The Day the 4% Rule Died

[McQ]

Bengen, in his 1994 paper, found that a balanced portfolio of stocks and bonds could support a withdrawal rate of 4%, adjusted each year for inflation. By “support,” he meant sustain those withdrawals for thirty years or more. https://retailinvestor.org/pdf/Bengen1.pdf

That paper, along with the later Trinity study (https://www.aaii.com/files/pdf/6794_ret ... inable.pdf), launched the Sustainable Withdrawal Rate literature, followed by the Variable Withdrawal Rate literature (see the wiki, and countless threads here at BH; most recently, and still live, this one: viewtopic.php?t=379159).

It’s not always recognized that Bengen only had 38 thirty-year rolls to work with (start dates 1926 – 1964 inclusive), nor that, for the bond portion, he used what today would be called a short Treasury, far from the Total Bond Index favored here at BH.org.

In any case, as we throw up our hands in horror here in mid-2022, I thought I’d try my hand at a little speculative fiction: Will 2021 be the start-year where the 4% rule goes down? Goal here is to construct a plausible scenario.

Assumptions

[you might want to glance at the spreadsheet image first]
1. Mr. Erstwhile takes a 4% withdrawal at the end of 2021, leaving $960,000 in the account at the start of 2022. Next withdrawal due at the end of 2022 and so forth.
2. Inflation at 8% in 2022, 6% in 2023, and then reverts to 3% (the US 90-year average).
3. Stocks suffer an ordinary severe bear market on the order of 1937-38, 1973-1974, and 2000-2002, aka go down not quite 50% in all: specifically, down 30% by the end of 2022, and down another 25% in 2023.
4. But the total bond index suffers a never-before-seen drawdown sequence: down 15% by the end of 2022 (getting there ), and down another 10% in 2023. After all, there had never been a 40-year bond bull market with sustained double-digit returns before, so … how bad might the aftermath be?
5. There’s a bounce for both assets in 2024 and 2025, as shown in the spreadsheet below.
6. But also a hiccup in 2030, call it the Second Ukrainian war, and then again a bounce, see the spreadsheet.
7. In all other years, stocks and bond returns revert to the post-1900 World average rather than to US long-term averages. For stocks, 4.3% real + 3% inflation = 7.3% nominal; and for bonds, 1.5% real (total bond adjustment of -0.5%, historical data were based on long bonds) + 3% = 4.5% nominal. [Credit Suisse / Dimson et al. data]

#7 may be the crucial departure: I assumed depressed stock market returns relative to US history, to reflect all the discussion one hears about how the 2009-2021 bull market borrowed returns from the future. So, payback time. This speculative fiction still assumes a substantial real return on stocks, and still produces an equity premium on the order of +3%, but again, the stock series is much lower than the returns historically seen in the US. However, the bond series is not depressed relative to the 1926 – 1980 US experience.



Results

Alas, Mr. Erstwhile succumbs to sequence-of returns risk and he runs out of money in the middle of 2040, after less than twenty years. Bye bye Bengen rule.

Plausible? I await your replies.

[You should be able to recreate the spreadsheet with a few keystrokes using Row #2 as a guide and run whatever alternative scenario you like.]

Homework: this thread by Nisiprius, aka “Bengen-retest-using-after-cost-mutual-funds-you-could-actually-have-bought.” It didn’t go so well. https://www.bogleheads.org/wiki/User:Ni ... tual_funds

Or my paper, “Bengen taken overseas and back in time.” Yikes. https://papers.ssrn.com/sol3/papers.cfm ... id=4001986

[Marseille07]

I think they make a point, but I don't like that random 6% inflation in 2030. It seems to fabricate a narrative they want to deliver.

[David Jay]

Your spreadsheet is an excellent illustration of Sequence of Returns. Significant market drops in the early years of retirement while maintaining inflation-adjusted withdrawals.

[Dottie57]

Why did Mr Erstwhile not start spending less?

[BigJohn]

I surely can’t say that the scenario is implausible but Begen didn’t get a 100% survival rate at 4% either. So a question, in your scenario what SWR allows Mr E to get to 30 years without a negative balance?

[afan]

Curious. If inflation goes down from 2022 to 2023 in your hypothetical, why do bonds take another big drop in '23?

In any case, agree that constructing a plan based on the highest withdrawals that did not fail in a relatively short time in one country while it was the dominant economy in the world may be optimistic.

Part of the reason I am still working.

[LilyFleur]

It is unfortunate that both stock index funds and total bond index funds are down at the same time.

For those of us who moved some of our bond index funds into our stable value funds, and withdraw from the stable value fund for 5 or 6 years into retirement, how would that affect the success of the plan?

[willthrill81]

I surely can’t say that the scenario is implausible but Begen didn’t get a 100% survival rate at 4% either.

Yes, he did. Bengen didn't find that the '4% rule' failed in any historic 30 year periods.

[SeasOfCheese]

If the market gets a 50% haircut in the first two years after you retire it probably is a good idea to moderate spending for a bit.

Maybe take up guitar instead of multi-engine aviation for a hobby.

[BigJohn]

I surely can’t say that the scenario is implausible but Begen didn’t get a 100% survival rate at 4% either.

Yes, he did. Bengen didn't find that the '4% rule' failed in any historic 30 year periods.

Hmmm… I thought it was 3.8% that never failed. Thanks for the correction.

[willthrill81]

I surely can’t say that the scenario is implausible but Begen didn’t get a 100% survival rate at 4% either.

Yes, he did. Bengen didn't find that the '4% rule' failed in any historic 30 year periods.

Hmmm… I thought it was 3.8% that never failed. Thanks for the correction.

It depends greatly on which data source is used. Bengen used different sources from the Trinity study's authors, which is why they came to slightly different conclusions.

[59Gibson]

Why did Mr Erstwhile not start spending less?

Right what happens if there is just a 5-10% cut, certainly anyone actually using and aware of SWRs could survive such a small lifestyle decrease.

[hnd]

I've definitely not thought about this in any real sense but most of these simulations have like hundreds of outcomes...this is some assumed outcome probably utilizing some made up scenarios where its half average of something, and half made up events. Its just tough to consider it.

[BigJohn]

I surely can’t say that the scenario is implausible but Begen didn’t get a 100% survival rate at 4% either.

Yes, he did. Bengen didn't find that the '4% rule' failed in any historic 30 year periods.

Hmmm… I thought it was 3.8% that never failed. Thanks for the correction.

It depends greatly on which data source is used. Bengen used different sources from the Trinity study's authors, which is why they came to slightly different conclusions.

Thanks. That’s why I asked my question. If 4% fails but 3.8% makes it 30 years I’m not sure I’d say it was dead…. maybe in ICU on life support, but not dead yet

[willthrill81]

All that has been done is to show that it's not impossible for the '4% rule' to fail. That's not news to BHs.

However, absolutely zero credence whatsoever should be applied to such make-believe data.

We don't have to use fictitious data to see that the '4% rule' can fail. Investors in many developed nations who only invested in their own home country would have seen the '4% rule' fail at some point in the last 50 years. But investors in nearly all developed nations who diversified their stock holdings globally would have generally seen the '4% rule' succeed, and only rarely would the 30 year SWR have been below about 3.5%.

[abc132]

It is certainly possible. The market generally prices at the events expected to happen and then reacts to what actually happens. There is nothing sacred about past historical returns that prevents the possibility of better or worse sequences in the future.

There are three major solutions:
1) ability to spend less
2) wait to retire until after a poor sequence (2 negative years in a row)
3) save past 25x expenses

The ability to do #1 means separating wants from needs and understanding expenses. Those at 25x expenses today might consider working until the end of 2023 if real returns stay down in 2022 and 2023. Many here accomplished #3 on accident and blew past 25x. They should be just fine at less than 4% withdrawal rates.

[arcticpineapplecorp.]

the trinity study has shown a failure rate of 5% for a 30 year period for a balanced portfolio:



https://www.bogleheads.org/wiki/Safe_withdrawal_rates

so i'd go with the more conservative estimates.

I'm also concerned about that one year diversion with 6% inflation in 2030.

guess this is why a Variable Percentage Withdrawal instead or some other dynamic method like Vanguard recommends using a ceiling and floor approach is more "guaranteed":

https://investor.vanguard.com/investor- ... ithdrawals
https://advisors.vanguard.com/insights/ ... ketworries
https://www.cbsnews.com/news/vanguard-p ... -paycheck/

[abc132]

A couple of points of discussion.

1) The SWR is just under 3% for this scenario ($29850 or 2.985%)
2) Working two more years and saving $20,000 each of those two additional years resulted in almost 4% SWR ($38540 or 3.854%). This was using the 1 million as the reference, and was a 29% increase in standard of living.

Note that to get to the desired 40,000 you would have had to save/generate and extra 300,000 prior to the crash as compared to working two extra years to add 120,000 (80,000 less in expenses and 40,000 invested). Working through the downturn is extremely valuable.

[nigel_ht]

Yes, it’s implausible (unlikely).

There’s no reason for the US market to revert to world average vs US long term average at this time.

If you want to postulate the US losing world superpower status and reserve currency in the near future that’s fine but that is also implausible.

[FiveK]

...the total bond index suffers a never-before-seen drawdown sequence....
...In all other years, stocks and bond returns revert to the post-1900 World average rather than to US long-term averages.

Sure, if the future is worse than the worst of the past then a guideline based on the worst of the past might not be appropriate.

If the future isn't worse than the worst of the past, then a guideline based on the worst of the past will have been appropriate.

[Cecelio]

If Mr. Erstwhile were 100% stock, he would have entered 2040 with $37,000 less. Thus failing in the first month of 2040.
100% stock: 5.12% avg annual return.
50/50: 3.95% avg annual return.
The large percentage drops in equities in the first 2 years were too great to overcome.

[marcopolo]

Bengen, in his 1994 paper, found that a balanced portfolio of stocks and bonds could support a withdrawal rate of 4%, adjusted each year for inflation. By “support,” he meant sustain those withdrawals for thirty years or more. https://retailinvestor.org/pdf/Bengen1.pdf

That paper, along with the later Trinity study (https://www.aaii.com/files/pdf/6794_ret ... inable.pdf), launched the Sustainable Withdrawal Rate literature, followed by the Variable Withdrawal Rate literature (see the wiki, and countless threads here at BH; most recently, and still live, this one: viewtopic.php?t=379159).

It’s not always recognized that Bengen only had 38 thirty-year rolls to work with (start dates 1926 – 1964 inclusive), nor that, for the bond portion, he used what today would be called a short Treasury, far from the Total Bond Index favored here at BH.org.

In any case, as we throw up our hands in horror here in mid-2022, I thought I’d try my hand at a little speculative fiction: Will 2021 be the start-year where the 4% rule goes down? Goal here is to construct a plausible scenario.

Assumptions

[you might want to glance at the spreadsheet image first]
1. Mr. Erstwhile takes a 4% withdrawal at the end of 2021, leaving $960,000 in the account at the start of 2022. Next withdrawal due at the end of 2022 and so forth.
2. Inflation at 8% in 2022, 6% in 2023, and then reverts to 3% (the US 90-year average).
3. Stocks suffer an ordinary severe bear market on the order of 1937-38, 1973-1974, and 2000-2002, aka go down not quite 50% in all: specifically, down 30% by the end of 2022, and down another 25% in 2023.
4. But the total bond index suffers a never-before-seen drawdown sequence: down 15% by the end of 2022 (getting there ), and down another 10% in 2023. After all, there had never been a 40-year bond bull market with sustained double-digit returns before, so … how bad might the aftermath be?
5. There’s a bounce for both assets in 2024 and 2025, as shown in the spreadsheet below.
6. But also a hiccup in 2030, call it the Second Ukrainian war, and then again a bounce, see the spreadsheet.
7. In all other years, stocks and bond returns revert to the post-1900 World average rather than to US long-term averages. For stocks, 4.3% real + 3% inflation = 7.3% nominal; and for bonds, 1.5% real (total bond adjustment of -0.5%, historical data were based on long bonds) + 3% = 4.5% nominal. [Credit Suisse / Dimson et al. data]

#7 may be the crucial departure: I assumed depressed stock market returns relative to US history, to reflect all the discussion one hears about how the 2009-2021 bull market borrowed returns from the future. So, payback time. This speculative fiction still assumes a substantial real return on stocks, and still produces an equity premium on the order of +3%, but again, the stock series is much lower than the returns historically seen in the US. However, the bond series is not depressed relative to the 1926 – 1980 US experience.



Results

Alas, Mr. Erstwhile succumbs to sequence-of returns risk and he runs out of money in the middle of 2040, after less than twenty years. Bye bye Bengen rule.

Plausible? I await your replies.

[You should be able to recreate the spreadsheet with a few keystrokes using Row #2 as a guide and run whatever alternative scenario you like.]

Homework: this thread by Nisiprius, aka “Bengen-retest-using-after-cost-mutual-funds-you-could-actually-have-bought.” It didn’t go so well. https://www.bogleheads.org/wiki/User:Ni ... tual_funds

Or my paper, “Bengen taken overseas and back in time.” Yikes. https://papers.ssrn.com/sol3/papers.cfm ... id=4001986

It is certainly possible.

But, item #3 and #7 seem to be double counting the not-unreasonable assumption that gains will be muted after the bug run up of the past decade. It is not clear to me how realistic it is that we have a 50% drop followed by decades of historically low returns.

[FactualFran]

It depends greatly on which data source is used. Bengen used different sources from the Trinity study's authors, which is why they came to slightly different conclusions.

Bengen used data in from the "Stocks, Bonds, Bills and Inflation: 1992 Yearbook". The Trinity study used data in the "Stocks, Bonds, Bills, and Inflation, 1996 Yearbook". The yearbooks for different years almost certainly had consistent data up to 1992.

What differed was the bond returns they used. Bengen used the returns of Intermediate-Term Government bonds while the Trinity study used the returns of Long-Term Corporate bonds.

For the years after 1992, Bengen used the average of the data in the yearbook. Using the actual data for the years after 1992 does not change the result that 50:50 and 75:25 stock:bond portfolios supported at least 33 years of inflation-adjusted withdrawals when the initial withdrawal rate was 4%.

[heyyou]

What retiree would always calculate then take every inflation boost, while never ever looking at how many multiples of that annual spending, are left in the portfolio?

[Marseille07]

What retiree would always calculate then take every inflation boost, while never ever looking at how many multiples of that annual spending, are left in the portfolio?

Those who want stable & COLA'ed withdrawals.

If you adjust withdrawals based on the portfolio balance, your withdrawals are no longer stable or COLA'ed.

[StevieG72]

What retiree would always calculate then take every inflation boost, while never ever looking at how many multiples of that annual spending, are left in the portfolio?

Those who want stable & COLA'ed withdrawals.

If you adjust withdrawals based on the portfolio balance, your withdrawals are no longer stable or COLA'ed.

They are diet COLA’ed.

[Marseille07]

They are diet COLA’ed.

That's a good one

[HomerJ]

4% hasn't died yet.

You just made up a bunch of junk.

If it happens, THEN you can say 4% died back in 2021-2022.

[Bfwolf]

4% hasn't died yet.

You just made up a bunch of junk.

If it happens, THEN you can say 4% died back in 2021-2022.

Exactly right. A plausible scenario where I win the lottery is I pick 12, 18, 22, 30, 34, and 49 and the numbers are 12, 18, 22, 30, 34, and 49. That doesn't mean it's something I should make life decisions on or spend any time thinking about--it's very unlikely.

[Lewd comment removed by moderator oldcomputerguy]

[JoeRetire]

I thought I’d try my hand at a little speculative fiction: Will 2021 be the start-year where the 4% rule goes down?

Plausible? I await your replies.

Fiction isn't really my thing. But anything can happen. Seems unlikely to me.

If you want to make it more plausible, don't posit a constant-except-for-2-years 3% inflation rate. That's exceedingly unlikely.

Oh, and I think you forgot to include the zombie apocalypse.

[JDave]

The 4% Rule has become the equivalent of Middle Ages debates over how many angels could dance on the head of a pin. Variable withdrawals solves the problem, but few people want to talk about them.

[nigel_ht]

The 4% Rule has become the equivalent of Middle Ages debates over how many angels could dance on the head of a pin. Variable withdrawals solves the problem, but few people want to talk about them.

Yeah, there’s only a bunch of threads about variable withdrawals and they are generally mentioned every time someone brings up SWR.

Variable withdrawal simply pushes the risk around, not solve it. You either can run out of money at the end of life or have periods where your portfolio doesn’t generate enough income to meet minimum expenses.

For any given required income then constant dollar is always the lowest risk because it’s also the lowest withdrawal method. You are always ONLY withdrawing the bare minimum required to meet your needs.

The trade off is generally you end up with a huge portfolio at the end.

You do not want to do variable withdrawals if your portfolio barely generates the income required to cover minimum expenses under SWR (ie your requires WR = SWR). That’s folks doing lean FIRE or didn’t manage to save a lot for retirement.

In any sequence of returns where SWR runs out of money or even gets close to running out of money will likely result in a dog food retirement.

Why? Because you are living above your means (what your portfolio can generate in the worst case) in the hopes you have a good sequence.

[firebirdparts]

To be honest, this has already been stated 1000 times and nobody really cares. We live in such a tornado of really cheap publishing. I guess it's fair to say that the 4% rule was stillborn. It's been dead a long time. Or not. There never was a consensus.

The big question always was "how much money do I need to retire" and they simply tried to answer that simple question with a very solid effort. It's very subjective, for a variety of reasons:
1. There never was a 4% rule outside the USA and it's obvious there never was.
2. Some people think there are "black swans" every year
3. Bank interest in 1980 was 16% and that's what makes the 4% rule the 4% rule to start with.
4. The Fed Put
5. Dividends are smaller now (don't troll me)
6. We got social security anyway
7. The truth is, once you retire, you'll make do.

People work pretty hard to make the 4% rule something that it's not.

[Leesbro63]

To be honest, this has already been stated 1000 times and nobody really cares. We live in such a tornado of really cheap publishing. I guess it's fair to say that the 4% rule was stillborn. It's been dead a long time. Or not. There never was a consensus.

The big question always was "how much money do I need to retire" and they simply tried to answer that simple question with a very solid effort. It's very subjective, for a variety of reasons:
1. There never was a 4% rule outside the USA and it's obvious there never was.
2. Some people think there are "black swans" every year
3. Bank interest in 1980 was 16% and that's what makes the 4% rule the 4% rule to start with.
4. The Fed Put
5. Dividends are smaller now (don't troll me)
6. We got social security anyway
7. The truth is, once you retire, you'll make do.

People work pretty hard to make the 4% rule something that it's not.

Some good points here, but I take issue with number 3. I believe the 4% still worked before the "big interest" years started factoring in. And "big interest" really isn't a good metric to look at. The correct metric would be REAL interest which considers inflation. The late 1970s/early 1980s had negative interest rates, like today. So the "sticker" value of 16% interest doesn't mean what you imply.

I do agree that no one I've ever heard of retires on 4% and blindly takes inflation raises each year. In fact, 12/31/21 retirees will be interesting to follow. Will they use the 12/31/21 portfolio value? Or will they re-evaluate in light of the 20%ish haircut most of us have experienced in the short time since then?

[HootingSloth]

Variable withdrawal simply pushes the risk around, not solve it. You either can run out of money at the end of life or have periods where your portfolio doesn’t generate enough income to meet minimum expenses.

As much as people don't want to admit it, stocks and bonds (possibly excepting duration-matched TIPS) are risk assets. No method of any kind will ever allow you to guarantee a minimum stream of future consumption from a portfolio made up of risk assets, regardless of the time period, country, diversification, market timing strategy, or whatever other bit of cleverness you want to employ.

We have essentially two choices of strategy that we can combine in various ways. First, we can choose to be flexible. For everyone, this means ditching ideas of "minimum spending," "maximum tolerable loss," backtesting-based "worst case" scenarios derived from ignoring actual history, and other such concepts that we use to trick ourselves into thinking that the future is somehow preordained. For those retiring early, it also means recognizing that a return to work in the future may be needed. Variable withdrawal methods are helpful in implementing this kind of flexibility as part of an overall plan.

Second, we can seek out the best (but still unavoidably imperfect) guarantees that are available to us, rather than depending solely on risk assets. If I own a paid-off, well-insured home, I have a pretty good guarantee of a stream of housing consumption. If I have a duration-matched TIPS ladder, I have a pretty good guarantee of a stream of consumption for a set period of years. If I have Social Security benefits and/or a COLA pension, I have a pretty good guarantee of a lifelong stream of consumption. If I have a SPIA, I have a pretty good guarantee of a lifelong stream of nominal spending (but not necessarily consumption). Etc.

These choices may not seem particularly desirable, but they are the ones that are based on the realities of life, rather than being based either on backtesting that selectively ignores those segments of history that produce results we do not want to face or on wishful thinking that, if we are just a little bit more clever, then maybe we can know what the future has in store.

[nigel_ht]

To be honest, this has already been stated 1000 times and nobody really cares. We live in such a tornado of really cheap publishing. I guess it's fair to say that the 4% rule was stillborn. It's been dead a long time. Or not. There never was a consensus.

The big question always was "how much money do I need to retire" and they simply tried to answer that simple question with a very solid effort. It's very subjective, for a variety of reasons:
1. There never was a 4% rule outside the USA and it's obvious there never was.
2. Some people think there are "black swans" every year
3. Bank interest in 1980 was 16% and that's what makes the 4% rule the 4% rule to start with.
4. The Fed Put
5. Dividends are smaller now (don't troll me)
6. We got social security anyway
7. The truth is, once you retire, you'll make do.

Lol…that brings to mind a Luke Skywalker quote but I digress.

1. SWR is calculated for each country/market. So yes, SWR is not 4% for say the EU or France. However if you hold total world market it’s going to be a lot more uniform…because it’s the total world market…
2. Obviously not and has nothing to do with SWR anyway.
3. Nope.
4. Nope.
5. Dividends vary across the historic periods.
6. You can compute SWR both with and without Social Security.
7. The objective of saving and investing is to minimize the amount of “making do” required.

People work pretty hard to make the 4% rule something that it's not.

Most folks make it out to be a useful rule of thumb for retirement planning and a good enough withdrawal strategy if you like simple and have enough to spend under 4% and find everything you want to do…

Which is pretty much what it is…

I suppose some folks like to make it a strawman…

[nisiprius]

The authors of the Trinity study wrote:

The word planning is emphasized because of the great uncertainties in the stock and bond markets. Mid-course corrections likely will be required, with the actual dollar amounts withdrawn adjusted downward or upward relative to the plan. The investor needs to keep in mind that selection of a withdrawal rate is not a matter of contract but rather a matter of planning.

That's more important than any of the numbers in the paper.

4% is a reasonable rule of thumb, and much better than numbers that had been commonly bruited about--which blindly assumed--without apparently testing--that historical average return = safe withdrawal rate.

As late as 1995, Peter Lynch--saying that "Fidelity's Bob Beckwith ran the numbers"--was claiming 7% for a portfolio of 100% stocks:

You could do this the easy way and invest in an S&P 500 index fund, currently yielding about 3 percent [dividends]. Or you could select a few "dividend achievers," as identified by Moody's.

There are a couple of basic problems with all of these studies. The first is that the determination of what is an acceptable failure rate is very important, and rarely discussed. Commonly, a 5% failure rate is taken to be acceptable, for no particular reason other than it sounds plausible if someone proposes it. It's even framed in optimistic terms, glass 95% full instead of 5% empty, i.e. as "95% success."

People calculate the SWR to two-place accuracy and think it's a big deal if some system gets 4.2% instead of 3.8%. Then they handwave the question of how accurate that 5% failure probability number is, if taken as a forecast for the future. And then they handwave the operational meaning and the consequences to a retiree if the 5% scenario turns up.

Note that 5% isn't any "black swan," its larger than the chances of rolling a 2 on a pair of dice.

Sure, we all agree that retirees won't literally spend themselves down to zero, but those "course corrections" the Trinity authors mentioned... how do you decide that it looks like you might be in one of those 5% oops-rolled-a-two scenarios, and what do you do? How much do you cut back, and when?

[Leesbro63]

Reply to Nisiprius, just above: I think what you are pointing out is that financial planning is both an art and a science/math. We Bogleheads like mechanical, math based rules to follow. But there are times when the "art" of financial planning is required. A few unresolved such issues are the idea of rebalancing into a long declining market. And the problem of people retiring at the end of 2007 ending up, a year later, with a much higher withdrawal rate than the end of 2008 retiree. (Looks like 2021 retirees will have the same issue). I've been thinking of starting a thread about such issues...problems that we have that can't be resolved with simple math based rules.

Anyway, good post Nisiprius. Investing is both an art and a science/math.

[firebirdparts]

Some good points here, but I take issue with number 3. I believe the 4% still worked before the "big interest" years started factoring in. And "big interest" really isn't a good metric to look at.

This is a good example of why I think it's hard to talk about this: People making the 4% "rule" something it's not. When you say the 4% rule "worked" in cases which weren't the worst case, it just shows that we're not seeing the actual meaning. Nobody should feel that there is additional information in the fact that it works in the not-worst-case. There are bad times and there are good times. They used the word "safe" because you don't know in the moment that you are living the worst case.

Now, if all the cases (not the worst case) interests you, I like this: I have a graph here that shows just how often the withdrawal rate is actually close to 4% and how often it's not, and how high it gets. (The usual assumptions here; 30 years, you die penniless) It's close to 4% a lot. I find that interesting. It's from an article on Advisor Perspectives called "Does International Diversification Improve Safe Withdrawal Rates" and it's a pretty good article but it has some format problems which you have to sort to appreciate the tables.

[oldcomputerguy]

Topic locked for moderator review

[LadyGeek]

Upon further review, this thread discusses actionable investing theory and is unlocked to continue the discussion.

[seajay]

"The [SWR] code is more what you'd call 'guidelines' than actual rules." -Hector Barbossa.

4% with linear inflation pacing results lasts 25 years, less if a bad sequence of returns, more if a good sequence of returns. A 65 year old retiree might under bad circumstances last less than 25 years, more under good circumstances. The odds are generally favorable, assuming equalization just 25% chance of good mortality, poor investment sequence of returns. In the more extreme case money might last only 15 years, see just 60% of its initial purchase power having been available, lost out to inflation, whilst for a lucky individual (subjectively) their mortality might see a 65 year old retiree living a further 40 years.

Actionable - if you endure a poor earlier years sequence of returns then smoke/drink/worry about it .. and your mortality will likely be realigned to still being a successful outcome

[AnnikaW]

Welp, after reading Bogleheads the past day or two (Japan, rampant inflation, Great Depression 2.0, etc), I guess 35+ years of savings might not matter after all. Kudos to people who never bothered to save at all for retirement. It's quite possible we will all end up in the same place....social security and gruel.

[Leesbro63]

Actionable - if you endure a poor earlier years sequence of returns then smoke/drink/worry about it .. and your mortality will likely be realigned to still being a successful outcome

+1. Great point!

[Leesbro63]

Welp, after reading Bogleheads the past day or two (Japan, rampant inflation, Great Depression 2.0, etc), I guess 35+ years of savings might not matter after all. Kudos to people who never bothered to save at all for retirement. It's quite possible we will all end up in the same place....social security and gruel.

I have days like that too, where I think the big spenders in my life will get the last laugh. Hopefully not.

[AnnikaW]

Welp, after reading Bogleheads the past day or two (Japan, rampant inflation, Great Depression 2.0, etc), I guess 35+ years of savings might not matter after all. Kudos to people who never bothered to save at all for retirement. It's quite possible we will all end up in the same place....social security and gruel.

I have days like that too, where I think the big spenders in my life will get the last laugh. Hopefully not.

Sometimes, you have to wonder.

I am starting to think we may not be able to draw down our retirement for fear of Armageddon.

[nigel_ht]

Welp, after reading Bogleheads the past day or two (Japan, rampant inflation, Great Depression 2.0, etc), I guess 35+ years of savings might not matter after all. Kudos to people who never bothered to save at all for retirement. It's quite possible we will all end up in the same place....social security and gruel.

I have days like that too, where I think the big spenders in my life will get the last laugh. Hopefully not.

Between inflation and losses I shoulda spent more last year

[AnnikaW]

Welp, after reading Bogleheads the past day or two (Japan, rampant inflation, Great Depression 2.0, etc), I guess 35+ years of savings might not matter after all. Kudos to people who never bothered to save at all for retirement. It's quite possible we will all end up in the same place....social security and gruel.

I have days like that too, where I think the big spenders in my life will get the last laugh. Hopefully not.

Between inflation and losses I shoulda spent more last year

You and me both.

[McQ]

[This post designed in part to bolster moderators' confidence in the decision to unlock the thread]

First, what can be the point of this sort of speculative future history?

Answer: to reveal the contingencies that have underwritten the success of the 4% rule historically in the US—by systematically violating these implicit rules, while at the same time, maintaining a veneer of plausibility as to projected returns.

To that end, on this second pass I got rid of the Second Ukrainian war. That takes Mr. Erstwhile out one more year, to 2041, still a substantial failure for the 4% rule. I also added the actual 2017-2021 returns, for context, and calculated the 30-year annualized values, 2017 – 2047, to help the reader decide whether plausibility has been maintained.

Here is the revised spreadsheet image.



The annualized values are not implausible, I believe. Yes, the real stock return is well under 1.0 Siegels (a Siegel equals 6.6%, the purported central tendency of real stock returns over the long run), but it is within the range of 30-year rolls in the US over the past two centuries. Likewise, the real bond return is quite low, but rather better than some periods between 1946 and 1980 in the US, plus the nominal return is better than anything recently seen in the US (before 6-13-2022, that is). And inflation is less than or equal to that seen in the US for every 30-year period with a start date between about 1945 and 1980.

No, the revealed contingencies lie elsewhere. I see two:
1. Stocks and bonds shall not tumble, hard, at the same time;
2. If stocks do take a tumble, then a few years later they must roar back with outsize gains.

The effect of the first is obvious; let me bring out the second by making one more change to the spreadsheet. As you may recall, after the tumble in 2007-08, stocks bounced back, as shown in 2024-2025 here; but beginning in 2013, they proceeded to roar ahead. In this next iteration, I drop in the actual 2013 - 2021 returns on VFIAX, beginning five years after the conclusion of the drop (gray highlight).



Et voila: Mr. Erstwhile now makes it to 2050, thirty years more or less. And if I ameliorated the bond decline in 2022-2023 (1st contingency), he’d be home free. Real stock returns are now almost exactly equal to 1.0 Siegels, and the nominal stock return is the customary 10%.

Now you know what it takes for the 4% rule to prevail, as it always has in post-1926 US history thus far, because in that history:
1. Stocks and bonds seldom tumble together
2. Stocks always roar back after taking a tumble
3. US stocks generally maintain a real return of 1.0 Siegels or better

Speculative history aside, the 4% rule should continue to work so long as those conditions are maintained.

PS: if your takeaway was something like, “Forget the quest for an SWR , I am going with a VWR framework,” I think that’s a very appropriate conclusion to draw.

[Marseille07]

4% rule for 60/40 may be dead, but the issue is that we don't know it is dead.

It's kind of like the Bengen study which didn't have 1966 and later discovered it failed. If 2022 were like 1966, we're like June of 1966; no way to know it'd fail 29.5 years later.