"4% rule" withdrawals from real mutual funds

[nisiprius]

Inspired by a book* on taking retirement withdrawals using the Vanguard Wellington Fund as the portfolio, I decided to look at how well several mutual funds with long histories would have fared using "4% rule" withdrawals.

Disclaimers: 1) The "4% rule" is what it is, I'm not recommending it as an actual system to follow literally. 2) All the funds that failed, failed at starting dates in the mid-1960s when inflation took off and the "death of equities" occurred. Not only is past performance not indicative of future results, but past performance in one particular narrow period of time is even less indicative. 3) These funds might have lower expense ratios now. 4) Therefore, my idiosyncratic analysis doesn't tell you much useful about the future of these funds.

"4% rule:" I assumed a starting portfolio of $100,000, and a $4,000 withdrawal at the start of the first year of retirement. In subsequent years, the withdrawal is kept at "$4,000 real," i.e. $4,000 adjusted for inflation. The time frame is taken to be thirty years. The regime "succeeds" if it is possible to take thirty $4,000-real annual withdrawals and have money left at the end, and fails if it runs out of money before making thirty full payments.

"Shortfall:" When the portfolio fails, I measure shortfall in years giving prorated partial credit for whatever withdrawal is possible in the last year. That is, if the 26th scheduled payment was supposed to be $14,000, but there is only $7,000 left in the portfolio, I count that as enough for 25.5 payments and score it as a "4.5-year shortfall."

The red bars descending from the top indicate shortfalls. The blue bars at the bottom indicate the balance of the portfolio at the end of thirty years.

I looked at five actual balanced funds; the Fidelity Puritan Fund, FPURX; George Putnam Balanced Fund, PGEYX; Dodge & Cox Balanced Fund, DODBX; T. Rowe Price Balanced Fund, RPBAX; and the Vanguard Wellington Fund, VWELX. I looked at a theoretical 60/40 portfolio based on the Ibbotson Associates data for the S&P 500 and predecessors for stocks, and intermediate-term government bonds; and finally, for 100% stock fans, the Massachusetts Investors' Trust (MITTX) fund, which is all stocks. I think I will just display the results without comment. In order of number of years the portfolio failed.

Well, two comments. 1) There is always a difference between theory and practice, and in investing practice usually falls short of theory. 2) Although FPURX has a larger stock allocation than the others, note that this isn't the whole story, as the 100%-stocks fund failed more often than four of the balanced funds.

--Theoretical 60/40 fund: 0 failures.
--FPURX: 0 failures.
--PGEYX: 1 failure, in 1969
--DODBX: 3 failures, in 1966, 1968, and 1969
--RPBAX: 3 failures, in 1966, 1968, and 1969
--MITTX (100% stocks): 7 failures, in 1930, all years from 1965-1970, and 1973
--VWELX: 8 failures, in all years from 1962-1969
















I liked the idea of using a real mutual fund instead of theoretical statistical returns, and I wanted to credit the source of the idea. I don't recommend the book, Retire with the Wellington Fund, by Josh Scandlen. The purpose of this posting is to examine the comparative results of real-world mutual funds, and the robustness of the "4% rule," not to critique the book.

[frankbrenowitz]

Thank you, Nisiprius! This kind of dispassionate analysis is very helpful to someone who tends to want to tinker with the portfolio.

[Normchad]

Outstanding post. Very thought provoking..... I’m going to have to do some thinkimg here......

[vineviz]

"4% rule:" I assumed a starting portfolio of $100,000, and a $4,000 withdrawal at the start of the first year of retirement. In subsequent years, the withdrawal is kept at "$4,000 real," i.e. $4,000 adjusted for inflation. The time frame is taken to be thirty years. The regime "succeeds" if it is possible to take thirty $4,000-real annual withdrawals and have money left at the end, and fails if it runs out of money before making thirty full payments.

Are there too many years on the x-axis? By my count, we haven't had 30 years since 2000 yet.

[nisiprius]

Are there too many years on the x-axis? By my count, we haven't had 30 years since 2000 yet.

Yikes.

(I can explain everything, I can explain everything.)

Fixed, now. Thank you.

[EddyB]

"4% rule:" I assumed a starting portfolio of $100,000, and a $4,000 withdrawal at the start of the first year of retirement. In subsequent years, the withdrawal is kept at "$4,000 real," i.e. $4,000 adjusted for inflation. The time frame is taken to be thirty years. The regime "succeeds" if it is possible to take thirty $4,000-real annual withdrawals and have money left at the end, and fails if it runs out of money before making thirty full payments.

Are there too many years on the x-axis? By my count, we haven't had 30 years since 2000 yet.

Thanks for nothing, vineviz. Nisiprius accidentally reveals the next ten years, and you had to point that out before I saw them.

[nisiprius]

"4% rule:" I assumed a starting portfolio of $100,000, and a $4,000 withdrawal at the start of the first year of retirement. In subsequent years, the withdrawal is kept at "$4,000 real," i.e. $4,000 adjusted for inflation. The time frame is taken to be thirty years. The regime "succeeds" if it is possible to take thirty $4,000-real annual withdrawals and have money left at the end, and fails if it runs out of money before making thirty full payments.

Are there too many years on the x-axis? By my count, we haven't had 30 years since 2000 yet.

Thanks for nothing, vineviz. Nisiprius accidentally reveals the next ten years, and you had to point that out before I saw them.

Yes, I get the joke. Nevertheless... thank you, Vineviz.

Details: For years after 1999, I was doing calculations over however many years remained. As it happens, all of the funds could support $4,000 real withdrawals through 2020. My calculation of "shortfall years" was the number of years during which a $4,000-real withdrawal could not be made, but in all case that was zero. Where I screwed up was in making the green bar length always equal to 30, rather than making it equal to the number of years remaining. I could have corrected this, but then the chart would have been harder to explain, and would have shown nothing interesting. Much easier and just as informative to display only years for which 30 years were available.

[rgs92]

Great post, fascinating analysis, thanks.
(Note: small typo in Fund list: DODBX, not DODGX.)

[willthrill81]

I'm not really surprised that the '4% rule' failed in the mid 1960s with live funds because they had expense ratios that Bengen's 1994 study did not include. Also, expense ratios were generally much higher back then than they are now.

[nisiprius]

(Note: small typo in Fund list: DODBX, not DODGX.)

Fixed, thanks.

[000]

I think it's time to retire the 4% rule.

[Stinky]

I'm not really surprised that the '4% rule' failed in the mid 1960s with live funds because they had expense ratios that Bengen's 1994 study did not include. Also, expense ratios were generally much higher back then than they are now.

Silly me! I thought that the “4% rule” worked darn near all the time.

Is the reason that these funds had multiple failures due only to the higher fund expenses from years ago? Or is there some deeper, darker underlying truth?

(Forgot to mention - absolutely excellent analysis, nisiprius. Kudos to you).

[willthrill81]

I think it's time to retire the 4% rule.

How can we retire a rule that virtually no one ever used?

[willthrill81]

I'm not really surprised that the '4% rule' failed in the mid 1960s with live funds because they had expense ratios that Bengen's 1994 study did not include. Also, expense ratios were generally much higher back then than they are now.

Silly me! I thought that the “4% rule” worked darn near all the time.

Is the reason that these funds had multiple failures due only to the higher fund expenses from years ago? Or is there some deeper, darker underlying truth?

(Forgot to mention - absolutely excellent analysis, nisiprius. Kudos to you).

Regarding 'darn near all the time', it largely did except for mid 1960s. Look at Bengen's 1994 study and the later Trinity study. They didn't make up the data. Kitces found the same thing.

Yes, expense ratios really do make a big difference. That's why keeping costs low is a BH principle. The '4% rule' would have obviously done better than it did in the past with today's lower ERs. But still, virtually nobody ever used the 'rule', nor should they, so I don't really see the point of lamenting about it.

[000]

I think it's time to retire the 4% rule.

How can we retire a rule that virtually no one ever used?

How can we retire the DJIA when virtually no one ever invested in the DJIA?

[MathWizard]

I think it's time to retire the 4% rule.

How can we retire a rule that virtually no one ever used?

I would disagree. Plenty of people use it for planning, to see if they have enough to retire.

I really appreciate nispirius' post.

Regarding fees, I take the 4% rule to be for after fees.
If you are paying 1% fees, then you had better use the 3% rule (4-1)%

I am paying 0.22% in my employer plan and about 0.05 % on my vanguard account. I will need to use the 3.78% rule if I stay with my employer plan, or use the 3.95% rule in my planning if I convert to Vanguard or similar low fee broker.

This is why I was horrified to hear a "free" financial advisor for my employer plan suggest that he take could take over if I should pass first, which is likely. When I asked the fee, it was 1.15% above any account fees. I told him that my wife was quite capable, having an accounting degree and having owned two businesses.

I also plan to have everything on autopilot long before I die, so that we can enjoy life.

[Bmac]

I'm not really surprised that the '4% rule' failed in the mid 1960s with live funds because they had expense ratios that Bengen's 1994 study did not include. Also, expense ratios were generally much higher back then than they are now.

Silly me! I thought that the “4% rule” worked darn near all the time.

Is the reason that these funds had multiple failures due only to the higher fund expenses from years ago? Or is there some deeper, darker underlying truth?

(Forgot to mention - absolutely excellent analysis, nisiprius. Kudos to you).

Regarding 'darn near all the time', it largely did except for mid 1960s. Look at Bengen's 1994 study and the later Trinity study. They didn't make up the data. Kitces found the same thing.

Yes, expense ratios really do make a big difference. That's why keeping costs low is a BH principle. The '4% rule' would have obviously done better than it did in the past with today's lower ERs. But still, virtually nobody ever used the 'rule', nor should they, so I don't really see the point of lamenting about it.

But this really vindicates the 4% rule. It nearly always worked in these scenarios. And as Nisi notes, there is theory and practice. In this exercise, the 4% rule including increases with inflation was followed blindly. In reality, annual withdrawal would likely be modified accordingly as needed to avoid failure. And in most scenarios the portfolios ended up larger than they started. It seems reassuring.

[AlwaysLearningMore]

I think it's time to retire the 4% rule.

How can we retire a rule that virtually no one ever used?

Michael Kitces has some simple "guardrail" suggestions for starting with 4% and then increasing/decreasing 10% (IIRC) depending upon the portfolio value. Saw this during a video interview.

4% isn't carved in stone, but to say it hasn't been used over the years as a rough planning tool would be incorrect.

[Marseille07]

I think it's time to retire the 4% rule.

How can we retire a rule that virtually no one ever used?

Michael Kitces has some simple "guardrail" suggestions for starting with 4% and then increasing/decreasing 10% (IIRC) depending upon the portfolio value. Saw this during a video interview.

4% isn't carved in stone, but to say it hasn't been used over the years as a rough planning tool would be incorrect.

Not exactly the 4% rule itself, but I find the tables of retirement horizon, AA, various SWR and the success rates extremely useful.

[willthrill81]

I think it's time to retire the 4% rule.

How can we retire a rule that virtually no one ever used?

I would disagree. Plenty of people use it for planning, to see if they have enough to retire.

In that regard, it's still fine, IMHO. Nobody should be cutting things so close that their retirement will 'fail' if they have to reduce their withdrawals somewhere along the way to cope with poor portfolio performance. But the data are quite clear that most people do this naturally anyway. Planning on withdrawing the same inflation-adjusted dollar amount from your portfolio for the next 30 years on day 1 of retirement with nary a glance at your portfolio balance would be insanity.

[BluesH]

Regarding fees, I take the 4% rule to be for after fees.
If you are paying 1% fees, then you had better use the 3% rule (4-1)%

I am paying 0.22% in my employer plan and about 0.05 % on my vanguard account. I will need to use the 3.78% rule if I stay with my employer plan, or use the 3.95% rule in my planning if I convert to Vanguard or similar low fee broker.

This has always been my interpretation, too. And is another way of looking at the desirability of low ER's. I'd rather live on 3.96% from a fund with a 0.04% ER, rather than live on 3.5% from a fund with a 0.5% ER.

It's also my understanding, from other posts on Bogleheads, that's one reason why some people like Wade Pfau have claimed lower SWRs, down to 3%, or lower (2.5%?). It's because he advocates using financial advisors with 1+% AUM fees, leaving only 3% or less from the original 4% SWR. (disclaimer - I haven't read the relevant papers by Pfau, so this is second or third hand).

Bob

[firebirdparts]

Neat piece of work, thanks!

[Buckrodgerz]

My Investment Policy says: "Limit distributions to 4% of the portfolio, and 3% if there is a year with a realized loss." It worked the first ten years with a Vanguard 45% stocks/55% bonds portfolio. My principle balance is about 10% higher after 10 years. 77 now. Hyper-inflation will kill most policies including mine.

[nisiprius]

My Investment Policy says: "Limit distributions to 4% of the portfolio, and 3% if there is a year with a realized loss." It worked the first ten years with a Vanguard 45% stocks/55% bonds portfolio. My principle balance is about 10% higher after 10 years. 77 now. Hyper-inflation will kill most policies including mine.

Just for the record, I remember the seventies, and there was high inflation--but it was not hyperinflation or anything close to it. The author who coined the word defined it as an inflation rate of 50% per month, which works out to multiplying by over a hundred times in one year.

Since the start of the consumer price index, the highest US inflation rate was less than 20% in any single year, and something like 13.5% in 1980.

[ram]

Excellent work Nisiprius.
It is fairly reassuring in the sense that the "4% rule" worked "most" of the time.

Most people on this board are more conservative than the 4%. Would it be possible to do the same analysis with a "3.5% rule". And if there are failures with the "3.5% rule", then repeat it with "3.33% rule" (Nest egg= 30 X). My hope is that a 3.33% rule will be successful 100% for each fund. If indeed it is then it will be very reassuring for people who are planning with a 3.33% rule.

[vineviz]

Excellent work Nisiprius.
It is fairly reassuring in the sense that the "4% rule" worked "most" of the time.

Most people on this board are more conservative than the 4%. Would it be possible to do the same analysis with a "3.5% rule". And if there are failures with the "3.5% rule", then repeat it with "3.33% rule" (Nest egg= 30 X). My hope is that a 3.33% rule will be successful 100% for each fund. If indeed it is then it will be very reassuring for people who are planning with a 3.33% rule.

Whether a particular withdrawal rate worked in the past doesn’t tell you much about whether it is likely to work now.

In other words, 3.33% did work in the past but that alone should provide no reassurance that it’s a good strategy for the future.

[willthrill81]

Excellent work Nisiprius.
It is fairly reassuring in the sense that the "4% rule" worked "most" of the time.

Most people on this board are more conservative than the 4%. Would it be possible to do the same analysis with a "3.5% rule". And if there are failures with the "3.5% rule", then repeat it with "3.33% rule" (Nest egg= 30 X). My hope is that a 3.33% rule will be successful 100% for each fund. If indeed it is then it will be very reassuring for people who are planning with a 3.33% rule.

Whether a particular withdrawal rate worked in the past doesn’t tell you much about whether it is likely to work now.

In other words, 3.33% did work in the past but that alone should provide no reassurance that it’s a good strategy for the future.

It really depends on how you choose to view the situation. Some believe that it's highly unlikely that their specific retirement period will be worse than the worst in ~100 years of U.S. history. Others believe that starting conditions today are the worst that they've ever been for retirees or at least close to it. Regardless, there certainly isn't an objectively correct answer for any specific retiree. Everyone has to determine what is the best overall strategy for themselves, weighing the many risks involved, not the least of which for many is working longer than necessary when they would much rather do something else with their remaining time on this earth, as best they can.

[alpine_boglehead]

Great work, thanks.

I'm not sure whether the log scale for the final portfolio value is a good idea, to me it gives the wrong visual impression. For the higher values, it seems to underrepresent by how much you overshot, and for the lower values how close you were to failure.

IMO a better measure would be remaining years of withdrawal with the same methodology you measured shortfall. That would nicely fit into the upper chart (which now is leveled off at zero, but could show the overshoot as well). That's what you would care about in the real world - how many years can I expect my portfolio to continue to sustain me (and compare that to your life expectancy).

The data suggest that results are polarized, most of the time you died rich, and in some years broke, but there's not so many middle results. The practical solution to this is to assume a positive outcome, but be completely prepared to cut expenses or try to earn some more should the bad times show up. That's likely what retirees in the late 1960s did, most people wouldn't have gone just burning through their portfolio without taking any measures after seeing it drop 50% in real terms.

The comparison of the 60/40 no-expense-ratio portfolio is a bit unfair, it would have been hard to pull this off in the days of outrageous fees.

Does anyone have an educated guess what made Fidelity Puritan Fund perform so consistently? I.e. the "seven lean years" that made the other funds fail several years in a row only show up as a minor blip.

[alpine_boglehead]

Excellent work Nisiprius.
It is fairly reassuring in the sense that the "4% rule" worked "most" of the time.

Most people on this board are more conservative than the 4%. Would it be possible to do the same analysis with a "3.5% rule". And if there are failures with the "3.5% rule", then repeat it with "3.33% rule" (Nest egg= 30 X). My hope is that a 3.33% rule will be successful 100% for each fund. If indeed it is then it will be very reassuring for people who are planning with a 3.33% rule.

Whether a particular withdrawal rate worked in the past doesn’t tell you much about whether it is likely to work now.

In other words, 3.33% did work in the past but that alone should provide no reassurance that it’s a good strategy for the future.

It really depends on how you choose to view the situation. Some believe that it's highly unlikely that their specific retirement period will be worse than the worst in ~100 years of U.S. history. Others believe that starting conditions today are the worst that they've ever been for retirees or at least close to it. Regardless, there certainly isn't an objectively correct answer for any specific retiree. Everyone has to determine what is the best overall strategy for themselves, weighing the many risks involved, not the least of which for many is working longer than necessary when they would much rather do something else with their remaining time on this earth, as best they can.

Yes. We can't know the future. A star-trek-esque guilded age might be just around the corner, or some 1984-style dystopia (there's signs for both). Take a good guess, and be prepared to adapt (but stay the course with your portfolio ).

[SimplicityNow]

Thanks for this post Nisi.

As always, I enjoyed your analysis and your insight.

[nisiprius]

For the record... I think this is mostly meaningless and should not be overinterpreted... just an "I-was-wondering..." the actual maximum sustainable withdrawal rate for these specific funds, for all years, over the specific time periods shown, according to the specific methods described above, was as follows:

[vineviz]

Whether a particular withdrawal rate worked in the past doesn’t tell you much about whether it is likely to work now.

In other words, 3.33% did work in the past but that alone should provide no reassurance that it’s a good strategy for the future.

It really depends on how you choose to view the situation. Some believe that it's highly unlikely that their specific retirement period will be worse than the worst in ~100 years of U.S. history. Others believe that starting conditions today are the worst that they've ever been for retirees or at least close to it. Regardless, there certainly isn't an objectively correct answer for any specific retiree. Everyone has to determine what is the best overall strategy for themselves, weighing the many risks involved, not the least of which for many is working longer than necessary when they would much rather do something else with their remaining time on this earth, as best they can.

To paraphrase Daniel Patrick Moynihan, “Everyone is entitled to their own withdrawal strategy, but not to their own facts.”

The probability of any particular withdrawal rate (e.g. 3.33%) working in the future is what it is. There is, in fact, "objectively correct answer" to the question "what is the likelihood that this asset allocation will support this rate of inflation-adjusted withdrawals. Of course there might be some uncertainty about that answer, but there is nothing subjective about it.

The subjectivity (aka the "personal" part of "personal finance") comes into play when a specific retiree is trying to make a decision in the face of the "objectively correct" probability. Different retirees will have different degrees of tolerance for risk, loss, or regret aversion, etc. as well as different capacities for risk and/or loss.

It should be THOSE differences in tolerance and capacity for risk that investors use to determine their "best overall strategy", IMHO, and NOT a misunderstanding about the objective probabilities that they face.

[FactualFran]

I'm not sure whether the log scale for the final portfolio value is a good idea, to me it gives the wrong visual impression. For the higher values, it seems to underrepresent by how much you overshot, and for the lower values how close you were to failure.

A log scale is very useful when dollar balances are charted. With a log scale the same linear distance along the scale represents the same percent change.

The comparison of the 60/40 no-expense-ratio portfolio is a bit unfair, it would have been hard to pull this off in the days of outrageous fees.

Does anyone have an educated guess what made Fidelity Puritan Fund perform so consistently? I.e. the "seven lean years" that made the other funds fail several years in a row only show up as a minor blip.

It is a fair comparison of actual mutual funds to the hypothetical portfolio used by Bengen. The hypothetical portfolio used by Bengen did not have fees or tracking errors that an actual investment would have had.

The Fidelity Puritan Fund at that time had investment managers who obviously made excellent decisions.

[willthrill81]

Whether a particular withdrawal rate worked in the past doesn’t tell you much about whether it is likely to work now.

In other words, 3.33% did work in the past but that alone should provide no reassurance that it’s a good strategy for the future.

It really depends on how you choose to view the situation. Some believe that it's highly unlikely that their specific retirement period will be worse than the worst in ~100 years of U.S. history. Others believe that starting conditions today are the worst that they've ever been for retirees or at least close to it. Regardless, there certainly isn't an objectively correct answer for any specific retiree. Everyone has to determine what is the best overall strategy for themselves, weighing the many risks involved, not the least of which for many is working longer than necessary when they would much rather do something else with their remaining time on this earth, as best they can.

To paraphrase Daniel Patrick Moynihan, “Everyone is entitled to their own withdrawal strategy, but not to their own facts.”

The probability of any particular withdrawal rate (e.g. 3.33%) working in the future is what it is. There is, in fact, "objectively correct answer" to the question "what is the likelihood that this asset allocation will support this rate of inflation-adjusted withdrawals. Of course there might be some uncertainty about that answer, but there is nothing subjective about it.

The subjectivity (aka the "personal" part of "personal finance") comes into play when a specific retiree is trying to make a decision in the face of the "objectively correct" probability. Different retirees will have different degrees of tolerance for risk, loss, or regret aversion, etc. as well as different capacities for risk and/or loss.

It should be THOSE differences in tolerance and capacity for risk that investors use to determine their "best overall strategy", IMHO, and NOT a misunderstanding about the objective probabilities that they face.

I didn't say anything about claiming certain facts to be true or not, so I really don't know where you're coming from with that assertion.

We can look at the very same facts and come to quite different conclusions. That happens all the time in the sciences.

We do not know with a very high degree of confidence what the probability of any particular withdrawal rate succeeding is. We can make an informed estimate using various statistical tools, but there is going to be a sizable interval of possibilities, and that's assuming that we have all of the relevant independent variables captured in our model, which we obviously do not know that we do.

If you have access to a far better method, then by all means submit it to a peer reviewed journal in the financial space. I'm dead serious.

I agree that individual retirees' differences and circumstances should be the major drivers of what they do as individuals.

[marcopolo]

Whether a particular withdrawal rate worked in the past doesn’t tell you much about whether it is likely to work now.

In other words, 3.33% did work in the past but that alone should provide no reassurance that it’s a good strategy for the future.

It really depends on how you choose to view the situation. Some believe that it's highly unlikely that their specific retirement period will be worse than the worst in ~100 years of U.S. history. Others believe that starting conditions today are the worst that they've ever been for retirees or at least close to it. Regardless, there certainly isn't an objectively correct answer for any specific retiree. Everyone has to determine what is the best overall strategy for themselves, weighing the many risks involved, not the least of which for many is working longer than necessary when they would much rather do something else with their remaining time on this earth, as best they can.

To paraphrase Daniel Patrick Moynihan, “Everyone is entitled to their own withdrawal strategy, but not to their own facts.”

The probability of any particular withdrawal rate (e.g. 3.33%) working in the future is what it is. There is, in fact, "objectively correct answer" to the question "what is the likelihood that this asset allocation will support this rate of inflation-adjusted withdrawals. Of course there might be some uncertainty about that answer, but there is nothing subjective about it.

The subjectivity (aka the "personal" part of "personal finance") comes into play when a specific retiree is trying to make a decision in the face of the "objectively correct" probability. Different retirees will have different degrees of tolerance for risk, loss, or regret aversion, etc. as well as different capacities for risk and/or loss.

It should be THOSE differences in tolerance and capacity for risk that investors use to determine their "best overall strategy", IMHO, and NOT a misunderstanding about the objective probabilities that they face.

I am not sure I have still seen an explanation of the objectively correct probability that you keep insisting exists.

Any projection of expected returns has to make some assumptions about how current metrics (bond yields, CAPE, interest rates, inflation rates) can be used to estimate future probabilities. Whatever assumptions you choose is a subjective decision, even if it based on past experience, as you said there is no assurance that those relationships will persist. Do we have any proof that the metric used to computed expected values are stationary?

No matter how you go about computing future expected returns and corresponding SWR, it seems there is a significant amount of subjectiveness to it.

[stocknoob4111]

so how much does the 4% rule need to be racheted down to get a 100% success rate in the above scenarios?

[Stinky]

so how much does the 4% rule need to be racheted down to get a 100% success rate in the above scenarios?

From the table that nisiprius posted it above, it looks like a 3.22% withdrawal rate would have been “successful” for every mutual fund and scenario tested.

Of course, past results do not guarantee future performance.

[chipperd]

Nisiprius, thanks for putting together this data.
Question: Do the charts and data you display take into account the fees, loads, expense ratios, and other actual costs corresponding to each year?
I was doing a quick search and couldn't find info on past expenses, from say the 60's and 70's.
Thanks again
Chipperd

[nisiprius]

Nisiprius, thanks for putting together this data.
Question: Do the charts and data you display take into account the fees, loads, expense ratios, and other actual costs corresponding to each year?
I was doing a quick search and couldn't find info on past expenses, from say the 60's and 70's.
Thanks again
Chipperd

They are based on Morningstar's performance data for the funds, which don't include loads, but--as with the "official" fund return data published by fund companies--reflects the actual amounts an owner would have received when redeeming shares. I don't know Morningstar's own source for the data, or how reliable it is for older data, though. That is, I don't know if the SEC disclosure rules were always the same.

so how much does the 4% rule need to be racheted down to get a 100% success rate in the above scenarios?

As posted above, these are the percentage withdrawal rates that exactly resulted in success for all starting years, and the starting year that was the worst case. But as I also said above, this is not very valuable information because of its being 20/20 hindsight.


Something I should have done long ago is to find Bengen's actual 1994 paper, Determining Withdrawal Rates Using Historical Data and his actual statements:

It is clear from Figurel(a) thatan"absolutely safe" (to the extent history is a guide) initial withdrawal level is 3 percent, in that it ensures that portfolio longevity is never less than 50 years. (This is also true for withdrawal rates as high as approximately 3.5 percent.) However,most clients would find such a low level of withdrawals unacceptable.)

[MathIsMyWayr]

Something I should have done long ago is to find Bengen's actual 1994 paper, Determining Withdrawal Rates Using Historical Data and his actual statements:

It is clear from Figurel(a) thatan"absolutely safe" (to the extent history is a guide) initial withdrawal level is 3 percent, in that it ensures that portfolio longevity is never less than 50 years. (This is also true for withdrawal rates as high as approximately 3.5 percent.) However,most clients would find such a low level of withdrawals unacceptable.)

They asked of the best opinion of his and reject it! Fight the numbers?

[willthrill81]

so how much does the 4% rule need to be racheted down to get a 100% success rate in the above scenarios?

We already knew that the '4% rule', based upon indexes, worked 100% of the time or nearly so, depending on precisely which assets and indexes you used.

The valuable takeaway from nisiprius' interesting study, by my estimation, is the impact of expense ratios and such on SWRs. The problem with ERs is that they remove some chunk of your portfolio every year, even when your portfolio's performance is suffering. That alone can significantly lower the SWR, even for 'modestly high' ERs in the range of .50%, which I'm guessing was the low end for most of these funds during the critical mid-1960s period.

For the same reason, this is part of the reason that so many BHs have cried foul on some of Wade Pfau's comparisons of SWRs vs. annuities; the combined AUM fees and ERs in at least one of his papers was something like 1.5%-2.0%. That's a huge drag on portfolio performance and is definitely not necessary any more for a DIY investor, who can pay no AUM fee and virtually (or, in some cases, literally) zero ERs.

[GuyInFL]

The Fidelity Puritan Fund at that time had investment managers who obviously made excellent decisions.

Yes they did.
OP did an excellent analysis but perhaps the nice results from Fidelity Puritan Fund are an example of data mining. And of course plenty of the funds over the time period didn’t survive.

[GuyInFL]

Michael Kitces has a great podcast with Bengen, Father of the 4% rule, and he said high inflation killed portfolios.
His current recommendation is...

“It’s not a great time to be taking high withdrawals now with the market so expensive, but it’s not awful either because inflation is very low. I think somewhere in 4.75%, 5% is probably going to be OK. We won’t know for 30 years, so I can safely say that in an interview.”

https://www.thinkadvisor.com/2020/10/19 ... guideline/

[vineviz]

Michael Kitces has a great podcast with Bengen, Father of the 4% rule, and he said high inflation killed portfolios.
His current recommendation is...

“It’s not a great time to be taking high withdrawals now with the market so expensive, but it’s not awful either because inflation is very low. I think somewhere in 4.75%, 5% is probably going to be OK. We won’t know for 30 years, so I can safely say that in an interview.”

https://www.thinkadvisor.com/2020/10/19 ... guideline/

I know that Bengen was making a joke, but "I'll be dead before you can prove I'm wrong" isn't terribly confidence-inducing.

[stocknoob4111]

Michael Kitces has a great podcast with Bengen,

Thank you! always good to have new interesting material for my next hike!!

[stocknoob4111]

One thing to note is that Bengen mentions that we are in a low inflation environment which is generally true except for renters... housing prices and rents are inflating at 6-7% a year which is triple the rate of the CPI and perhaps also more than expected returns from equities and housing is one's biggest expense. So 4-4.5% may be too optimistic for renters unless they plan to do something like geo-arbitrage.

[zarci]

Out of pure curiosity, are those Python plots?

[grabiner]

Well, two comments. 1) There is always a difference between theory and practice, and in investing practice usually falls short of theory. 2) Although FPURX has a larger stock allocation than the others, note that this isn't the whole story, as the 100%-stocks fund failed more often than four of the balanced funds.

--Theoretical 60/40 fund: 0 failures.
--FPURX: 0 failures.
--PGEYX: 1 failure, in 1969
--DODBX: 3 failures, in 1966, 1968, and 1969
--RPBAX: 3 failures, in 1966, 1968, and 1969
--MITTX (100% stocks): 7 failures, in 1930, all years from 1965-1970, and 1973
--VWELX: 8 failures, in all years from 1962-1969

How much of the shortfall with actual funds is the result of their expenses? If a fund has 0.5% expenses, that reduces the 30-year sustainable withdrawal amount by about 8% (the exact amount will depend on sequence of returns).

And on the other hand, choosing these funds creates a survivorship bias; there are lots of balanced funds that would have failed to sustain 30 years of withdrawals but were closed first.

It is not surprising that 1966 is the start of many of the failures (and even the expense-free S&P/Treasury portfolio barely made it). 1966-1982 is the longest period in which US stocks underperformed inflation.

[nisiprius]

Out of pure curiosity, are those Python plots?

No, they're Excel. I hate Excel charts but I can't seem to break myself of the habit. It's hard to know where to begin but their terrible, awful, no good, very bad log scales are just one problem.

[tonyclifton]

I assumed a starting portfolio of $100,000, and a $4,000 withdrawal at the start of the first year of retirement. In subsequent years, the withdrawal is kept at "$4,000 real," i.e. $4,000 adjusted for inflation.

Thank you for putting this together. If inflation was “expected” wouldn’t bond and equity performance incorporate the “expected” inflation? If so, couldn’t you just withdraw 4% each year and not the $4,000 + inflation?