By Dr. Jim Dahle, WCI FounderI've written numerous times about the concept of a safe withdrawal rate (SWR), including these posts:
- The 4% Rule and Safe Withdrawal Rates
- 3% Is the New 4%? The Safe Withdrawal Retirement Rate
- How Much Will You Probably Die With?
However, the silliness of it all continues, and today I'm going to rant like never before on this topic in the hopes of realigning how people think about this concept.
What Is a Safe Withdrawal Rate?
The concept of a safe withdrawal rate is that a retiree can withdraw a certain amount of the portfolio value at retirement each year (usually indexed to inflation) with little risk of running out of money. That amount is the safe withdrawal rate. Note that it is NOT a GUARANTEED withdrawal rate. It is a SAFE withdrawal rate. If you want guarantees, look into insurance products such as Single Premium Immediate Annuities (SPIAs). Unfortunately, you can't really get inflation-indexed SPIAS anymore. Your best option there is to delay Social Security to 70.
Where Does the Safe Withdrawal Rate Concept Come From?
While not the first people to look at the concept, the SWR concept was popularized by the Trinity Study in the 1990s. Back then, financial advisors were telling their clients that if their portfolio was earning 8% a year on average, they could spend 8% a year without any fear of running out of money. Some were even more aggressive with spending because of the outsized equity returns of the late 1990s. However, the problem with this approach (aside from the fact that late 1990s equity returns were far above average) was that it did not address the problem of Sequence of Returns Risk (SORR). In a nutshell, SORR is when, despite having adequate (again, let's say 8%) average portfolio returns throughout retirement, the portfolio will run out of money prior to death if the crummy returns show up first during early retirement. Withdrawing a large amount from a portfolio while it is falling in value can decimate a portfolio relatively quickly.
The Trinity researchers looked at the best database available (US stock and bond returns from 1927 on in rolling time periods), applied various time periods (15, 20, 25, and 30 years) and various asset allocations ranging from 100% bonds to 100% stocks, and then applied withdrawal rates from 3%-12% to see what percentage of the time the portfolio survived. This is what the data showed:
For a 30-year retirement with a 50/50 portfolio, a 3% withdrawal rate survived 100% of the time, a 4% withdrawal rate survived 96% of the time, a 5% withdrawal rate worked 2/3 of the time, a 6% withdrawal rate worked half the time, and an 8% withdrawal rate worked 10% of the time. Thus, advisors telling clients they could take out 8% were doing a huge disservice to them.
This is the birth of the 4% Rule. Four percent is a “safe withdrawal rate.” Maybe you could call 3% a “guaranteed withdrawal rate” (although there is never a guarantee in life, especially when it comes to market returns).
More information here:
4 Methods of Reducing Sequence of Returns Risk
What Is the Problem with the Safe Withdrawal Rate?
The problem with the SWR is engineers. What do I mean by that? Well, nothing too sinister. My father is an engineer. But engineers love to solve problems. And they love data. And they love spreadsheets. And they love planning way in advance. We need engineers for many things in life. But we don't need them when it comes to planning your retirement. What happens when you get engineers involved? You get people arguing about whether 3.67% or 3.74% is the correct safe withdrawal rate. Reminds me of the old joke:
“How do you know economists have a sense of humor?
They use decimal points.”
If you think economists use decimal points, you should check out engineers.
Injecting precision that doesn't exist into a process might make you feel better, but it certainly doesn't make the data any more useful. Consider the source of the data: less than 100 years of market returns in a single country. While there might be dozens of rolling 30-year time periods, there are fewer than four independent 30-year data points. You can't use a dataset like that and then pretend you can get some sort of precise answer out of it. The correct answer to “what is a safe withdrawal rate?” is “something around 4%, probably in the 3%-6% range.” Not 3.59% or 4.21%.
Pessimism Is Sexy
The other problem with the folks that get into this SWR stuff is that they like to look at all the possible reasons why a SWR could be much less than 4%. Maybe they project lower equity returns or bond yields are lower or the world seems scarier or whatever. They forget that the Trinity Study database (including updates) includes events such as:
- The Great Depression
- World War II
- The inflation of the 1940s
- The stagflation of the 1970s
- The Cold War
- October 19, 1987 (look it up)
- The Asian Contagion
- The meltdown of long-term capital management
- The dot.com bubble and bust
- The Global Financial Crisis
- Several cryptocurrency winters
- The COVID pandemic
That's a lot of bad stuff. So, when people are saying that the future will be worse, wow, that's pretty pessimistic. Especially since economic history should be titled “The Triumph of the Optimists.” But pessimism is sexy. In his excellent book “The Psychology of Money,” Morgan Housel wrote:
“Optimism sounds like a sales pitch. Pessimism sounds like someone trying to help you.”
“Pessimism just sounds smarter and more plausible than optimism.”
“Pessimism isn’t just more common than optimism. It also sounds smarter. It’s intellectually captivating, and it’s paid more attention than optimism, which is often viewed as being oblivious to risk.”
“It’s easier to create a narrative around pessimism because the story pieces tend to be fresher and more recent. Optimistic narratives require looking at a long stretch of history and developments, which people tend to forget and take more effort to piece together.”
“The intellectual allure of pessimism has been known for ages. John Stuart Mill wrote in the 1840s: ‘I have observed that not the man who hopes when others despair, but the man who despairs when others hope, is admired by a large class of persons as a sage.'”
Get a bunch of engineers cranking numbers using past data and current yields, and, all of a sudden, they start competing to see who can be the most pessimistic. Some small percentage of them might even be neurotic, maybe even a little OCD. But all of a sudden, people are talking about 2% being the safe withdrawal rate. Maybe 1.5% if you retire early. It's bonkers. That's what that is.
The Silliness of Safe Withdrawal Rates
If you go down this rabbit hole too far (and you don't have to go very far to go too far), you will meet a lot of strange people who try to convince you otherwise. They are mistaken. That's God's honest truth about withdrawal rates. Go back to that chart. Take a look at the 75/25 30-year line. It's 100%. In all of recorded financial history, a 4% (adjusted for inflation) withdrawal rate NEVER ran out of money. In fact, on average after 30 years. the portfolio was 2.7X the original portfolio value. Even if you decide you're going to do 5% of the original portfolio value (again, adjusted upward each year with inflation), it worked more than four out of five times.
How Long Is Retirement?
The average American retires at 64. The average life expectancy of a 64-year-old is 17 (male) to 20 (female) years. What's the table look like for a 20-year retirement? The 75/25 20-year line shows that you can withdraw 9% a year and only run out of money half the time. Nine percent! And you've got these crackpot engineers running around telling people they can only take out 2% a year. By the way, if you're a financial advisor and you're telling your 92-year-old client she can only spend 4% a year, you're a jerk. You can have very aggressive withdrawal rates for short time periods without any problem. There's a reason the RMD for a 92-year-old (9.8% of portfolio) is more than twice as large as that of a 75-year-old (4%).
More information here:
The TIPS SWR
Financial advisor Allan Roth took a look at a portfolio that was 100% TIPS at our current yields and determined that one could take out 4.36% a year, adjusted to inflation, and your money would be guaranteed to last 30 years. You don't even have to take any risk, and you can get 30 years of 4%+ inflation-adjusted withdrawals. You're not immortal. It's OK to spend some principal. Now, maybe you're retiring at 45, and 30 years isn't quite enough to feel comfortable. Fine. The difference in withdrawal rate between a portfolio that will last 30 years and a portfolio that will last forever is pretty trivial. Besides, how many 45-year-old retirees do you know who NEVER earn another dollar? There are not very many of them. Even the most diehard physician FIRE dude I know left medicine at 43, and he is still earning a little income.
Larimore vs. Pfau and Big ERN on SWR
I have three friends. We'll call the first one Taylor Larimore. Let's say he's a 100-year-old World War II vet who retired in 1980 at 55 on $1 million, and he has been living on it ever since. His philosophy on spending money in retirement is to “adjust as you go.” You simply take a look at how much you have and adjust how much you spend based on that. In a year when you earn a lot, you spend more. After a bad year or two, maybe you spend and give a little less. He bought a couple of SPIAs in his ninth decade of life and mostly lives off those and Social Security these days.
Let's call my second friend Wade Pfau. Let's just say he's a “retirement researcher” who makes his living off publishing study after study about safe withdrawal rates. Boy, does he ever like to get into the weeds on this stuff! Publish or perish you know, and this area of research has been fertile ground over the years. He has argued that the SWR is 2.4%.
Let's call my third friend Big ERN. While not an engineer (he's an economist by training), he has published a long, detailed blog series about safe withdrawal rates that rivals Pfau's work. He has also argued for a sub-4 % withdrawal rate.
Who's right and who's wrong? People say Taylor was just lucky. He retired at the beginning of the longest bull market in history. He won't be the last person who retires at the beginning of a bull market. If Pfau likes 2.4%, he's going to love the 0% withdrawal rate many retirees are currently using. Big ERN says adjusting as you go might mean a 15- or 20-year adjustment in spending, and he calls that “failure.” Calling having to make an adjustment “a failure” seems overly harsh to me. That's just how life works, both in accumulation and decumulation.
More information here:
Let’s Celebrate Taylor Larimore’s 100th Birthday by Asking Him 4 Questions About Money
SPIAs
If you go down the rabbit hole and end up with something like a 2% withdrawal rate, you need to seriously reconsider your approach. If you are so worried about running out of money that you are only willing to spend 2% a year, you should do something to reduce that anxiety. Here are some great options:
- Delay Social Security to age 70. This is the best deal out there for an inflation-indexed annuity.
- Buy a SPIA or two. Use these together with Social Security and any pensions to put a floor under your retirement. Even if your portfolio totally fails, you won't be eating Alpo.
- Ladder some TIPS. It's 4.36% for 30 years at today's yields. Yes, you could live longer than 30 years, I suppose. Maybe cut it back to 4% instead of 4.36%.
A Real Retirement Spending Plan
Few who have looked at this problem seriously actually advocate for a systematic withdrawal plan based solely on the initial portfolio value that never changes over the following decades. Most recommend a variable withdrawal plan based on returns as you go along. Whether you want to formalize that or just eyeball it like Taylor is up to you. In essence, the risk is SORR, i.e., the idea that you get terrible returns in the first few years of retirement. If that happens, you stick with something like 4% or even less. If it doesn't happen, you can spend 5%, 6%, or even more and be fine.
But you don't have to stick with the same stupid withdrawal rate for 30 years if your first three years go terribly. Just like you adjusted your spending to your income during your working years, you can do the same during your retirement years. If your retirement spending is 100% fixed and predetermined, you probably retired too early with too little money. It's one thing if you were forced to do that, but if you did that voluntarily, that's on you.
My advice on retirement spending is this:
Start at something around 4% of the portfolio value and adjust as you go.
That's it. It's that simple. The more uncomfortable you are with that plan, the more you should put into SPIAs and TIPS ladders. The more comfortable you are, the more you can leave in stocks and bonds and the more you can likely spend in retirement and leave to heirs.
Reverse Engineering
Perhaps the biggest benefit of the 4% Rule Guideline is figuring out how much you need to retire. For example, consider a doctor who needs $120,000 a year in today's dollars to retire and who saves $50,000 a year in today's dollars for retirement and earns 5% real (after inflation) on the investments. How long does this doctor need to work and save? It depends on the withdrawal rate.
- 5%: 25 years =NPER(5%,-50000,0,120000/5%)
- 4%: 28 years =NPER(5%,-50000,0,120000/4%)
- 3.5%: 30 years =NPER(5%,-50000,0,120000/3.5%)
- 3.33%: 31 years =NPER(5%,-50000,0,120000/3.33%)
- 3%: 33 years =NPER(5%,-50000,0,120000/3%)
- 2.5%: 36 years =NPER(5%,-50000,0,120000/2.5%)
That's an 11-year difference between someone willing to “risk it” with a 5% withdrawal rate (which will probably work, especially if you're willing to be flexible) and someone who went down Wade Pfau's rabbit hole. Maybe 11 years doesn't seem that long to you. But think about it this way.
Retirement is often divided into the “go-go years,” the “slow-go years,” and the “no-go years.” Perhaps, on average, it stacks up like this:
- Go-go years: Age 65-75
- Slow-go years: Age 75-85
- No-go years: Age 85+
Let's say you're a doctor who started saving upon exiting residency at age 33. Age 33 plus 36 years equals age 69. That leaves you six go-go years. On the other hand, if you retired after 25 years, you retired at 58. That leaves you 17 go-go years. That's almost three times as many of the best years of retirement. Does it matter? Absolutely it does.
More information here:
Doesn't Matter for Many
For many white coat investors, however, the debate about safe withdrawal rates is purely academic. Because they enjoyed their work so much and were so talented at earning, saving, and investing, they hit FI long before they were actually ready to be done working. As you can see, it doesn't take that many years to go from a 4% withdrawal rate to a 2.5% withdrawal rate. Perhaps eight years. Maybe less with an inheritance or other windfall or a particularly high savings rate. If you were planning to retire at 55 but hit FI at 47, you now face an enviable dilemma. You can retire early. You can spend more money in retirement. You can give more money away during retirement. Or you can stress less about running out of money and probably give away more at death. I hope we all get to face THAT dilemma! You can whine about your RMD problem while you're at it.
Rich, Broke, or Dead
The SWR movement is composed of a lot of well-meaning but anxious people advocating for ridiculous less than ideal recommendations. Understand the concepts, but don't fall for their arguments. It's OK to spend your money. You're far more likely to end up dead with a huge portfolio than broke and wishing you had more. Don't forget the lessons of the Rich, Broke, or Dead graph. Green is rich. Red is broke. Black is dead. In the end, we're all dead, and most of us were rich right up until the end.
What do you think? Why do people get so pessimistic about withdrawal rates? What is your retirement spending plan?
Love and 100% agree with the part of your article regarding why SWR is such a bad idea. I’d go further to say that any of the hacks to improve SWR, such as many that have been posted on the bogleheads forum and/or in the ERN series are also poor choices. I’d characterize most of them as “Lipstick on a Pig”.
Yes, Taylor got lucky and not just because he retired into a great bull market but also because he purchased a SPIA back when rates were also very high. On top of that he’s won the longevity contest as well.
BigERN continues to try and make SWR better, with a couple of exceptions where he briefly looked at some other methods that weren’t simply dressed-up SWR.
Agree that Dr. Pfau is pessimistic and that his research does show how a SPIA properly applied and at the right point in retirement can certainly help.
Now this particular engineer feels somewhat maligned (in a tongue in cheek way) by the sweeping statement regarding the problem with SWR being engineers. This particular engineer enjoys math and does love his spreadsheets and after years of looking at this stuff, including endless SWR simulations, I landed on amortization being the most logical withdrawal choice and had everything set up when I retired last year.
Amortization can be the most basic version: VPW (Variable Percentage Withdrawal). It can be as simple as pre-calculating a withdrawal rate for each year of retirement or, if it suits one’s particular situation, looking up the withdrawal percentages on VPW’s wiki: https://www.bogleheads.org/wiki/Variable_percentage_withdrawal It doesn’t require an engineering, math or finance degree to go with method, either. Or one can use the VPW spreadsheet and take into account Social Security and/or a Pension.
Or it can be the more complex ABW (Amortization Based Withdrawals) which adjusts for changes in expected returns of assets and builds in time-value-of-money concepts for future cash flows: https://www.bogleheads.org/wiki/Amortization_based_withdrawal
Finally, it can be the most complex version of all, which is TPAW (Total Portfolio Allocation and Withdrawal), where it has all of the concepts of ABW above, but effectively creates a personalized, dynamically adjusted Asset Allocation Glidepath: https://www.bogleheads.org/wiki/Total_portfolio_allocation_and_withdrawal.
The point is, there are logical methods one can use for withdrawal which have no links back to SWR, nor do they require an ad-hoc approach of starting with around 4% and adjusting as you go.
Now, about those physicians….. 🙂
One of Jim’s strongest pieces of writing followed up by one of the best comments. Thanks for the links and the broader view on engineers and their pessimism vs optimizations
Dear Engineer,
This comment was brilliantly written. I hadn’t heard of any of these other techniques. Thank you! Jim, looks like we still need those engineers (but in general I agree with your article and thing Pfau is too doom/gloom).
Grateful MD
I don’t think you and I are using the “SWR movement” term in the same way. You’re using it to just apply to using the 4% rule as your withdrawal method. I think everybody agrees that’s not very smart. I’m using it to refer to all this talk and discussion about 8 or 10 different methods to withdraw from your portfolio. Basically the whole last half of your comment demonstrates what I mean when I say “engineers do this stuff.” It’s not necessarily JUST engineers, but you being an engineer and all…
I actually do think that Wade Pfau and Big ERN have done some AWESOME work on this stuff that the rest of us are fully in their debt for having done. However, I think there is a lot of missing the forest for the trees happening in the academic work done by them and their peers as well as the typical discussions about it like the one happening right here in the comments section and two other forums where this post is being discussed. THAT is what this rant is about:
# 1 Too much precision when the data does not merit that level of precision
# 2 Too much pessimism when that level of pessimism is not warranted
# 3 Too much formality to the methodology when eyeballing it works at least as well
I’m not referring to the 4% rule (which should never be referred to as a “rule”), but SWR in general.
Per the definition of SWR on the bogleheads wiki:
“A safe withdrawal rate (SWR) is the quantity of money, expressed as a percentage of the initial investment, which you can withdraw per year from a portfolio, for a given quantity of time, including adjustments for inflation, without portfolio failure, where ‘without failure’ is a 95% probability of not running out of money at any time within the specified period.”
That is, one starts with a certain percentage to withdraw from one’s portfolio. Each year the withdrawal is adjusted by the rate of inflation of the prior year. Alternatively, one can think of the rate as a fixed rate, in real terms. It’s “safe” in so much as the percentage chose results in the retiree not running out of assets in some large percentage of historic cases, given a particular asset allocation. I do see people sometimes referring to SWR to mean any sort of withdrawal method one comes up with that can backtest and give satisfactory results, but I don’t think that’s correct.
One can now take the initial concept of SWR, as defined above, and hack it to try and make it better and find yourself less likely to run out of money prematurely. As noted above, that includes most of the studies that BigERN has done, what Kitces ratcheting method does, as well as floor/ceiling rules, Gummy’s sensible withdrawals, ad nauseum. These all start with a percentage withdrawal based on SWR and then, effectively, modify the withdrawals along the way by doing more than just adjusting for inflation.
The methods I show, at least the first two, were from a class of withdrawals known as the “actuarial method”. The math is the same math used to calculate loan payments, which most certainly wasn’t invented by engineers, though we have more than sufficient training to understand it. These sorts of withdrawal methods have been used by financial planners for quite a while, but it’s relatively recently that DIY’ers have taken it up and provided some tools for those who are interested in using them.
Even with these methods (well, except maybe for TPAW because it’s quite complex) there is no reason to be a slave to a tool or spreadsheet. I’m in retirement and currently using my own version of ABW and I treat it only as a “max” I can withdraw. Otherwise, I withdraw what I think I need. My backup plan for my wife, should I pre-decease, is to go with something similar to the first one, VPW. In the end, all it requires is a year by year table of the percentage that she can withdraw from our portfolio. I think anybody can do that since, after all, this is exactly what they will need to do once RMD’s kick in.
At the end of the day, I’m completely aligned with your rant for #1 and #2 above, but less so for #3 in your list, unless the retiree has amassed so much and lives so far below their means that most anything will work.
The discussions and arguments on this subject will continue on the forums just as they always do when factors, holding international assets or not, gold, crypto (when allowed to be discussed), tilting, etc. are brought up.
Ben Mathew from Bogleheads here. I am the developer of Total Portfolio Allocation and Withdrawal (TPAW). I’ve also contributed to the Amortization Based Withdrawals (ABW) wiki on Bogleheads, which is the withdrawal strategy used by TPAW planner as well as the standard withdrawal method coming out of economic models that study portfolio allocation and withdrawals.
You don’t seem convinced about the utility of these tools. So I’d like to explain why I think they are the right tools for calculating withdrawals.
I agree with your assertion in the article that retirees should be flexible and adjust their withdrawals in light of portfolio performance. But the key question is how should they adjust? How should a retiree a decade into retirement and who has faced a poor sequence of return adjust their withdrawal? How much more can they withdraw if the market turns around and does well? You seem to be saying that we can rely on our intuition for this. This is where I disagree. This might be true when the portfolio is well in excess of desired spending and withdrawals are very conservative. But when the portfolio does not obviously exceed desired spending (which I think is the situation that most people would be in), an explicit calculation is needed. Should the adjustment be 10%? 20%? These are materially different options and hard to say without some explicit modeling. I don’t think people have great intuition around this—any more than they have intuition for how much to save during their working years. The math of compounding, variability of returns, sequence of return risk all combine to make this quite hard—probably impossible—to intuit. The question is not whether we should use a model to calculate this, but what type of model and calculation to use.
One option is to base the calculation on SWR itself by recalculating SWR with updated portfolio balance and horizons. This turns out to have some unintended behavior when you actually model it out explicitly. ABW is a simpler, more direct approach that behaves as expected. Also, as I mentioned earlier, ABW is the withdrawal strategy that comes out of economic models that study precisely this question of portfolio withdrawals. SWR, in contrast, is an ad hoc methodology that does not withstand careful analysis.
I have written a lot about the advantages of ABW over SWR on Bogleheads. I’d be happy to do a guest blog post here to go over in more detail why you and your readers should consider ABW over SWR tools to construct and get answers to questions about variable withdrawals.
Here’s the guest post instructions:
https://www.whitecoatinvestor.com/contact/guest-post-policy/
We probably need to run a post giving a broad outline of the various types of withdrawal methods prior to a post talking about just one, but I’ve got that post planned anyway and can get it done before running a guest post.
I think somebody counted 296 different withdrawal methods in the literature recently. The likelihood that any one of them is “the best” is obviously pretty low and almost surely cannot be known a priori.
And no, I’m not yet convinced that any formalized method is necessarily better than an ad hoc/eyeball style method for the few people that actually need to closely align their spending with any method. Keep in mind the target audience for this site is relatively high income and presumably eventually relatively high wealth compared to even the average Boglehead. But the amount of people who need to pay close attention is a completely separate topic and we both agree that there are at least some who do.
It’s going to be difficult to compare withdrawal methods with different objectives.
The amortization methods that Ben and I are bringing up will not result in a prematurely depleted portfolio. Via the math involved, they will achieve this objective systematically, but they don’t do so by looking at past sequences of returns or monte carlo methods in an attempt to find a number that worked based either on past sequences or similar distributions of returns that past sequences had.
BigERN has already noted that the magnitude of cuts required in spending during a serious downturn might not be small. Amortization takes care of this systematically without a need to guess. And even if it overshoots, which can only be known in hindsight, the portfolio will not deplete prematurely. Again, to a large extent, the magnitude of the cuts will largely be under the retirees control based on how much of their spending is derived from low risk income sources such as TIPS ladders, SS, SPIAs, pensions, etc.
Regarding the income level & savings of physicians vs. bogleheads, this was posted on WCI a few years ago and if one scrolls down to the “Doctors don’t save section”, where there are some interesting charts. https://www.whitecoatinvestor.com/how-much-money-to-retire/
Now it’s been quite a while since Bogleheads allowed polls, and they stopped the “net worth” poll back in 2015. But the results of the last poll can be still found on this googlesheet on the summary/plot tab.
https://docs.google.com/spreadsheets/d/1IwBXes8GdYOmOrtKnYLd5Vw8M7TALte_nk63hbUfAIY/edit?pli=1#gid=1649182824
Even with the difference in the year that these polls were made, there isn’t that much difference in overall wealth vs age from what I can see. Which is to say that one might expect a similar range of considerations when it comes to how these two populations might wish to decumulate their portfolios.
Looking forward to the future post on this entire subject!
Great. I’ll write a potential guest post on this and send it your way.
As a follow up to my previous response, for those who might be interested in amortization.
Why do I think it’s a better method? First, there is no escaping sequence of return risk for a portfolio that contains risky assets. But you can change how it manifests itself. What amortization does is guarantee that your portfolio lasts exactly as long as you want it to last. The tradeoff is that instead of a risk of prematurely running out of assets, your withdrawals will vary from withdrawal period to withdrawal period such that there may come a time when the calculated withdrawal is too low to pay your bills.
This can be mitigated somewhat by your AA choices and/or by taking into account low-risk sources of income such as Social Security, a Pension, or a TIPS ladder to form a relatively safe income floor.
The opposite can also happen where a calculated withdrawal is significantly more than what you need to pay your bills or to provide for any discretionary spending. You don’t have to actually spend it all. You can keep whatever you don’t actually need invested and think of your calculated amortization based withdrawals as being a “maximum”. Or you can withdraw it and put it in some sort of rainy day fund that you can draw on for those too-low withdrawal years. The point is, you have options.
Another feature of amortization is that you don’t have to target a $0 final value. It can be any dollar amount you wish, allowing you to pre-plan for a legacy or to use as additional margin for your planning, which can then be spent later in retirement if it turns out that it looks like you’re going to be around longer than you expected.
Perfect? Heck no! But it does provide a systematic way of adjusting withdrawals based both on how your portfolio is performing as well as how long you’re planning to be around.
Thank you, Jim. After reading this, I feel pretty comfortable starting off with a 6% or high withdrawal rate. I definitely plan on delaying social security until 70, but I’m not exactly banking on it being around in its current generous form 50 years from now (I could very well still be alive then).
I like working so I don’t really see myself 100% stopping working until I just can’t anymore, which hopefully is at least in my 70’s. If I “retire” and have a series of poor returns early on, it would be easy in my specialty to do a moonlighting only gig for a few hours here and there or work a few weeks a year to make a fairly significant income to replenish some of those losses. If cognitive decline precludes that, then I’m probably not going to be in any kind of shape to be spending as much in discretionary leisure and traveling anyways and my spending will naturally adjust downwards.
I don’t see any reason to plan for any lower than a 6% withdrawal rate in my situation. But given how long I plan to work, my wife and I will probably have enough saved to withdraw at a lower rate anyways. That really just gives us the freedom to spend more now and not feel bad about it.
Yeah I think one of the huge benefits of being able to forecast reasonably is it takes a lot of stress out along the way. It’s such a relief to know what “enough” is
If you work into your 70s and your spending is very flexible, I don’t think starting at 6% is all that risky at all.
Thanks for being a voice of reason on this topic. I can’t recall how many threads on Bogleheads I’ve seen that are some variation of “I have 8 million dollars, spend 150k per year, and am 59 years old. Do I have enough to retire?” And lots of people actually chime in and say “No”!
The level of conservativeness when it comes to withdrawal rates is pathologic at times.
Exactly what I’m ranting about. Well, one of the things I’m ranting about.
Great article. As usual, have to chime in with “Soc Sec”is one of the great inflation indexed “annuities”, but I would include Military Retirement (starts as early as age 38, maybe 42-46 for a physician.) No penalty if you keep working. Adjusts exactly like Soc Sec does. Comes with cheap (in every sense, sadly) healthcare insurance.
-Usual caveat – joining the military for the financial benefits is not usually the best move for physicians – but if you want to serve anyway for 20+ years, the retirement is uniquely good. My 30 year retirement started in 2020, and has adjusted with inflation to $125k/year.
It’s wonderful the government can offer something relatively unique these days (i.e. an inflation protected pension) to attract talented folks to serve our country. We need military docs and thank you for pointing out benefits to this service.
Yea, it’s even better than SS since it can start as early as 38 and includes healthcare.
But the point of including delaying SS to 70 in this article was that if you’re so worried about low returns crashing your portfolio, you should spend the portfolio now and let the guaranteed benefit of SS grow at 8%.
Great post!
One thing that I do not see addressed with the SWR is factoring in income from dividends, interest, real estate investments. How should that factor in? Especially, given that dividends and interest will change over time as one spends down their principle.
Thanks again for all the great insight!!
“dividends, interest, RE investments” – that’s all just investment returns. Of course that’s factored in to your total portfolio returns!
Have a rental property that provides rental income? That’s akin to a pension/income in retirement and can reduce the amount of portfolio withdrawals you need and therefore portfolio size.
cfireSIM is great at easily adding items like that into your projections
An alternative way to treat rental income (especially since withdrawals of principle are so hard with such an illiquid asset) is to treat it like SS, i.e. it reduces the amount of income you need but you don’t include it in your portfolio. However, I think it should be discounted for the fact that it is less certain than true guaranteed income streams. Maybe only use 75% of cash flow for instance instead of 100% like you would for SS when you run your numbers. I don’t think I’d do that with publicly traded securities though.
It doesn’t factor in. Obviously you spend the dividends, interest, realized/distributed gains from your taxable account, RMDs, SS and other taxable income first, but if that doesn’t get you to 4%/what you need to spend, then you sell some shares to get there. Whether income makes up 1% or 3% of that 4% doesn’t really matter.
For the TLDR crowd:
“Start at around 4% of the portfolio value and adjust as you go.”
That sums it up nicely. Based on past performance, we can’t expect too much precision from a ballpark number. Larry Kotlikoff argues we have 30 years of data, but it is really a sample of six non-overlapping periods. We can’t do too many stats with the small sample size. We’ve had 40 years of declining interest rates boosting asset prices. Who knows what the future holds? This is about the best guess we have.
I plan to go from Rich to Dead without going Broke. It would be a lot easier to plan if I knew when that transition will occur.
Two things convinced me to consider the 4% rule a guideline instead: your presentation at WCICON 2023 and the book “Die With Zero.” The fact that the majority of retirees don’t spend their principal in retirement is proof that we need to stop talking about withdrawal rates less than 4%. For my parents’ retirement, I use a modified, reverse value averaging withdrawal. Value averaging (endorsed by Bill Bernstein) is tricky in the accumulation phase, but it’s much easier in the decumulation phase.
I really appreciated your deep dive into figuring out a safe amount to withdraw from retirement savings without hitting a snag down the road. It was incredibly insightful to see how various factors, like market trends, spending habits, and even our personal feelings about money, can impact how long our savings last.
One thing that caught my attention was the absence of a discussion on taxes and their impact on withdrawal rates. Since studies like the Trinity Study usually present these rates before considering taxes, I’m curious about how taxes fit into the whole equation. Did the Trinity Study factor in any specific tax rates? And how should we think about our tax responsibilities when applying these safe withdrawal rate principles to our own retirement planning? Understanding the tax side of things seems pretty crucial, considering the different ways retirement accounts are taxed and the possibility of tax laws changing down the line.
The taxes (and advisory fees) have to come out of the 4%/amount withdrawn from the portfolio. So if you really need to spend $100K, and getting that $100K is going to cost $20K in taxes, you’d better have a portfolio large enough to sustain $120K withdrawals going forward.
Possibly the best single article I’ve read on SWR – thank you. So much wringing of hands and gnashing of teeth about a question to which the simple answer is, “just figure it out as you go, don’t do anything crazy, and you’ll be fine.” (OK, that’s my version of it. 🙂 )
Next up for you: The guardrails approach and Monte Carlo – why use 90% when 50% is fine?
Thanks, glad you liked it.
I think an article summarizing the dozen or so various approaches would be useful, but I feel like THIS ONE needed to be written first. The various methods are trees. THIS ONE is the forest. Especially since “just eyeballing it” might turn out to be at least as good or even better than most or all of the methods like the Guardrails approach.
The thing about the 4% rule is that no one actually uses it. It’s a rough guideline but people spend what they spend and that’s going to be over and under 4%.
Christian, I agree that few if any strictly adhere to a SWR annual spending target in retirement. I think it serves more like a “what can I afford” benchmark (relative to the SWR model) to help “live within your means.” It’s much easier to know whether you are living within your means when you are working for income than when retired. SWR provides a benchmark to tell you “is my spending this year increasing or decreasing my risk of having my money last through retirement”.
There is a big lie that you need more money for expensive health care as you age. That is part of the fear so people never feel it is quite enough in the US and keep calculating the SWR for future risk assessment, because in the US a semi-decent nursing home costs at least $10k a month, or when aging the back/knee/wrist/shoulder/heart hurts and surgery and rehab cost a ton, etc. Most chronical diseases could be addressed by reading Dr. John Sarno’s and Dr. Gabor Mate’s books. If one truly understands their emotions that have direct impact on physiology and immune systems, one does not need the expensive medical technology or drugs to address the symptoms. They would be able to eliminate the symptoms and address the root causes of the diseases. FIRE would feel more liberated and people don’t have to be kidnapped in the world’s most ridiculous, expensive yet never solving any root problems type of health care systems. By the way, not dragging forever with expensive healthcare is also very sustainable to the planet.
The above comment proves that non-physicians read this web site as well.
Your emotions don’t cause cancer, arthritic joints, Parkinsons disease, dementia etc, although an optimistic outlook on life is certainly helpful for coping. The US healthcare system is the best in the world, and that’s also why it’s the most expensive. Public health is a different matter, because that depends on behavior out of a physicians control: teen pregnancy, obesity, smoking, violence, addictions etc.
Bottom line: I doubt we will see you in the Philippines or at a psychologists office when you fall and fracture your osteoporotic hip!
US healthcare system is the most expensive but definitely not the best for average people. The best doesn’t have to be the most expensive, and the most expensive doesn’t mean it is the best. Dr. Sarno is an MD and so is Dr. Mate. If you read their books carefully and apply your medical knowledge, it would make sense. It is not about positive outlook on life but about truly understanding one’s unconscious minds. Physicians don’t know more about making a person healthier but definitely have detailed knowledge of advanced tests, drugs, procedures, which in some cases help but in a lot of cases do not. If a person falls and is injured, of course they need healthcare, but not expensive health care like in the US. The most expensive health care expenses are on chronical diseases and those are preventable. If you visit some other countries like Japan, Singapore, Finland, etc. you will find much efficient and cheaper healthcare that take care of your problems right there.
All chronic diseases are not preventable. That’s ridiculous. Many are, of course. And mindset does matter (anxiety is a very expensive disease that consumes tons of health care.)
The US does some things better than anyone else but lots of things worse. The US is a great place to prevent colon cancer, for instance. Best in the world.
Many of the people I knew when I was at a boy’s “school” in the 1960’s are dead; drugs and suicide. (The facility was closed, the director sent to prison in the 1980’s – Edgemeade.) I am a survivor and work on my mental health every day; it’s exhausting and as specialists in PTSD tell me, and as I am finding out, it gets worse with age. Saying or even implying that our trauma now is self induced and causes our physical illness is not helpful in both our mental health and physical health.
This really highlights the unique benefits of social security as a quasi-inflation adjusted annuity.
In residency, I was exempted from social security because of my public hospital employer’s 401(a) Defined Contribution Social Security Replacement Plan. As an attending at a different hospital, i’m now back with a public hospital with a 401(a) Defined Contribution Social Security Replacement Plan.
For physicians choosing employers, are there general rules of thumb as to whether it is better to work for standard W2/1099 employers paying into social security, or to be exempt through a 401a SS replacement employer?
That issue would be very low on my list of priorities when comparing two jobs. Which one is better would depend on how much of a SS benefit you’re already getting and what the SS replacement employer is offering instead.
Pessimism is Sexy…if one were a real pessimist, one might spend nothing or everything since everything but pork and beans would be useless.
My family has deep roots in Virginia. The economic collapse surrounding Bacon’s Rebellion in 1676; the economic chaos of the American Revolution and the subsequent depression of the 1780’s and of course the Civil War where even gold, during the fall of Richmond in 1865 was useless (a person can’t eat gold). What had value was pork and beans and once the liquor warehouses were destroyed to prevent drunken mobs from looting, booze was valuable commodity.
The real pessimists ought to be far more worried about nuclear war than running out of money in retirement or being sued above policy limits. Many estimates are that there is a 10-85% chance of a nuclear war in the next 100 years. That’s a pretty huge risk, even at the lower end. And it’s entirely possible that a nuclear war not only wipes out most people in the Northern Hemisphere directly in the first hour, but that the ensuing decade long nuclear Winter leads to a extinction event for our species.
William Bernstein had a fun short post on this and other risks, and why there are diminishing returns when trying to improve chance of financial independence beyond 80%.
https://www.efficientfrontier.com/ef/901/hell3.htm
Yes, very familiar with that one and totally agree with it. There is little benefit shifting from something with an 85% success rate to a 95% success rate.
I think one thing the post overlooked was how much wiggle room/unplanned events you have too.
For instance, I plan to retire at 35 with an anual spending of $4,000/mo. I plan to live to 95 for planning purposes. But my core expenses come in at around $2,200/mo. So if I had to I could cut back almost $1,800/mo. So having a slightly higher withdrawal rate that I adjust every few years works well. I think people on the leaner side of their budget would need to adjust their withdrawal rate accordingly.
Another thing is comfort. I know that with an ~4% withdrawal rate over 60 years I have a decent chance of portfolio survival without adjustments. But I also have a lot of assumptions in my numbers. Currently I’m single but by some point in my retirement I may not be, I’ve tried to plan for those assumptions as best I can by lowering my withdrawal rate initially and then increasing it if/when these events happen.
Overall I’ve always suggested somewhere between 3.5%-4% for periods longer than 30 years if one is adjusting every couple of years though. Personally I shoot for a 3.5% rate initially since I have so many unknowns.
Hi EngineeringFIRE,
You seem to be younger (like me), so it’s refreshing to see your perspective. However, I have several questions for you:
1. Do you never, ever plan to get married, have kids, or get the white picket fence house in the suburbs with a dog? Living off of $4000 a month for perpetuity as a single guy seems like a great financial plan to die early and alone.
2. How are you planning for unexpected shocks? It seems like I get an unexpected expense like clockwork every month, ha – car battery, vet bill, wedding trip for a friend, my plants died from the winter freeze and I had to ask my landscaper to plant 10 new bushes, I need to buy girl scout cookies from the daughter of a coworker, etc. I’m a lot looser and more generous with my money than a typical FIRE archetype, but life happens every month.
3. Do you think $4000 is enough in retirement? My parents had to hire a health aide for several years and then eventually put my mom into assisted living for dementia. Those places start at $6,000 for a half decent location.
Hey Joel,
I am younger, 24 at the moment. But my logic is based on a few things.
1) I’d like to get married if I meet the right person. Kids are optional but a nice to have. That’s part of the reason for being so far above my “core expenses” in retirement. While $4000 isn’t a lot, it will be almost double what I’m spending (although my spending will increase in retirement) to help account for what a spouse/kids increase this by. I’m also assuming a potential spouse brings $0 in existing retirement savings, which is also unlikely. But I’d even go back to work/do part time for a few years to reinforce this number to reflect such an increase in spending.
As for housing, I’m planning to own in full in a low cost of living state unless interest rates on a mortgage are crazy low again. Although that $2000 covers maintence.
2) I have a healthy cushion each month. I have $1800 to cover unexpected expenses. Currently I have unexpected things come up every month. I had a slightly higher than expected tax bill last month, for instance. Plus I’m planning for a lower withdrawal rate so if I push it up closer to 4% for a month or a year or two it won’t be the end of the world. It’s kind of like what would happen now if I needed to make that expense.
3) $4000 will be from my portfolio alone so I’ll have some help from things like Social Security. Plus when I’m beginning on end of life care, it makes sense for me to adjust my withdrawal rate much higher, since it no longer needs to last 30-60 years.
This is also assuming I quit earning money the second I hit FI. I may continue to do some work on the side or some of my projects may net me money. If my spouse wants to continue working that is their choice.
I don’t factor social security into my planning just to make sure my retirement is fairly airtight. Although I am doubtful it will
Another note, I’m an engineer not a doctor, so I don’t make the big bucks. So for me a fat fire would be closed to the $5000-$6000 range anyway.
I planned around the old saying “no plan survives first contact with the enemy.” My plan is extremely robust, at least I think, so that when I first hit retirement I’m well set for unexpected shocks and can adjust accordingly.
Thanks, I need a laugh today 🙂
Sorry, but your plan is not “extremely robust”. No offense, but at 24 (when you wrote this) you have no idea what your life is going to be like. (Enjoy that! Your future is wider open than for us older folks.) You’re extremely naive because you haven’t had a lot of experiences yet (never been married, never had kids, never owned a house, maybe never been fired, never moved abroad, etc.). One example, you think your core expense prediction of $2200/month “covers maintence [sic]” of a house. A new roof or a new HVAC system could cost $20-30k, and there are home ownership costs that are much harder to predict than these. I’m seeing an average home maintenance cost of about $16k/year right now, but that’s a lumpy thing (some years it’s much less, while other years it’s _much_ more). If you get married, yes, your spouse likely will bring in income—but they also might say, “I want new kitchen cabinets, no, not those ones, these ones” and you spend $40k on that.
At your age, you shouldn’t be worrying about SWR. It’s not useful for you, because your future is so unconstrained right now that the noise of your future swamps any signal from a SWR analysis. Focus instead on your savings rate and your earnings. When you think you’re 5 years from retirement, maybe 10, conduct a spending analysis _then_ and start planning on withdrawal strategies.
Agreed. Spending is not as flat as the models tend to make it.
No way would I retire at 35 on an income of $4K a month though. Shoot, a majority of people on this site were barely beginning their earning careers at 35. Is there really no paid work you’re interested in that FIRE at 35 on $4K/month is more attractive to you? Especially with the possibility of dependents in the future?
I bet there is some sort of part-time work out there where you could have your cake and eat it too. I love working part time. All the benefits of work and all the benefits of FIRE with none of the downsides of either.
Jim, great post, but in the spirit of Final Four weekend, don’t show ’em your whole game! For all the time I’ve spent on this the only percentage that really matters is projected monthly total expenses to after tax monthly defined benefit retirement income. I can report 75% works really great, and the only SWR or ratio I care about with respect to my retirement accounts is what percentage of my RMD will go to QCDs. My experience in studying this is a defined benefit retirement asset is profoundly undervalued, probably because the valuation reference is often in relation to the cash surrender value that could be rolled into an IRA. I think that is mistaken and may be a device to acquire assets under management for fees, perhaps too cynical there. The valuation method I prefer is a defined benefit’s present value is the future value of the benefit divided by a reasonable rate of return times the probability of payout. We don’t count them as balance sheet assets, but they are footnoted because they are big numbers and should be known. What do you think? oh, and by the way, a great way to help stay the course.
When I have to console myself re being a low budget WCI reader- will never get to $5M I’m pretty certain- I calculate the present value of our pensions and declare myself almost as rich as some of the private IM docs 😉 . I like your formula, but what do you do about couples and pensions which end at death of one spouse (or both- my kids will inherit a lot less than someone with actual assets of what I can claim we have at the moment). At present since husband will only lose $1k/month where I lose >$3K/mo iif widowed, I calculate 2 numbers- ‘ours’ and ‘mine if widowed’. His PV if widowed will not be my problem.
The stupid thing about all of this is the inflation adjusting. Why not just a fixed, higher withdrawal rate that is NOT increased each year with inflation? Is everybody just going to be CPI watching and do the math each year and increase their 4% by 3.1% in year two…2.8% in year three, etc.?
I think Nick Magguilli linked to an article recently in this vain: something like a 6% (or even higher…I can’t remember the exact % in the article) withdrawal rate each year but not inflation adjusted. By definition you’d be withdrawing less of a portfolio when the market is down and more when the market is up like the Trinity Study and I believe the odds of running out of money are just as low.
All of you engineers can probably poke holes in this but to me the simplicity is appealing
I find your reply curious. I read the (awesome) article and had few nits with it. But emphasis on inflation could have been increased I thought. 4% unadjusted withdrawal with 2% inflation is MASSIVELY different from 4% corrected. The consequences are huge
Yes, and hence the static HIGHER withdrawal rate %. No one is arguing that a static rate would be as low as the 4% in the study
Why not turn the SWR on it’s head and instead of playing around with a 4% rate, look around with what you can lock in today’s environment? For example, I’ve seen recent SPIA rates between 7 to 8%. Don’t know why you wouldn’t take a guaranteed income at 2x the withdrawal rates of the SWR. With today’s income investment options of REIT’s, BDC’s, options, etc. should be able to come up with a much higher blended rate than 4%.
Charlie, a couple points here.
To compare a SPIA to SWR, you have to look at an inflation adjusted annuity, since SWR includes inflation adjustment. Even with modest inflation, that is a major effect over 30 years. There are no inflation adjusted commercially available SPIAs anymore (delaying Social Security to 70 is the only inflation adjusted “annuity” available), so the closest available commercially would be with a 2% or 3% fixed COLA SPIA (roughly in line with average inflation).
Second, SWR comes with optionality (choice to spend more or less each year) and may leave inheritance, whereas SPIAs comes with reduced risk but without optionality or inheritance. So, depends on the individual, which they value more.
As to REITs, BDCs, etc, generally you’re paid more to take on more risk. 4% SWR is already based on blended stock and bond returns, which are much higher than 4% average return, but drops to 4% withdrawal because of Sequence Of Returns Risk (SORR). You can’t estimate SWR based purely on investment returns from risky investments, as it ignores SORR.
I agree with your first couple of points, but it doesn’t need to be a binary choice of doing all of your retirement or not with an SPIA (i.e. you can build a blended portfolio with just a portion dedicated to an SPIA). You are sacrificing optionality now in exchange for a guarantee income later.
Your sequence of returns risk is false under these alternative investments if your not never sell your portfolio, but instead living on generated income. SORR applies when you are selling your portfolio into a down market and not able to recover the capital investment.
I’d still challenge 4% in today’s environment. Too many people do the quick math and realize that just to generate $40K of passive income they need $1M under the SWR method and then think there’s no way they can retire.
Charlie, I agree with much of your response.
I agree SPIA isn’t an all or nothing choice. I’d start with delaying Social Security (SS) to 70 (for individual or higher wage earner in couple) before considering a SPIA to cover any desired reliable income beyond SS, as that is inflation adjusted, comes with joint survivorship, and no sales commission, but otherwise agree, and it is definitely not a binary choice. I’d point to Retirement Researcher’s Retirement Income Style Awareness (RISA) assessment as a way to determine suitability of reliable income vs total returns approach based on individual preferences. People differ on these tradeoffs.
Regarding SORR not being applicable to living off generated income only and not touching principal, I agree – with a caveat. Assuming income generated by risky investments, the returns aren’t guaranteed, and fluctuate with market conditions. So, relying on a fixed (or minimum) return from a variable return investment is still risky (depending on the degree those investment returns fluctuate with market conditions). Generally alternative investments are riskier than normal stocks and bonds, which is why they often are available only to accredited investors (as accredited investors are believed to better be able to evaluate the risk).
As to fixed 4% SWR, I don’t think a quick rule of thumb is the best approach. Rather, I would recommend estimating individual SWR based on Big ERN’s SWR calculator/spreadsheet, which allows factoring in SS, pensions, annuities, and any other reliable income cash flows and expenses, planning intervals (years money needs to last), and years until planned retirement, to get a much more personalized SWR (it takes into account both historical and future projections based on current valuations and interest rates). The SWR calculator sometimes come up with less than 4%, but sometimes much higher, based on SS, pensions, years from retirement, etc. But I certainly would not “wet finger” estimate a SWR, as that is just too risky to be way off, discovered at a time when earned income is impractical.
It’s only risky if you can’t adjust. The more flexibility you have, the less risky it is to “wet finger” it. If you only need your SS + 1% of portfolio to meet your fixed expenses, you’ve got a lot of flexibility. If you need SS + 5% to meet your fixed expenses, you don’t have any flexibility at all.
Jim, agreed, it depends a lot on spending flexibility. The lower a WR needed to support desired spending, the lower the risk, and the more flexibility, allowing less precise WR management.
With a lot of spending flexibility once retired, wet finger estimating is likely fine. Even so, I’d want rigorous estimates (backed by a second analysis using a different methodology) prior to making the often hard-to-reverse decision to retire (e.g. careful SWR analysis to determine if you have enough to safely retire), but being more flexible after retiring, as long as WR was sufficiently below 4% to not require the rigor.
Jim, agreed, it depends on spending flexibility.
If you only need 1%-2% WR to support desired spending, you don’t need to be rigorous.
But before making the often hard-to-reverse decision to retire, doing rigorous SWR analysis (backed by a second analysis using a different methodology) would still be prudent.
SPIAs aren’t indexed to inflation. So 8% now might look great, but 8% vs 4%+ inflation might not look so good in 20 or 30 years.
Be careful with an income focused approach. It has its downsides, both in accumulation and decumulation.
https://www.whitecoatinvestor.com/the-pros-and-cons-of-income-investing/
Jim, you got me on the decimal point precision. But given the way you’ve characterized me in this article, I think you might have forgotten what we talked about when you interviewed me at the Boglehead’s Conference last year. Here’s a few highlights…
Choosing a Portfolio Distribution Strategy
1. Inflation-Adjusted Amounts: It’s a baseline for comparison, but not efficient or advisable in practice. Others do better
…
Variable Spending Strategies with Calibrated Downside Risk
Spending Strategy Initial Spending Rate
Inflation-Adjusted Amounts (BASELINE)
3.62%
(3.83% w/ 10% failure rate)
Fixed Percentage Rule
8.54%
Dollar Floor-and-Ceiling Rule
4.14%
Ratcheting Rule
3.59%
Spending Guardrails Rule
4.53%
Inflation Rule
4.67%
Modified RMD Rule (Adjustment Factor: 1.56x)
4.25%
Also, I talked a lot about how taxes do not work as constant inflation-adjusted amounts, so that even if you were using one of these SWRs on a pre-tax basis, it’s going to end up being quite variable on a post-tax basis. And most people will have varying income flows and fluctuating expenses, so SWRs can’t really translate into actionable strategies. I don’t think anyone uses this in practice. Financial planners tend to use software that test a set of spending liabilities against the asset base with the software’s capital market assumptions. Knowing what the SWR is in such software can only be reverse engineered by vastly simplifying the real world information people are able to enter into the software. SWRs are really just an artifact for debates and discussion that only lives on in places like the Boglehead’s Forum.
Wade-
It’s great to have you commenting here and awesome chatting with you at Bogleheads last year. I think this post may have even been written before that chat. And now that I see you and Big ERN hanging out in the comments I’m left to reflect on the fact that I should have been more complimentary toward both of you and your overall work than I was in the blog post itself, and feeling a little ashamed that I was not. I think some of your best work has been building the framework on which to place the various decumulation strategies. You have taken an incredibly complicated subject and simplified it down to an understandable framework. You’ve made it as simple as it can be, but not simpler.
The problem I see happen to some people (and not necessarily you) is that they get lost in the details of the various strategies. They don’t see the forest for the trees. They think that if they crunch the numbers long enough they can figure out the perfectly optimized strategy which, in reality, can only be known in retrospect.
I don’t really have a problem with any of your work. As you’ve explained before, you put your assumptions out there and if someone disagrees with them, they can change the assumptions and crunch the numbers themselves. The problem I have is with what some people do with your work. For example, some insurance companies point to your work with whole life and annuities and use it to sell those products inappropriately. Likewise, some “engineer types” (it really isn’t occupation specific) see you come up with a 2.6% or whatever SWR rate and then try to convince everybody else that “starting at 4% and adjusting as you go” is totally cavalier.
I’m not sure you and I (and maybe everybody else) use the term “SWR” exactly the same way either and maybe that’s the issue. Maybe some people are looking at it as more of a guarantee than I am. When I say something is “safe” I mean “it works most of the time and you can see it not working early enough to make some adjustments that will make it work.”
I plan to do some follow up posts about the various strategies people who want something more formal than “start around 4% and adjust as you go” can use and their pluses and minuses (likely using your framework) but I felt like this “don’t miss the forest for the trees” article needed to be run first. I’ve got another one coming that points out that most of what people see that makes decumulation hard is really only hard for those who didn’t do a very good job of accumulation.
I totally agree with you that this discussion mostly just lives in places like the Bogleheads forum and this post is meant to push back against the false precision and pessimism seen in those discussions.
Thanks for all your great work.
As an engineer who really enjoys your work, I just have one question. What safety factor would you recommend using on my 3.74% SWR? 😉
I love it. It’s obviously unfair to paint all engineers with one brush of course. That bit was supposed to be just a little humor.
I can’t believe nobody is answering his question – after many calculations and backtests with the utmost precision, if he sets his SWR at 3.739877922355% he is almost certainly assured of never running out of money. 🙂
Jim, you make some excellent points, such as delaying Social Security to 70, possibly with additional annuity or TIPS ladder to reduce risk, precision exceeding accuracy, that SWR is most helpful for figuring out how much you need to retire, and your simple approach of start with 4% (I would recommend 3%-3.5% for FIRE adherents, though) and adjust.
That said, I see a few issues with what you said too. First, while I’m sure you’re a great doctor, financial analysis is not your *professional* expertise, whereas BigERN and Wade Pfau both have Ph.Ds in fields far more relevant to evaluating the 4% rule (guideline) than medicine. BigERN was a quant for the Federal Reserve, and Wade Pfau created the Retirement Income Certified Professional (RICP) professional designation for financial advisors. He literally defined the field of Retirement Income financial planning. Frankly, I’m far more likely to trust their analysis than yours (when you disagree) on retirement income finance topics, just as I would be far more likely to trust yours than theirs on medical topics.
Conservative retirement financial planning makes sense for many reasons. Most people underestimate remaining life expectancy at current or retirement age, especially for a couple, and ignore that there is a 50% chance of living longer than life expectancy (who would want a retiremement income plan with a 50% failure probability?, but planning only for life expectancy does just that). Things go wrong, and it becomes harder and harder to earn substantial income the older you get and the longer you’ve been out of the workforce. So, it makes sense to build healthy safety margins into any plan without assuming post retirement earned income (post retirement earned income becomes a bonus, rather than a plan dependency).
I would encourage you to read one specific blog post from BigERNs SWR series: https://earlyretirementnow.com/2023/06/16/flexibility-swr-series-part-58/, as it shows why flexibility in withdrawal rates is highly overrated. In a real SORR scenario, drastic cuts in spending would be required to recover.
BigERN isn’t nearly as pessimistic as you think. Using his SWR spreadsheet/calculator, for not so early retirees with decent Social Security and/or pension benefits, his calculator spits out quite high SWRs (way above 4%).
He also makes the valid argument (in response to my same pushback as you said about the limited dataset), that the number of effectively independent datasets is more than we think because only the first decade or so has a big effect on SORR, so it is the number of first decades in the dataset that matter in terms of independence, not the number of 30 year periods. Plus, he uses both forward projection and backwards looking data for his analysis.
Wade’s numbers were very low fairly recently because interest rates were at rock bottom values, so the fixed portion of the portfolio was severely hampered at the same time that stock valuations were at historic highs. Interest rates have recovered since then, as have his estimates. Morningstar does similar SWR estimates, which are reasonably in line with Wade’s (both improving as interest rates returned to more normal levels).
Thanks for your great comment. I agree their work is excellent and deserving of far more praise than it received in the short mention I had of it in the rant above.
It’s hard to disagree with a word like “conservative” without you actually defining it. If you mean that dude retiring at 57 with a 52 year old wife might not want to start at 4% but instead at 3.5%, fine. If you mean 2% is a reasonable place to start for someone who would prefer to spend more, I think that’s bonkers.
I’ll get into more in depth discussions about things like you are mentioning in the second half of your column in future posts. But I agree that in a drastic SORR situation, drastic cuts might be required. But I think those cuts probably won’t be as hard as you might imagine for many retirees. Let’s say someone starts at 6% and then has some nasty returns. So they cut their withdrawals to 2% for a few years. Pretty drastic, right? Well, not if SS makes up 2/3 of their spending and all of their fixed spending. So they go on one international vacation instead of three each year for 4 years. Not exactly eating Alpo.
I think we’re fairly close on what I meant by conservative (more your first example than your second), though I’d recommend generating a tailored SWR from Big ERNs calculator/spreadsheet, rather than just picking a set percentage SWR.
On how drastic the cut might be, certainly it depends on whether essential expenses are covered by reliable income or not. For someone relying primarily on reliable income, not a big deal. For someone relying primarily on total returns from an investment portfolio, drastic cuts would be drastic.
I have now started to base my annual withdrawals in retirement using a very simple method suggested by Jonathan Clemens of the Humbledollar.com : take your net worth on January 1st, and just multiply it by between 4-5%. There’s your spending money for the year ! Repeat the first of each year. Simple and clean. You can’t run out of money, and lets you spend more in the good times and less in the not so good times.
1) I have chosen to use 4.25%
2) as one’s net worth can fluctuate quite a bit over a given year, in concert with market gyrations, I multiply 4.25% x my average net worth at 6 month intervals ie, spending for the first half of 2024 will be 4.25% x the average of my net worth on 1/1/23, 7/1/23 and 1/1/24. When we get to July of 2024, it will be 4.25% x the average of my net worth on 7/1/23, 1/1/24 and 7/1/24. By using the average of several 6 month values, it keeps spending more aligned with what the market/net worth is at, as it can obviously change in either direction rapidly over a short period of time.
One of many reasonable methods for those who feel a need for a method. Probably overly conservative (most who use a percentage of their current portfolio pick something higher than 4%) and a little weird that it is based on net worth rather than portfolio size. I mean, consider two people one of whom has 90% of their net worth in their house and the other with 10% of their net worth in their house.
You are correct – I used net worth above, rather than portfolio size – i don’t remember which he used in his article . I meant portfolio size. I removed home equity from my calculations years ago after realizing it was not a liquid asset which readily provides retirement income.
The RMD portfolio spending method is your age-based annual RMD % applied to each recent annual portfolio value, plus spending dividends and interest. Due to inherited IRAs that are required to be spent down, there are RMD percentages for all ages. The RMD spending method was developed by Sun and Webb at Boston College’s Center for Retirement Research.
Note how the retiree can see next year’s spending amount vary as this year’s portfolio value fluctuates, so there is not a big surprise on December 31st . I welcome those annual spending adjustments instead of the threat of needing a significant withdrawal change at some later date.
Delaying Social Security (with COLA!) until age 70 has also been helpful.
I love the simplicity of the RMD method. One of my favorites after “start around 4% and adjust as you go”.
When RMD based spending is combined with delaying Social Security (for the SS recipient with the higher Primary Insurance Amount/benefit), this has been called the Spend Safely in Retirement Strategy (SSiRS) by the Stanford Center on Longevity, who found it compared favorably against 292 retirement spending strategies. The approach inherently backloads spending (which protects against SORR), as that is how RMDs work, so it may result in too low early year incomes for FIRE adherents, and others who want to follow the normal spending smile (spend the most in early go-go years). Stanford’s study offered some variations to reduce the backloading effect somewhat.
For early retirees, I think SSiRS would be most useful only after RMDs actually kick in, due to a combination of its simplicity and that by then, withdrawal percentages should be high enough to somewhat counter the front loading effect.
Fixed inflation adjusted SWRs come closer to matching the retirement spending smile (especially if assuming 1% less than inflation for the inflation rate and with an LTC plan in place, and some reserves assets available for rising medical expenses).
So, one option would be to start by using inflation indexed SWR (as a “spend within my means” guideline, not a strict rule) and switch to SSiRS after RMDs kick in (by which time, so would Social Security).
The Stanford study I referenced above is here: https://longevity.stanford.edu/viability-of-the-spend-safely-in-retirement-strategy/
292 different strategies. Unbelievable. It’s become like the investing factor thing. Every academic or financial advisor seems to need their own perfectly optimized strategy. No wonder everyone is so confused.
I believe several of the strategies were small variations of each other, so far fewer major strategies than it sounds like. That said, the purpose of the study was to simplify by seeing if there was a “best” strategy. Of course, best isn’t really possible because different strategies optimize for different tradeoffs at the expense of other factors. But, they concluded SSiRS strikes a pretty good balance for a lot of people (“compares favorably”).
As a licensed professional engineer I resemble those remarks! One of my wiser mentors once told me that engineers would weigh something on a hay scale (crude scales that only gave an approximate weight), and then correct for barometric pressure.
Stereotypes are fun! Sometimes they’re even true.
It seems to me that any “SWR” model especially if trying to maximize the rate of withdrawal needs to have a periodic “course correction” built in. Especially if the person does not have substantial annuity,Social Security,or pension income to provide a backdrop.
David, Big ERN offers a very simple adjustment model to maximize withdrawal. Re-run the SWR calculation in any future year you want to see if you can increase spending (his calculator/spreadsheet allows you to specify the number of years to plan for the money to last, so you can take into account fewer remaining years in future – I recommend planning well beyond life expectancy, though, as there would be a 50% chance you’d live longer). If it is higher, switch to new SWR as though you were just retiring at that time. If not, stick to the old one. Sort of works like an RMD based approach but allows for more spending earlier on.
I like this method because it allows you to adjust (usually up) as you go.
The point is just that you have to pay attention and be willing to adjust.
I agree.