you'll make a mistake with estimating expenses than it is for the 4% rule to fail. 

Here's an example - divorce.  As a 45 year old, employed white male, I have a 7% chance of getting divorced.  That rises to over 20% by the time I am 60.  So my RE is far more likely to fail from divorce than it is for the 4% rule to fail.

http://flowingdata.com/2016/03/30/divorce-rates-for-different-groups/

Its 3x to 5x more likely that divorce will do me in than a 4% rule failure.

I went to your link, maybe I'm looking at the wrong graph, but if it's the right one it is labeled "divorced or married more than once" which would sound like it includes both your risk of divorce AND your risk that your spouse passes away and, after an appropriate interval you meet someone else and remarry (which would explain why the risk goes up so much as you get older).

You're right, I was in a hurry and didn't read it as closely as I should have.  This has better data - http://www.pewresearch.org/fact-tank/2017/03/09/led-by-baby-boomers-divorce-rates-climb-for-americas-50-population/

And still, 21% chance of divorce for 40-49 year olds and 10% for 50 and over.  Way higher fail rates than the 4% rule.

I've already gone through my divorce, so at 33, my Stash will only go up from here ;-)

[Monkey Uncle]

My take on the BH side of things is not based on hanging out on those forums just the high savings targets and low WR rates that get reported here. I think the assumptions around that approach needs to be challenged because the trade off in time working at the prime of your life is a really high cost to pay. Both for the specific health and relationship impacts on the person working as well as their loved ones and also because of the environmental impacts that affect us all from high spending lifestyles. My party comment about BH was flippant shorthand for over working vs. spending time on relationships. If anyone was offended I apologize and I'll delete the comment.

That, and the fact that there is no evidence to support the need for such super-low withdrawal rates, other than the world as we know it might end.  Well, yes, it might.  In which case we are all screwed, and your WR won't matter.

[honeyfill]

I have a better idea.  Instead of trying to make your money out last your life. Why not try to make your life end before your money?  Therefore I have started drinking heavily, taken up smoking and got into sky diving and rock climbing!  I bet I dont last 5 years at this rate.  There fore I am pulling out 20% a year!!

[Retire-Canada]

Right, and even more incredible is that Retiree A could use the bulletproof 3% SWR and have $30,600 to spend in 2008, with 100% historic success, and Retiree B still only gets $25,200 with 96% success!

If you try and use a hammer as screwdriver you are bound to be disappointed in its effectiveness as a tool. Despite it being pretty great at banging nails into wood. The 4% Rule tells you what would have happened based on various starting dates throughout history. With only 1 input [starting portfolio value] it can by design not factor in recent market events that happen in your pre or post retirement. Your retirement will succeed or fail the 4% Rule doesn't tell you what will happen in your case.

Retiree A is aware that his poor early returns are potentially putting his FIRE success at risk. He can ride it out as planned and see where he's at after a few years or he can take corrective action early just in case he's in a starting year that fails. Retiree B knows that he's either got to FIRE into a crash and ride the upswing out or put off full retirement for a bit until the recovery is complete.

The 4% Rule did its job which was to get them both to a point where they had $1M saved after probably 10-20yrs of effort shooting for that target. Expecting the 4% Rule to solve all their problems around how to deal with the 2008 crash would be expecting too much from a simple tool. 

[Exflyboy]

I have a better idea.  Instead of trying to make your money out last your life. Why not try to make your life end before your money?  Therefore I have started drinking heavily, taken up smoking and got into sky diving and rock climbing!  I bet I dont last 5 years at this rate.  There fore I am pulling out 20% a year!!

As an ex-skydiver, rock climber (and engineer, therefore risk adverse). I can tell you that skydiving and rock climbing are both very safe sports if done properly.. I recommend driving and taking opioids as "better" alternatives..:)

[Monkey Uncle]

I have a better idea.  Instead of trying to make your money out last your life. Why not try to make your life end before your money?  Therefore I have started drinking heavily, taken up smoking and got into sky diving and rock climbing!  I bet I dont last 5 years at this rate.  There fore I am pulling out 20% a year!!

As an ex-skydiver, rock climber (and engineer, therefore risk adverse). I can tell you that skydiving and rock climbing are both very safe sports if done properly.. I recommend driving and taking opioids as "better" alternatives..:)

Actually, just continuing to work a high-stress job would probably be the most dangerous thing you could do...

[steveo]

My take on the BH side of things is not based on hanging out on those forums just the high savings targets and low WR rates that get reported here. I think the assumptions around that approach needs to be challenged because the trade off in time working at the prime of your life is a really high cost to pay. Both for the specific health and relationship impacts on the person working as well as their loved ones and also because of the environmental impacts that affect us all from high spending lifestyles. My party comment about BH was flippant shorthand for over working vs. spending time on relationships. If anyone was offended I apologize and I'll delete the comment.

That, and the fact that there is no evidence to support the need for such super-low withdrawal rates, other than the world as we know it might end.  Well, yes, it might.  In which case we are all screwed, and your WR won't matter.

I think that this is the issue. There is no evidence supporting a really low WR other than being extremely risk averse or wanting to somehow become rich in a monetary/bank balance sense.

I can understand both points of view in that I would hate for my retirement to fail and I don't really want to go back to work plus I suppose at some point I'd like to think I've got enough money to be fine forever or just to have a bank balance and go "woah".

In stating that these are psychological issues and not statistical/math based issues in relation to safe WR's.

I still think a 4% or higher WR should be really safe especially if you have some flexibility. If you are really worried about your WR I also think focussing on having some buffer in your expenses is probably a safer approach. I think the ERE approach of getting really low expenses is less robust than a MMM approach of having what I consider reasonable expenses. If your expenses are really tight and you get hit by a big expense that could throw you I think a lot more than having reasonable expenses and being able to scrimp and save on those expenses. For instance go without a car or buying some new clothes for a year or two.

[Padonak]

Similar to what ExFlyboy asked earlier (a page ago), but with a twist.  What if Retiree A and B have $1M.  Retiree A calls it quits in 2007 and withdraws his inflation adjusted $40k for 30 years.  Retiree B hangs in a little longer only to suffer the ~37% decline in 2008.   Retiree A has roughly 605k and Retiree B has roughly 630k. 

How much can Retiree B plan to spend for the next 30 years?

Based on the 4% "rule", 4% of $630K.

Of course, the "4% rule" is not as straightforward as, say, the solution to the quadratic equation.  It is the result of backtesting a simplistic withdrawal strategy.  Perhaps a more convoluted withdrawal strategy could be backtested, such as "use (a + b*X) as the initial withdrawal amount, where X is the CAGR of the previous 3 years for the S&P500 and 'a' and 'b' are fit to the historical data."  But ol' Occam would object, and I think rightly so.

Using 4% (or its inverse, 25X) to determine if one is "in the ballpark" for retirability is reasonable.  Asking for exactitude in predicting the future is not.

This is the circular logic that gets missed IMO. 

So in 2008 assuming 2% inflation, Retiree A would pull out $40,800 of the $605,000 but based on 4% SWR they should still have a 96% chance of success for the money lasting them 29 more years.

Retiree B can only take out $25,200 in 2008 but still only has a 96% success probability. 

This is something I can never seem to reconcile. If the odds are the same then Retiree B should be able at least take out as much as Retiree A. 

This is obviously sequence of return risk (Retiree A) and what never gets talked about sequence of returns opportunity (Retiree B).  But to believe in either you must have to also believe in some form of market fundamentals matter or market timing or whatever because it is absolutely certain that the risk profile between the two is drastically different.

The bolded part is profound. The church of the 4% would set you on fire as a heretic if we lived in the middle ages.

On a serious note, I came across a spreadsheet (could have been earlier in this thread) with different safe withdrawal rate estimates for different PE10 bands. It's pretty common in science and analytics to split the population into bands or buckets based on important parameters (cyclically adjusted PE in this case). For example, the expected safe withdrawal rate for a person retiring when PE10 is in the 5-10X bucket is much higher than for somebody who retires when PE10 is in the 25-30X bucket. I'll post the spreadsheet if i find it. It's not perfect, and we can argue about the cutoff points and whether we should use PE or perhaps a combination of PE, Price to Book, Price to Sales and other metrics, but I think this approach is much more accurate than just blindly using the 4% rule regardless of where we are in the economic cycle.

Edit: found the spreadsheet and related post. I didn't check it's accuracy. I'm not affiliated with this blog.
https://earlyretirementnow.com/2017/01/25/the-ultimate-guide-to-safe-withdrawal-rates-part-7-toolbox/

[sol]

The bolded part is profound.

It's not profound, it's just math.  Math that has been discussed for pages and pages in this very thread, and a bunch of others on this forum.

If this issue is still confusing to anyone, or if it seems paradoxical or confusing, I suggest you go back to page one of this thread and start following all of the links posted here.  You've misunderstood what the 4% SWR is if it doesn't make perfect sense to you why these two retirees have different dollar amounts based on the 4% rule.

If you need more guidance than "please read the thread you're currently posting in" then allow me to suggest you spend an hour or two googling "sequence of return risk" and trying to understand why the worst year in market history never seems to happen immediately after the second worst year in market history.

[Goldielocks]

ERE sometimes takes a bunker mentality. Become self-sufficient in case TSHTF.

MMM's optimism seems unique among the ER crowd, which is why it's such a draw for many of us.

Yes, it means we are rosy about the 4% rule, but it also means if it doesn't pan out, we'll happily get back to work (literally or figuratively) and make adjustments and be fine. :)

haven't followed Ere recently,  but there is a huge difference between planning to retire early on a bare minimum expenses with no SS likely to come in future,  and no way to cut out 30% of your costs to hold out a series of bad return tears.... And planning to retire on 50k plus per year with SS part of that.

[Monkey Uncle]

Similar to what ExFlyboy asked earlier (a page ago), but with a twist.  What if Retiree A and B have $1M.  Retiree A calls it quits in 2007 and withdraws his inflation adjusted $40k for 30 years.  Retiree B hangs in a little longer only to suffer the ~37% decline in 2008.   Retiree A has roughly 605k and Retiree B has roughly 630k. 

How much can Retiree B plan to spend for the next 30 years?

Based on the 4% "rule", 4% of $630K.

Of course, the "4% rule" is not as straightforward as, say, the solution to the quadratic equation.  It is the result of backtesting a simplistic withdrawal strategy.  Perhaps a more convoluted withdrawal strategy could be backtested, such as "use (a + b*X) as the initial withdrawal amount, where X is the CAGR of the previous 3 years for the S&P500 and 'a' and 'b' are fit to the historical data."  But ol' Occam would object, and I think rightly so.

Using 4% (or its inverse, 25X) to determine if one is "in the ballpark" for retirability is reasonable.  Asking for exactitude in predicting the future is not.

This is the circular logic that gets missed IMO. 

So in 2008 assuming 2% inflation, Retiree A would pull out $40,800 of the $605,000 but based on 4% SWR they should still have a 96% chance of success for the money lasting them 29 more years.

Retiree B can only take out $25,200 in 2008 but still only has a 96% success probability. 

This is something I can never seem to reconcile. If the odds are the same then Retiree B should be able at least take out as much as Retiree A. 

This is obviously sequence of return risk (Retiree A) and what never gets talked about sequence of returns opportunity (Retiree B).  But to believe in either you must have to also believe in some form of market fundamentals matter or market timing or whatever because it is absolutely certain that the risk profile between the two is drastically different.

The 4% rule does NOT say that retiree A has a 96% chance of success at the end of 2008, one year into retirement and after a huge market decline.  It says that, in 2007, on the cusp of retirement and with no crystal ball knowledge about what the future holds, retiree A has an average probability of success of 96%, assuming that the future is no worse than the past.   Once his retirement has started, he is on a specific trajectory and the universe of possible trajectories that was open to him at the outset (i.e., the average) no longer exists.  He might be on one of the 4% of initial trajectories that lead to failure.  If you want to know his probability of success at the end of 2008, it would be easy enough to calculate using his balance at that time, withdrawal rate, and life expectancy at that time. But again, that would assume no crystal ball knowledge about the future, so if the result is, say, 68% (I'm just making that up), that doesn't mean his retirement going forward will be a 32% failure.  He is still on a trajectory, and that trajectory might wind up in the 68% of possible scenarios that succeed.

[EscapeVelocity2020]

Similar to what ExFlyboy asked earlier (a page ago), but with a twist.  What if Retiree A and B have $1M.  Retiree A calls it quits in 2007 and withdraws his inflation adjusted $40k for 30 years.  Retiree B hangs in a little longer only to suffer the ~37% decline in 2008.   Retiree A has roughly 605k and Retiree B has roughly 630k. 

How much can Retiree B plan to spend for the next 30 years?

Based on the 4% "rule", 4% of $630K.

Of course, the "4% rule" is not as straightforward as, say, the solution to the quadratic equation.  It is the result of backtesting a simplistic withdrawal strategy.  Perhaps a more convoluted withdrawal strategy could be backtested, such as "use (a + b*X) as the initial withdrawal amount, where X is the CAGR of the previous 3 years for the S&P500 and 'a' and 'b' are fit to the historical data."  But ol' Occam would object, and I think rightly so.

Using 4% (or its inverse, 25X) to determine if one is "in the ballpark" for retirability is reasonable.  Asking for exactitude in predicting the future is not.

This is the circular logic that gets missed IMO. 

So in 2008 assuming 2% inflation, Retiree A would pull out $40,800 of the $605,000 but based on 4% SWR they should still have a 96% chance of success for the money lasting them 29 more years.

Retiree B can only take out $25,200 in 2008 but still only has a 96% success probability. 

This is something I can never seem to reconcile. If the odds are the same then Retiree B should be able at least take out as much as Retiree A. 

This is obviously sequence of return risk (Retiree A) and what never gets talked about sequence of returns opportunity (Retiree B).  But to believe in either you must have to also believe in some form of market fundamentals matter or market timing or whatever because it is absolutely certain that the risk profile between the two is drastically different.

The 4% rule does NOT say that retiree A has a 96% chance of success at the end of 2008, one year into retirement and after a huge market decline.  It says that, in 2007, on the cusp of retirement and with no crystal ball knowledge about what the future holds, retiree A has an average probability of success of 96%, assuming that the future is no worse than the past.   Once his retirement has started, he is on a specific trajectory and the universe of possible trajectories that was open to him at the outset (i.e., the average) no longer exists.  He might be on one of the 4% of initial trajectories that lead to failure.  If you want to know his probability of success at the end of 2008, it would be easy enough to calculate using his balance at that time, withdrawal rate, and life expectancy at that time. But again, that would assume no crystal ball knowledge about the future, so if the result is, say, 68% (I'm just making that up), that doesn't mean his retirement going forward will be a 32% failure.  He is still on a trajectory, and that trajectory might wind up in the 68% of possible scenarios that succeed.

This is getting a bit convoluted, but I think it was clear that the 96% probability was for 2007 with the hope that the trajectory was overwhelmingly poised for success.  With this in mind, Retiree A can decide if they want to continue with 40k + inflation, or maybe trim it back a little for a year or two to maybe 30k.  I disagree though, for Retiree A, that they would want to re-run the probabilistic outcome based on the new WR.  What could possibly be gained by adding another % probability that isn't telling you what the 4% rule was intended for anyway?

The question was more for Retiree B - do they really take such a massive haircut vs. Retiree A, or have to put in OMY(s)?

[maizefolk]

Retiree B can spend the same amount as Retiree A, and have the same odds of success.

However, the benefit of addition data (one year of post FIRE returns), is that it allows us to update our prior estimate for the probability of success for retiree A (and hence B) -- given completely inflexibly 4% spending for the next 29 years, which is probably an unrealistic assumption in a recession -- as being less than 96%.*

The difference between A & B isn't necessarily their net worth. It is that retiree B has an extra year of data which retiree A lacked.

[EscapeVelocity2020]

Retiree B can spend the same amount as Retiree A, and have the same odds of success.

However, the benefit of addition data (one year of post FIRE returns), is that it allows us to update our prior estimate for the probability of success for retiree A (and hence B) -- given completely inflexibly 4% spending for the next 29 years, which is probably an unrealistic assumption in a recession -- as being less than 96%.*

The difference between A & B isn't necessarily their net worth. It is that retiree B has an extra year of data which retiree A lacked.

How can this be consistent with the 4% rule if Retiree B is embarking on a 30 year retirement at the end of 2008?  Also, Retiree A has the benefit of one less year of retirement to fund over Retiree B and they took out money before the decline.  The 4% rule was not designed to calculate the probability of a 6% WR that starts after a 37% decline lasting 29 years...

But if Retiree B takes a 6%WR, FIRECalc argues that Retiree B now has a 45.3% probability of sucess.

[maizefolk]

If retiree B wants to spend 4% of their current stash for 30 years they have a 96% chance of success.

If retiree B wants to spend ~6.7% for 29 years (the same spending plan at retiree A at that point) they have the same odds of success as retiree A. And given the availability of an extra year of data which wasn't know to retiree A when they started on their retirement, those odds are now less than 96% for either A or B.

If retiree B wants to spend ~6.7% for 30 years (a more ambitious plan than retiree A), then B's odds of success are presumably slightly lower than A's, although likely not by a measurable amount.

Edit: based on the update, yes I agree with you, just plugging the numbers of retiree after one year of retirement have passed isn't going to provide an ideal estimate of success rate. This starts to get complex, so I'm going to start with four statements, I think we can all agree to (but let me know if you don't).

#1: I'm not sure if we have a good way to quantify the amount of change in success rate based on the new information, but we know most failures of a 4% WR strategy do result from sequence of withdrawal risk early in the retirement, so a 1/3 decline increases the likelihood that we are in one of those failure scenarios relative to the odds of being in a failure scenario when retiree A pulled the trigger.

#2: Similarly, we know that if retiree A and retiree B spend the same inflation adjusted amount as a percentage of their starting portfolio for the same number of years, their odds of success are identical.

#3: We know that adding additional years to a FIRE scenario will never increase success rates, and sometimes decrease it. (So retiree B retiring for 30 years rather than 29 assumes some, unmeasured, additional quantity of risk).

#4: We know that spending a smaller proportion of your stash (especially at high withdrawal rates like 6-7%) decreases risk. (So if retiree B retires spending the same amount as retiree A -- and hence a slightly smaller percent of assets -- they reduce their risk by some, unmeasured, additional amount).

What we don't know is the size of any of these effects. If the size of effect #3 is greater than the size of effect #4, then retiree B has slightly more risk than retiree A. If #4 is greater than #3, retiree B has slightly less risk than retiree A.

But I think points #1 and #2 explain the parts of the original scenario that people were considering someone paradoxical, or representing a flaw in the logic of the 4% rule.

[Retire-Canada]

I think it would be most helpful to people reading this thread if we stopped saying stuff like "The 4% Rule says Retiree A has a 96% chance of success...." The 4% Rule doesn't know shit about Retiree A. All it says is that based on historical data a 4% WR had a 96% success rate in the past. That's it. As soon as you project its results into the future with statements like "...assuming the future is no worse than the past..." you are on shaky ground. I get what you are trying to say, but I think it's just more easily understood to say nothing about the 4% Rule and the future.

Now you may say to yourself based on past results with a 4%WR I am happy to base my retirement on it because I am optimistic for the future. That's a reasonable statement and I think it puts the 4% Rule input in the correct context as a guideline for FIRE planning not a predictive tool. In particular because nobody I know of is actually going to follow the exact WR process used by the research that forms the basis of the 4% Rule.

[brooklynguy]

If retiree B wants to spend 4% of their current stash for 30 years they have a 96% chance of success.

Although I believe it is already implicit in this shorthand statement intended for an audience well-versed in SWR methodology, in order to avoid the type of confusion that steered this thread down this path to begin with, I would explicitly note that in order for this statement to be technically correct it needs the following caveat:  "assuming that such spending plan's historical success rate is a perfect proxy for its present probability of success (which, incidentally, is not an accurate assumption)."

(Maizeman, you hung an asterisk on the 96% figure in your post two posts up but then didn't insert any footnote language, so maybe you intended to make a similar point?)

Edit:  I see Retire-Canada beat me to it and just made the same point.

[maizefolk]

Big edit to my last post based on EV's update (thanks for the PM to let me know you'd added extra text).

brooklynguy, no that's an important point but my forgotten footnote was just going to be the point that I don't know of a good way to estimate how much the estimate of the odds of retiree A should be reduced relative to the estimate of success when they pulled the trigger on FIRE, just that the change from the new data is non-zero and negative.

[Mr Mark]

I think it would be most helpful to people reading this thread if we stopped saying stuff like "The 4% Rule says Retiree A has a 96% chance of success...." The 4% Rule doesn't know shit about Retiree A. All it says is that based on historical data a 4% WR had a 96% success rate in the past. That's it. As soon as you project its results into the future with statements like "...assuming the future is no worse than the past..." you are on shaky ground. I get what you are trying to say, but I think it's just more easily understood to say nothing about the 4% Rule and the future.

Now you may say to yourself based on past results with a 4%WR I am happy to base my retirement on it because I am optimistic for the future. That's a reasonable statement and I think it puts the 4% Rule input in the correct context as a guideline for FIRE planning not a predictive tool. In particular because nobody I know of is actually going to follow the exact WR process used by the research that forms the basis of the 4% Rule.

+1

I like to invoke the power of the optimism gun.

[PizzaSteve]

Big edit to my last post based on EV's update (thanks for the PM to let me know you'd added extra text).

brooklynguy, no that's an important point but my forgotten footnote was just going to be the point that I don't know of a good way to estimate how much the estimate of the odds of retiree A should be reduced relative to the estimate of success when they pulled the trigger on FIRE, just that the change from the new data is non-zero and negative.

As stated, it is important to remember thst the % success rate the study provides is the result of a model, meant to be applied specifically at the point of retirement.  The assumptions include withdraw rate and sequence of returns consistent with the recent past.  That is it.  It is not a life plan, nor a true  statement of success probability, though we assume the correlations are high.

In the hypothetical scenario, each retiree should evaluate their situation and either accept failure risk or decline failure risk by deciding how much to spend.  What the model may have stated as a probability years earlier is not relevant. The stock returns are in a sense memoryless, though one might make assumptions that clustered bad years raise the liklihood of recovery (which assumes good management of the economy, by the way...the steady returns are highly related to our nations good governance model...so all bets are off if we kill off our govt.).

Anyway, just my opinion.  The hypothetical A vs B example is not very useful, mathematically, because it is over reaching about what the model tells us. If does seem useful for starting a dialog about the behavior aspects of spend and withdraw  rate.  Should we stick with the stready amount throughout retirement?  If we luck out with high returns in our early years, should we spend more ? Should we rerun the numbers every year and adjust our lifestyle or hedge more and possibly over save?  Etc.  All good questions, but it is up to each person to save, spend or worry to their own level of comfort.

[tooqk4u22]

I love it, so hysterical.  General theme is that 4% rule is rock solid and you should absolutely base your retirement (but always with the caveat to be flexible)....until it is challenged of course.

Question two different scenarios for timing of FIRE....all these hedges come out
...its not a rule
...based on historical data
....its just a model and doesn't mean anything going forward
....you can't control the future
....retiree would or should reduce their withdrawals
...its not realistic to just blindly follow it
...blah blah blah.

All that may be true or not but the fact of the matter is that the 4% rule doesn't go far enough (but still may be valid) and the two retiree scenario posted before would have drastically different withdrawals based on the time they retired (absent adjustments/flexibility/current market knowledge as none that has any factor on WR analysis). 

The scenario presented was the negative view - markets go down, but the inverse is also a problem.....Retiree A goes in x year and retiree B goes one year later but after a 30% return.  Same thing applies Retiree A is living off of $40k (+inflation) and Retiree B is living off of $52k (+inflation) (based on $1mil and 4%WR).  Yet both still have the same probability at the start of the retirement. 

I would be far more willing to accept a higher WR after a downturn or lower rate after a good run...call it logic or intuition or whatever....but then if I do this I am basically conceding that the 4% rule is worthless and should instead focus on my own views (that doesn't seem quite right either).

We have all heard and believe that during the accumulation phase that savings rate is more important than returns....well I think based on all of this entire thread's discussion and discussions elsewhere that we could probably all agree that during retirement that being flexible is far more important than WR, which kind of sucks bc then it diminishes the more data driven approach.

[arebelspy]

Plenty of people believe in a WR based on market conditions. MF has a whole post on this, IIRC.

[maizefolk]

All that may be true or not but the fact of the matter is that the 4% rule doesn't go far enough (but still may be valid) and the two retiree scenario posted before would have drastically different withdrawals based on the time they retired (absent adjustments/flexibility/current market knowledge as none that has any factor on WR analysis). 

I don't understand your point here. The 4% rule is based on having no knowledge of what the stock market is going to do the year after you retire. Of course, once you've lived through that year, you can make a better estimate of whether your stash is going to last the remaining 29 years than you could before. That's how adding more information works.

But until we have a way to predict what next year's returns are before we get to next year, I'm not sure the above observation is actually helpful to people trying to make decisions about when to FIRE.

The scenario presented was the negative view - markets go down, but the inverse is also a problem.....Retiree A goes in x year and retiree B goes one year later but after a 30% return.  Same thing applies Retiree A is living off of $40k (+inflation) and Retiree B is living off of $52k (+inflation) (based on $1mil and 4%WR).  Yet both still have the same probability at the start of the retirement. 

But this is just another example of "yes, you can improve your predictions over time as you get more data about how your own retirement window has already behaved."

It's like saying "your odds of being attacked by a shark are 1 in 11.5 million" but if you go to the beach, and you're out surfing and you see a shark fin circling around you, at that point your odds of being attacked as significantly higher than 1 in 11.5M. That doesn't make the first estimate wrong, invalid, or misleading. It just means that as time passed and more data accumulates on your risk factors, we can do a better job of estimating the chances that you're going to fall into that 1 in 11.5 million.

[secondcor521]

It's like saying "your odds of being attacked by a shark are 1 in 11.5 million" but if you go to the beach, and you're out surfing and you see a shark fin circling around you, at that point your odds of being attacked as significantly higher than 1 in 11.5M. That doesn't make the first estimate wrong, invalid, or misleading. It just means that as time passed and more data accumulates on your risk factors, we can do a better job of estimating the chances that you're going to fall into that 1 in 11.5 million.

Thick but possibly useful reading for people:

https://en.wikipedia.org/wiki/Conditional_probability

[Mr Mark]

I love it, so hysterical.  General theme is that 4% rule is rock solid and you should absolutely base your retirement (but always with the caveat to be flexible)....until it is challenged of course.

Question two different scenarios for timing of FIRE....all these hedges come out
...its not a rule
...based on historical data
....its just a model and doesn't mean anything going forward
....you can't control the future
....retiree would or should reduce their withdrawals
...its not realistic to just blindly follow it
...blah blah blah.

All that may be true or not but the fact of the matter is that the 4% rule doesn't go far enough (but still may be valid) and the two retiree scenario posted before would have drastically different withdrawals based on the time they retired (absent adjustments/flexibility/current market knowledge as none that has any factor on WR analysis). 

The scenario presented was the negative view - markets go down, but the inverse is also a problem.....Retiree A goes in x year and retiree B goes one year later but after a 30% return.  Same thing applies Retiree A is living off of $40k (+inflation) and Retiree B is living off of $52k (+inflation) (based on $1mil and 4%WR).  Yet both still have the same probability at the start of the retirement. 

I would be far more willing to accept a higher WR after a downturn or lower rate after a good run...call it logic or intuition or whatever....but then if I do this I am basically conceding that the 4% rule is worthless and should instead focus on my own views (that doesn't seem quite right either).

We have all heard and believe that during the accumulation phase that savings rate is more important than returns....well I think based on all of this entire thread's discussion and discussions elsewhere that we could probably all agree that during retirement that being flexible is far more important than WR, which kind of sucks bc then it diminishes the more data driven approach.

I guess what you are asking is: Is there a correlation between maximum 30 year SWR and the stock market performance the year before FIRE?

[Monkey Uncle]

All that may be true or not but the fact of the matter is that the 4% rule doesn't go far enough (but still may be valid) and the two retiree scenario posted before would have drastically different withdrawals based on the time they retired (absent adjustments/flexibility/current market knowledge as none that has any factor on WR analysis). 

I don't understand your point here. The 4% rule is based on having no knowledge of what the stock market is going to do the year after you retire. Of course, once you've lived through that year, you can make a better estimate of whether your stash is going to last the remaining 29 years than you could before. That's how adding more information works.

But until we have a way to predict what next year's returns are before we get to next year, I'm not sure the above observation is actually helpful to people trying to make decisions about when to FIRE.

The scenario presented was the negative view - markets go down, but the inverse is also a problem.....Retiree A goes in x year and retiree B goes one year later but after a 30% return.  Same thing applies Retiree A is living off of $40k (+inflation) and Retiree B is living off of $52k (+inflation) (based on $1mil and 4%WR).  Yet both still have the same probability at the start of the retirement. 

But this is just another example of "yes, you can improve your predictions over time as you get more data about how your own retirement window has already behaved."

It's like saying "your odds of being attacked by a shark are 1 in 11.5 million" but if you go to the beach, and you're out surfing and you see a shark fin circling around you, at that point your odds of being attacked as significantly higher than 1 in 11.5M. That doesn't make the first estimate wrong, invalid, or misleading. It just means that as time passed and more data accumulates on your risk factors, we can do a better job of estimating the chances that you're going to fall into that 1 in 11.5 million.

Thanks for that, Maizeman.  The concept that "more data changes the prediction" is what I was trying to get at, but you did it much more eloquently.

[EscapeVelocity2020]

Similar to what ExFlyboy asked earlier (a page ago), but with a twist.  What if Retiree A and B have $1M.  Retiree A calls it quits in 2007 and withdraws his inflation adjusted $40k for 30 years.  Retiree B hangs in a little longer only to suffer the ~37% decline in 2008.   Retiree A has roughly 605k and Retiree B has roughly 630k. 

How much can Retiree B plan to spend for the next 30 years?

So is there a consensus on how much Retiree B can spend?  I've heard the strict interpretation of the 4% Rule = $25,200 (inflation adjusted) for the next 30 years and also heard 'something closer to $40,800, since that what Retiree A can spend'.  There is a pretty big difference between those answers, and a whole lot of handwaving around 'being flexible' doesn't really help Retiree B know what to do.

[Retire-Canada]

So is there a consensus on how much Retiree B can spend? 

You'd have to define can more precisely to get a specific answer most people agree on.

[TempusFugit]

Similar to what ExFlyboy asked earlier (a page ago), but with a twist.  What if Retiree A and B have $1M.  Retiree A calls it quits in 2007 and withdraws his inflation adjusted $40k for 30 years.  Retiree B hangs in a little longer only to suffer the ~37% decline in 2008.   Retiree A has roughly 605k and Retiree B has roughly 630k. 

How much can Retiree B plan to spend for the next 30 years?

So is there a consensus on how much Retiree B can spend?  I've heard the strict interpretation of the 4% Rule = $25,200 (inflation adjusted) for the next 30 years and also heard 'something closer to $40,800, since that what Retiree A can spend'.  There is a pretty big difference between those answers, and a whole lot of handwaving around 'being flexible' doesn't really help Retiree B know what to do.

I think the answer in terms of the 4% rule is that *in the past* it has worked out 96% of the time if a person with that amount of stash spent 4% of the balance of that stash each year.   

The 4% rule isn't really a prediction, it is a statement about what has happened before.  Now if you want to use Person A's conditions to determine Person B's spending, the fact is that even Person A's plan isn't looking too great now.  Even though they followed the 4% rule properly, the fact is that even the historical model shows that it does fail sometimes (4% of the time during the historical period).  The new information that we now have - the large market drop one year into Person A's retirement) now means that we have more information to plug into that historical model to see how he *would have done* in the past.  You would have to look at the subset of periods in which a similar market drop occurred right after a person retired. He might still be fine, both in the past model and in the future reality.  It would now be the role of the thinking rational human being to look at all of the available information and re-evaluate. 

The 4% rule is kind of an anchor point that gives us a model of the best information we currently have about how market performance and withdrawal rates ultimately affect a retirement plan.  We can look at all of the historical circumstances such as the great depression, the tech bubble crash, the inflationary 70's, the crappy stock market returns of the late 60's - early 70's and use that knowledge to help us put a confidence level on our historical model. 

No matter what cFireSim tells you when you plug in your numbers, you might still run out of money.  Even if you use 2% instead of 4%, you might run out of money.  Because no one can predict the future. 

But it's a trade-off.  Running out of money is one kind of failure.  Dying one year into your retirement is another kind of failure.  Having a long retirement of worrying about money and not travelling to see your family or not enjoying your life but then leaving a million bucks in the bank when you kick off is another kind of failure. We all have to use our own judgment and our own priorities to determine where we feel comfortable.

[boarder42]

Similar to what ExFlyboy asked earlier (a page ago), but with a twist.  What if Retiree A and B have $1M.  Retiree A calls it quits in 2007 and withdraws his inflation adjusted $40k for 30 years.  Retiree B hangs in a little longer only to suffer the ~37% decline in 2008.   Retiree A has roughly 605k and Retiree B has roughly 630k. 

How much can Retiree B plan to spend for the next 30 years?

So is there a consensus on how much Retiree B can spend?  I've heard the strict interpretation of the 4% Rule = $25,200 (inflation adjusted) for the next 30 years and also heard 'something closer to $40,800, since that what Retiree A can spend'.  There is a pretty big difference between those answers, and a whole lot of handwaving around 'being flexible' doesn't really help Retiree B know what to do.

I think the answer in terms of the 4% rule is that *in the past* it has worked out 96% of the time if a person with that amount of stash spent 4% of the balance of that stash each year.   

The 4% rule isn't really a prediction, it is a statement about what has happened before.  Now if you want to use Person A's conditions to determine Person B's spending, the fact is that even Person A's plan isn't looking too great now.  Even though they followed the 4% rule properly, the fact is that even the historical model shows that it does fail sometimes (4% of the time during the historical period).  The new information that we now have - the large market drop one year into Person A's retirement) now means that we have more information to plug into that historical model to see how he *would have done* in the past.  You would have to look at the subset of periods in which a similar market drop occurred right after a person retired. He might still be fine, both in the past model and in the future reality.  It would now be the role of the thinking rational human being to look at all of the available information and re-evaluate. 

The 4% rule is kind of an anchor point that gives us a model of the best information we currently have about how market performance and withdrawal rates ultimately affect a retirement plan.  We can look at all of the historical circumstances such as the great depression, the tech bubble crash, the inflationary 70's, the crappy stock market returns of the late 60's - early 70's and use that knowledge to help us put a confidence level on our historical model. 

No matter what cFireSim tells you when you plug in your numbers, you might still run out of money.  Even if you use 2% instead of 4%, you might run out of money.  Because no one can predict the future. 

But it's a trade-off.  Running out of money is one kind of failure.  Dying one year into your retirement is another kind of failure.  Having a long retirement of worrying about money and not travelling to see your family or not enjoying your life but then leaving a million bucks in the bank when you kick off is another kind of failure. We all have to use our own judgment and our own priorities to determine where we feel comfortable.

in the given scenario if 100% of assets were invested in VTSAX both could have withdrawn 40k per year adjsuted up at 3% per year for much higher than inflation withdrawals and been left today sitting with 1.3MM and 1.4MM ... so with the added data that has been proposed from the future in this situation it is currently further proving the safety of the 4% swr regardless of the huge hit you took in year one.  when a normal human would likely cut some spending or earn some extra income.

[Exflyboy]

It is interesting to me that as a group we tend to worry about the "running out of money" type of failure. I think we see this kind of thinking because to savers are are wired that way.

Now I have family members who are broke, will never retire and the thought of ROM failure doesn't even enter their heads.... "Oh retirement, that's what that Social Security thing is for right?". Of course they are head directly for a ROM type of failure BEFORE they FIRE. A path I don't want to be on.

The challenge to us worriers (and yes I am one) is to learn to be comfortable spending our allocation after a lifetime of saving. I am attempting to do this by being comfortable blowing 3%.. As in blowing it and not giving a second thought.

Sounds easy, but when it comes to paying for unsubsidised Health care I immediately "optimise" and go for the max subsidy, i.e spend way less.

Its a disease I tell ya..:)

[Anon in Alaska]

Has anyone done the math based on monthly withdrawals? Suppose I retire and withdraw tomorrow. I don't need to draw out 4% right away, I only need to draw out a months worth, or 1/12 of 4% or 0.33%. That would mean that the other 3.67% would be earning for me for another 1-11 months.

[matchewed]

Has anyone done the math based on monthly withdrawals? Suppose I retire and withdraw tomorrow. I don't need to draw out 4% right away, I only need to draw out a months worth, or 1/12 of 4% or 0.33%. That would mean that the other 3.67% would be earning for me for another 1-11 months.

That is a strategy I guess but would generate a great deal of transactions. Also if your plan relies on trying to ensure that everything but what you spend is actively in some market maybe your plan isn't flexible enough.

A much more common strategy would be to have some cash or easily available position that you can access and rely on in a downturn. One which allows you to fill that cash or cash alternative position with the other funds when the market recovers. That position would be based on your risk tolerance and historical recovery periods.

In short I don't think anyone has done that math on such a granular level as given the above I'm not sure it's a solid strategy.

[TempusFugit]

Has anyone done the math based on monthly withdrawals? Suppose I retire and withdraw tomorrow. I don't need to draw out 4% right away, I only need to draw out a months worth, or 1/12 of 4% or 0.33%. That would mean that the other 3.67% would be earning for me for another 1-11 months.

I suspect that if you did this it wouldn't really change anything.  It would all even out. 

As a practical matter, monthly withdrawals would probably be difficult, since things like dividend payments are quarterly or annual.  It would depend upon the investment vehicle in which your money was stashed. 

[SwordGuy]

Has anyone done the math based on monthly withdrawals? Suppose I retire and withdraw tomorrow. I don't need to draw out 4% right away, I only need to draw out a months worth, or 1/12 of 4% or 0.33%. That would mean that the other 3.67% would be earning for me for another 1-11 months.

You are correct.  You will get a difference.  If the market is going up, it will be in your favor.  If down, then not. 

The thing that would gum it up would be transaction costs.   So as long as they aren't odious, all should be well.

Dividends, that someone else mentioned, aren't a problem.  You'll get more of them by having more money in the market longer.

[Anon in Alaska]

Has anyone done the math based on monthly withdrawals? Suppose I retire and withdraw tomorrow. I don't need to draw out 4% right away, I only need to draw out a months worth, or 1/12 of 4% or 0.33%. That would mean that the other 3.67% would be earning for me for another 1-11 months.

That is a strategy I guess but would generate a great deal of transactions. Also if your plan relies on trying to ensure that everything but what you spend is actively in some market maybe your plan isn't flexible enough.

Twelve transactions a year does not seem like a lot to me. I'm currently buying more than 12 times a year; I'm not sure why selling 12 times a year would be any different.

Flexible is good. Earning money for me until I need it is also good.

[Anon in Alaska]

Has anyone done the math based on monthly withdrawals? Suppose I retire and withdraw tomorrow. I don't need to draw out 4% right away, I only need to draw out a months worth, or 1/12 of 4% or 0.33%. That would mean that the other 3.67% would be earning for me for another 1-11 months.

I suspect that if you did this it wouldn't really change anything.  It would all even out. 

Lets say I'm earning 7% per year. If I keep my money invested for an extra 0-12 (mean 6) months than I'm going to earn an extra 3.5% on that 4% a year I'm spending. (The market isn't that smooth of course, I'm simplifying.) Suppose I had $500,000. I retire, withdraw the $20,000 I need for the first year, and at the end of that year I have $513,600 (480,000 x 1.07). If I keep that $20,000 invested until I need it, and draw it out once a month, I'm going to have an extra $700 at the end of the year (0.07/2 x 20,000)). My investments will have earned $34,300, not $33,600. Why wouldn't I want this?

Oh sure if the market is going down I'll be a little worse off but you can't time the market and it goes up more than it goes down, so this should work most of the time, right?

I just need to have my money invested in something where my transactions cost is low enough that it would be less than $700/year for an extra 11 transactions. I do.

[matchewed]

Because a $400 difference in investment return over the course of a year is indistinguishable from noise. It's not about wanting it, the statement was that you probably wouldn't notice a difference. So more effort for an indistinguishable from normal fluctuation gain doesn't seem like something that is a large enough impact to chase.

[TomTX]

I just need to have my money invested in something where my transactions cost is low enough that it would be less than $700/year for an extra 11 transactions. I do.

Yeah, I don't see why transaction cost should be a barrier. Example:

If you have $100k (trailing 3 month average) combined between Merrill Edge and Bank of America - you get 100 free stock or ETF trades every month and zero fees.  Put your money in VTI (or whatever ETFs you want).

Withdrawing monthly costs nothing.

Hell, I could make my withdrawals 3 times a DAY and pay nothing for transaction costs.

[secondcor521]

It's not the transaction costs, it's the hassle factor.  FIRE people get lazy quickly.  But if it's worth the hassle to you, knock yourself out.

I will point out that a quick calculation suggests that moving maybe 2% of that hypothetical $500K from bonds to stocks - i.e., a very slightly riskier asset allocation - gets you that same $700 per year with a lot less hassle.

But yeah, the principle of moving money monthly instead of yearly is sound.  Again, just the hassle factor.

[Livewell]

The challenge to us worriers (and yes I am one) is to learn to be comfortable spending our allocation after a lifetime of saving. I am attempting to do this by being comfortable blowing 3%.. As in blowing it and not giving a second thought.

Sounds easy, but when it comes to paying for unsubsidised Health care I immediately "optimise" and go for the max subsidy, i.e spend way less.

I look at this as an experiment to prove out your risk profile.  Conservative and you’ll want to bake in every possibility  and spend as little as possible.  I think there is a danger in doing this.  Not that saving and planning is bad, they are required, but at some point you have to trust that things will work out. 

For my journey, first it was accepting that my target needed to be higher because I preferred to be married and I needed to compromise with my wife!  That was tough because it’s added years to my career.  However those years have been much better than I thought and after much thought I’m fine with 4.5% because we do spend a decent amount (which could be cut back) and we do live in a HCOL area (and can relocate if needed).

The point is all my initial worrying has proven unnecessary, and while I am not yet FIRE, I can see why it’s likely to be just fine later.  There is a point of over engineering with this stuff, and it’s a good thing to challenge yourself around that every so often.

[EscapeVelocity2020]

Has anyone done the math based on monthly withdrawals? Suppose I retire and withdraw tomorrow. I don't need to draw out 4% right away, I only need to draw out a months worth, or 1/12 of 4% or 0.33%. That would mean that the other 3.67% would be earning for me for another 1-11 months.

That is a strategy I guess but would generate a great deal of transactions. Also if your plan relies on trying to ensure that everything but what you spend is actively in some market maybe your plan isn't flexible enough.

A much more common strategy would be to have some cash or easily available position that you can access and rely on in a downturn. One which allows you to fill that cash or cash alternative position with the other funds when the market recovers. That position would be based on your risk tolerance and historical recovery periods.

In short I don't think anyone has done that math on such a granular level as given the above I'm not sure it's a solid strategy.

OK, I cannot resist.  If your strategy hinges on such a negligible advantage, then you are working way too hard.

Honestly, on a forum like Bogleheads or ER.org, you'd be laughed at for needing to sit in front of your computer so much to squeeze out such a small gain.  And, as others have pointed out, it is not always a straightforward win - it's not like the market is linear and you just do this for the first few years.  In order to benefit from the market average, you need to put in 30 years...  So yes, if it is worth it to you to try to make more transactions and ultimately benefit from the added tax paperwork and playing this game, then go for it.  It neither sounds fun nor educational, so I'll stick to annual withdrawals and enjoy the added free time to peruse optimal health or business opportunities.

Ultimately, as human beings, time is the final limiting factor.  Don't trade it away for vanishing gains.

[Roothy]

Most places will let you automatically withdraw a fixed amount every month, so no need for it to be a hassle.  For an extra few hundred bucks a year, absolutely worth it.  I certainly plan to do this. 

[Roothy]

Has anyone done the math based on monthly withdrawals? Suppose I retire and withdraw tomorrow. I don't need to draw out 4% right away, I only need to draw out a months worth, or 1/12 of 4% or 0.33%. That would mean that the other 3.67% would be earning for me for another 1-11 months.

And as for whether anyone has done this math, earlyretirementnow has: https://earlyretirementnow.com/2017/01/25/the-ultimate-guide-to-safe-withdrawal-rates-part-7-toolbox/

Their version of the Trinity Study not only has withdrawals happening monthly,  but their historical returns are also monthly.

[Classical_Liberal]

Regarding monthly withdrawals.

There are entire threads on this forum devoted to making a few hundred a year on CC bonuses.  Some people here enjoy this type of thing, its what differentiates MMM from Bogleheads.  So I see no reason to criticize this approach for that particular reason.

Essentially its the same as lump sum vs dollar cost averaging.  Lump sum is better, because equities go up more often than they go down.  I can see this strategy increasing wealth in the "good times", maybe working OK in period of high inflation, but likely working in reverse for periods of fast, temporary, 1-3 year, down turns. 

I think its important to ask a question though.  Is the rate of portfolio failures the most important thing to consider in withdrawal strategy.  Example(s), a strategy with a 10% failure rate, in which half of those failures occur more than 10 years before predicted life expectancy.  A withdraw strategy with 15% failure, but there are no failures before 2 years of projected life expectancy.  Which is the stronger strategy? 

IMO It's just as important to see how (and by how much) a portfolio & WR fails, as it is to see how often is fails.

[boarder42]

Imo a plan failing within 2 years of 40 or 10 years of 40 isn't relevant bc most will need 50-80+ years of withdrawals.

[Classical_Liberal]

Imo a plan failing within 2 years of 40 or 10 years of 40 isn't relevant bc most will need 50-80+ years of withdrawals.

I didn't specify 40 years, but in fairness, it is in the general range I was thinking.  While I appreciate your optimism; the reality of life expediencies are a bit different.  Unless someone FIRE's several years before conception an 80yr span is probably not needed, nor is it a statistically significant difference from a perpetual WR.

[boarder42]

Imo a plan failing within 2 years of 40 or 10 years of 40 isn't relevant bc most will need 50-80+ years of withdrawals.

I didn't specify 40 years, but in fairness, it is in the general range I was thinking.  While I appreciate your optimism; the reality of life expediencies are a bit different.  Unless someone FIRE's several years before conception an 80yr span is probably not needed, nor is it a statistically significant difference from a perpetual WR.

Correct. 40 years is greatly determined to mean money should last bear indefinitely if it makes it there. But my point was. Why do I care if I ran out at 30 years vs 38 years it still failed and didn't support me and I had to change my plan to make something else happen.

And I think it's shortsighted to plan based on current life expectancy. They keep rising at least for the healthy non drug users.

[Classical_Liberal]

Correct. 40 years is greatly determined to mean money should last bear indefinitely if it makes it there. But my point was. Why do I care if I ran out at 30 years vs 38 years it still failed and didn't support me and I had to change my plan to make something else happen.

Good question!

If all of my historical failures are only coming up short 10-50K, I'm much more confident in my ability to make minor course corrections to rectify minor short falls as I see them materializing.  If I have several failures coming up multiple hundreds of thousands short, it becomes more difficult to correct. I would also argue that it becomes increasingly difficult to correct for shortfalls(at least through income) after a certain age.

There is a notable difference between 30-40 years and perpetual, almost nonexistent at 80 years vs perpetual.

And I think it's shortsighted to plan based on current life expectancy. They keep rising at least for the healthy non drug users.

Touche.  However, keep in mind, making it to 80-90 is lucky (or skilled) enough.  Making to that age and still enjoying a reasonable active and healthy life, much less likely.  Everyone has an anecdote of their great aunt who is 100 and still walks miles to the grocery store every day; but that is a far, far outlier.  I hope its you and I, but it probably wont be, so I'm not willing to risk quality life now on that assumption.  Much like the likelihood of dying vs running out of money graphs up thread show.