The Ultimate Guide to Safe Withdrawal Rates – Part 24: Flexibility Myths vs. Reality

It’s been three months since the last post in the Withdrawal Rate Series! Nothing to worry about; this topic is still very much on my mind. Especially now that we’ll be out of a job within a few short weeks. I just confirmed that June 1 will be my last day at the office! Today’s topic is not entirely new: Flexibility! Many consider it the secret weapon against all the things that I’m worried about right now: sequence risk and running out of money in retirement. But you can call me a skeptic and I like to bust some of the myths surrounding the flexibility mantra today. So, here are my “favorite” flexibility myths…

Myth #1: Flexibility is cheap

We just have to be flexible. Great! How flexible, though? Reduce withdrawals by 1%? Or 10%? Or 100%? Obviously not 100%, because that’s the level of “flexibility” inherent in running out of money with the 4% Rule. But how substantial would the flexibility have to be?

Well, let’s look at the two classic historical retirement cohorts where the 4% rule didn’t work: The Great Depression and the 1970s/80s. Some of the worst failures of the 4% Rule occurred for the September 1929 cohort and the January 1966 cohort. And now let’s do the following thought experiment: How much more money would we need to add to the initial portfolio to make the $40,000 annual withdrawals last the entire 50-year horizon, and not run out of money halfway through retirement? I got good news and bad news: To roughly double the lifespan of the retirement portfolio we need a lot less than double the initial portfolio value. Makes perfect sense because of the time value of money: the withdrawals in the second half of the 50-year retirement horizon require a lot less initial capital. But one would still need $226,019 in additional funds in the case of the 1929 cohort and $145,579 to make ends meet for 50 years. Double that money for the folks who want to withdraw $80k out of a $2m portfolio. That’s a substantial sum, so let’s forget about any kind of cockamamie flexibility rule like “I’ll forego the Starbucks Lattes for a year” or “I will work some Uber/Lyft weekend shifts for a year!” Unless those Lattes and Uber shifts can generate a six-figure sum over a year!

With a 4% Rule, we’d have run out of money only about half-way through the 50-year retirement. It took “only” between $146k and $226k make the $40k annual withdrawals last the entire 50 years! (assuming monthly withdrawals, 80/20 portfolio)

But that said, here’s one flexibility rule that I think will save your retirement: The flexibility to work one or two extra years past that 25x annual consumption target. That would be one form of flexibility that I have always wholeheartedly supported. That’s what I did and it wasn’t so bad! Flexibility is easy pre-retirement. Flexibility post-retirement becomes a little more challenging once you go past the hand-waving and wishful thinking and look at some hard numbers!

In any case, here’s how the time series of portfolio values would have looked like for the two retirement cohorts with and without the additional funds, see the chart below. Notice the difference the cash infusion at the beginning would have made: It would have translated into about $400,000 in year 23 and 28, respectively, and that was enough to make the portfolio last until year 50! Pretty cool! Even though, for me personally, this wouldn’t have been a very stress-free retirement either: having drawn down the portfolio by 60% half-way through early retirement might look a bit scary if you didn’t know about the subsequently strong returns that lifted the portfolio again!

What a different the small increase in the initial net worth makes. increasing the initial net worth by 14.5-22.6% will roughly double the lifespan of the retirement portfolio!

Myth #2: Flexibility has to last only as long as the downturn

How long do I have to be flexible? That’s pretty pertinent information! Though nobody in the flexibility crowd seems to really care very much about this issue. Or could it just be that one has to tighten the belt only for as long as the bear market lasts? Maybe a few years?! Sounds intuitive, right? But there are (at least) two flaws with this logic:

  1. Inflation! In nominal terms your portfolio recovers. But remember that your withdrawals are adjusted for inflation. Thus, the portfolio has to recover in real inflation-adjusted terms, too. So, it’s true that a sample retirement portfolio with 80% stocks and 20% bonds would have experienced some drawdowns in 2002 and 2008/9 it now would have gained almost 180%. See the blue line in the chart below. But the real return was lower and the drawdowns would have lasted longer when taking into account inflation, see the green line, being mostly underwater between 1999 and early 2012!
  2. Withdrawals! Your portfolio would have only recovered so fast if you hadn’t withdrawn anything. Unless you’d have found a supplemental income source to fund your entire (!) early retirement budget between 1999 and 2012, the portfolio would have been dragged down by withdrawals. And those withdrawals during drawdowns are especially painful, compliments of Sequence of Return Risk! So, the red line in the chart below simulates a constant 4% withdrawal from this portfolio (0.333% every month, recalculated every month) and the result isn’t that impressive anymore: Even though the portfolio excluding withdrawals might have recovered, the one taking into account the withdrawals is still underwater by the end of 2017!
Simulated portfolio values for the January 2000 cohort. Ignoring the withdrawals, portfolio drawdowns look quite benign. But the portfolio of the 2000 cohort would still be under-water when factoring in inflation and withdrawals!

So, in summary, people who adhere to this flexibility mantra might suggest that one could have weathered the last 18 years pretty easily by tightening the belt during the two recessions (one lasted only 8 months, the second lasted 18 months). But that’s bogus! You would have tightened the belt, measured as the inflation-adjusted withdrawals for the entire 18 years!     (Side note: If this finding sounds familiar, it’s a similar logic to the one we pointed out in the SWR Series Part 12 on why a cash cushion doesn’t really work)

Myth #3: I can always be flexible in some other way that you haven’t even considered yet!

Discussing flexibility pitfalls with folks in the FIRE community is a bit like a game of whack-a-mole; after you show that one approach to flexibility would have failed in historical simulations people come up with another method. I can fix that! How about I list a bunch of different flexibility schemes and simulate them all for a few episodes where the 4% Rule would have failed. Here are the simulation assumptions:

  • A 50-year horizon.
  • $1,000,000 initial portfolio, $40,000 annual spending target.
  • All portfolio values and consumption values are displayed in real, CPI-adjusted terms!
  • Two start dates: 9/1929 and 1/1966, i.e., months where the 4% Rule would have failed.
  • A portfolio comprised of 80% equities (S&P 500) and 20% government (10-year) bonds.
  • Assume 0.05% p.a. expense ratios.

Under the fixed and non-flexible 4% rule (withdraw 4% p.a. initially and then adjust by CPI inflation regardless of portfolio performance) we’ve already seen above that we’d run out of money after 23 and 28 years, for the 1929 and 1966 retirement cohorts, respectively. So let’s look at how different withdrawal patterns using “flexibility” would have fared:

  • Perfect Foresight: Adjust the withdrawal amount to exactly exhaust the capital after 50 years. This would imply $32,626 p.a. for the 1929 cohort and $34,917 for the 1966 cohort. Of course, this is more of thought experiment because no one would have ever known what was the “correct” initial withdrawal rate at that time. Consider this a baseline against which we measure the other flexible rules!
  • Tighten Belt (5Y): Adjust the withdrawal amount for 5 years only and then revert back to the baseline $40,000. Pick the initial withdrawal amount to exactly exhaust the money after 50 years. This is also a flexibility rule that can’t be implemented in real-time because nobody in 1929 and 1966 would have known how much you’ll have to tighten the belt to make the $40k p.a. withdrawals work for months 61 to 600. But it’s a good baseline to compare with!
  • Guyton-Klinger: This rule uses guardrails around the 4% withdrawal rate. We keep withdrawing the same amount (adjusted for inflation), but if the effective withdrawal rate (current withdrawal amount divided by current portfolio value) goes above 4.8% (i.e., the portfolio value has dropped enough) then we reduce the withdrawals by 10%. Likewise, if the effective withdrawal rate drops below 3.2% (because the portfolio did really well) then increase the withdrawal amount by 10%. It’s essential to note that the ratcheting up/down can (and will!) occur multiple times, so Guyton-Klinger will potentially reduce the withdrawal by much more than 10%, something that is sometimes lost in the discussion of GK! See Part 9 and Part 10 for more of my thoughts on this sometimes overrated flexibility approach.
  • Guardrail to work (30%): See our post from a few months ago (Part 23), i.e., whenever the portfolio falls to 30% under the target level, go back to work and make enough money to finance a 30% drop in withdrawals (or simply cut consumption or a combination of supplemental income plus consumption cut, worth $12,000 p.a.). Discontinue the side hustle when the portfolio recovered again to 80% of the target value. (Side note: Why not 100%? See the discussion on that issue in SWR Part 23!)
  • Guardrail to work (50%): Same as above, but wait until we drop to 50% under the target portfolio level and then get a job to cover 50% of the expenses. Set the upper guardrail (stop working) when we reached 60% of the target.
  • Inflation Rule: Similar to the Guardrail to Work approach, let’s assume if the portfolio drops by than 20% we start slowly (!) reducing the monthly real withdrawals by 0.1652% (2.00% annualized) to mimic an often suggested flexibility method: Simply don’t do inflation adjustments while the portfolio is underwater! But when the portfolio has recovered again start increasing the withdrawals by twice 4% faster than inflation to ratchet up consumption again!
  • Constant 4%: The simplest flexible rule. Simply withdraw an amount equal to 4% annualized (0.333% per month) of the portfolio value of the previous month’s end. This guarantees that you’ll never completely run out of money. But the withdrawals might be seriously depressed for extended periods, see the calculations for Myth #2 above!
  • Bogleheads VPW: Similar to the Constant 4% rule but we increase the withdrawal rate to take into account that the retirement horizon is getting shorter and shorter. This will allow higher withdrawal rates but also engineer a complete portfolio depletion as the retiree gets older!
  • CAPE-based rule (a=1.5%, b=0.50): Look up the Shiller CAPE value from last month, invert it to generate the cyclically-adjusted earnings yield (CAEY) and use an annualized withdrawal rate of WR-annual = 0.015 + 0.5xCAEY. Divide this by 12, though, because this is a simulation at monthly frequency!
  • CAPE-based rule (a=1.5%, b=0.50) with VPW adjustment: Same as the CAPE based rule, above but also walk up the intercept parameter “a” by the same amount as the VPW. That’s because the other CAPE-based rule with a constant “a” parameter was calibrated to preserve capital. So all we’re doing here is to gear this rule deplete capital as the retiree gets older, just like VPW.

Let’s look withdrawals under the different flexibility rules. There are too many of them to display in one single chart so for each cohort, we’ll have two charts. Let’s start with the 1929 cohort, Part 1, see below:

  • As we already knew, the inflexible 4% rule runs out of money after only 23 years. So the baseline is $40,000 withdrawals for 276 months and zero dollars after that. Ouch!
  • With perfect foresight, we’d need to reduce the withdrawals to $32,626 over the entire 50 years…
  • … or tighten the belt for five years to about $12,400 withdrawals (-69%, ouch!!!). But again, these are not exactly viable flexibility rules because nobody would have known about the future returns!
  • The inflation rule would have slowly walked down the withdrawals by 2% every year, pretty much starting from the beginning. But over a much longer horizon that might not be palatable for the average early retiree. You would have slowly eroded your purchasing power over 30+ years to a little more than half the initial withdrawal. And then only started to recover between years 35 and 50.
  • The guardrail to work program would have been active for a total of 272 months (almost 23 years) if the side gig brings in $12,000 a year. Or 186 months (15.5 years) for the 50% version, i.e., with $20,000 supplemental income and/or consumption cut. Not a very pleasant way of spending my first 25 years of retirement! I did this to get away from work not to be forced to do side hustles for almost half my early retirement!
Simulated time series of withdrawal amounts under different flexibility rules. 80/20 portfolio. 1929 cohort, Part 1.

And for the 1929 cohort, Part 2 with the remaining 5 flexibility rules:

  • All rules would have suffered serious declines in withdrawal amounts for about 25 years. The Constant 4%, the Bogleheads VPW and the Guyton-Klinger approach would have tracked each other extremely closely. For most of the first 25 years in retirement, withdrawals would have been around $25,000 on average. 38% less than the $40,000 annual baseline! Then between years 25 and 45 you’d have withdrawn a lot again, only to decline back to $40k and even a bit below during the final 5 years of the simulation.
  • The CAPE-based rules experienced much smaller declines but they would have also started with much smaller withdrawals (around $30,000) due to the high initial CAPE ratio in 1929! And they wouldn’t have risen quite as much during years 25-45. Overall not a pretty picture!
Simulated time series of withdrawal amounts under different flexibility rules. 80/20 portfolio. 1929 cohort, Part 2.

Let’s look at the other cohort: January 1966. Again start with the same first subset of the flexible rules, see chart below:

  • With the benefit of perfect foresight, a constant withdrawal amount of about $35,000 or a temporary “tighten the belt” rule dropping the withdrawals to only $12,000 would have ensured the survival of this portfolio for the entire 50 years.
  • The inflation rule would not have kicked in until about 5 years into the retirement. But then would have eventually dragged down withdrawals to about $24,000 per year (-40%!) after 30 years. Withdrawals would have recovered to their initial $40k only by year 48!
  • The guardrail to work program would have been in effect for a total of more than 16 years for the 30% version and more than 12 years for the more aggressive 50% version. Working in my 60s is not what I would have envisioned in my early retirement!
Simulated time series of withdrawal amounts under different flexibility rules. 80/20 portfolio. 1966 cohort, Part 1.

And on to the other 5 rules for the 1966 cohort, see chart below:

  • The constant 4%, VPW, and Guyton-Klinger dynamic withdrawal rules held up all right during the first five years. That’s because the real bad returns didn’t start until later in retirement. But the 1973-75 recession and the early 80s definitely took a toll. Just like during the Great Depression, withdrawal amounts will be depressed by around 50% for an extended period. 20 years into retirement there is a bit of a reprieve and the withdrawal amounts return to the $30-40k range. Still below the initial withdrawals but at least off the really nasty trough (in 1982). Only after about 30 years will we see withdrawals that are above initial amounts. Not a pretty picture!
  • The CAPE-based rules start at about $36k and are dragged down to about $25-27k. You held up better during the 1970s and 80s but in exchange, the drawdown in withdrawals lasts longer! There is no free lunch!
Simulated time series of withdrawal amounts under different flexibility rules. 80/20 portfolio. 1966 cohort, Part 2.

What about other rules to flexibly withdraw money? Well, I’ve looked at a lot of rules with very unpleasant withdrawal time series. Notice that it’s numerically impossible to come up with a new flexibility rule that would look better than one of “my” rules at every horizon (at least for a given passive asset allocation). Sure, one can find a rule that would look better in year X, but that means that it has to look worse in year Y. You already see some of this “squeezing a balloon” business in the rules I presented: The inflation rule does a better job at not dropping too drastically during the early part of retirement, in contrast to the back to work guardrail scheme or a VPW where you drop the withdrawals substantially during the first half of retirement. But then in the second half of retirement, it’s reversed: you no longer need the side gig and VPW withdrawals recovered, but the inflation rule is still depressing withdrawals! Pick your poison!

But just to be sure, I’ll be up for the challenge: If you can think of a better rule, let me know and I’ll include it here! We’re at close to 3,000+ words already and I got a lot more material. So, there will be a followup post with several more flexibility myths to bust and reader suggestions for other rules!

Thanks for stopping by today! Please leave your comments and suggestions below! Also, make sure you check out the other parts of the series, see here for a guide to the different parts so far!

Picture Credit: Wikimedia

116 thoughts on “The Ultimate Guide to Safe Withdrawal Rates – Part 24: Flexibility Myths vs. Reality

  1. Don’t know who said it but the phrase “bad luck is when lack of preparation meets reality” springs to mind. Too often the bad luck stories of being caught out by market downswings in early retirement paper over the cracks (crevasses?) of incomplete preparation.

    I read the title of your past, had formulated some initial comments before I read the first paragraph and then I find you hit the nail in the head on the section on the benefit of one/two more years work. Completely agree!! The additional benefit is that it is generally at (1) the peak of earnings for many folks. Throw on top of that (2) bonuses and/or stock options and (3) bull market tail winds, you have a trifecta of factors that build in a lot of flexibility if you already have a tidy-sized portfolio.

    The other factor we are building in is on lifestyle. If we need to forego the more opulent extended European trip or Asia adventure for a more frugal US-based trip, that is what we will do. Yes, first world problems, I hear ya…..

    Congrats on the June 4 date. Awesome. My last day is June 1. However, a (not so ) little nuance here – I have 24 unused vacation days so I will essentially take them through July 10 and get paid in full for those ~5 weeks of vacation. Swing by work on July 10 to “officially finish” and hand back the lap-top, iPad and iPhone and building access card.

    Let the adventures begin!!

    1. Thanks! Completely agree! At least for those of us who’re not too desperate to leave, it’s best to stay flexible and work for a little longer. Will pay off handsomely later!
      So you Fritz and I will all retire within a few days. Awesome! Best of luck!!!

  2. FWIW, I think a lot of the talk about “flexibility” is bogus and not real. It mostly comes from two places:

    – Internet bloggers that retired on 3% SWRs (i.e. super conservative) and are making $40,000 from their FIRE blog anyway and their wife worked for 3 years after they retired and….all the talk about “flexibility” is ridiculous because it was never part of their plan and they’re just writing an article for the clicks.
    – People who are years away from FIRE and it is a purely theoretical thing for them. I’m not saying they are bad or lying or anything. Just….there are no real consequences for them staking out this position.

    In the real world — when you look at people who are right on the cusp of retirement — what we virtually always see is people working more. (That is, exactly what you suggest.) There’s a reason why One More Year gets written in capital letters because it happens so dang much……

    Because for a lot of people if they have to “be flexible” it defeats the purpose of retiring. They could have kept working (and making quite a lot of money, kept the company health insurance, and 401k match, and….) if they wanted to “be flexible” with their hobbies and vacations and whatnot.

    1. jp6v, that’s a pretty good assessment of the FIRE bloggers overall, but it’s up to you, a reader, to participate or not in their charade. Instead of wasting your energy and time posting messages against the cultivated mantra of the blog, you just ‘help’ or ignite even more clicks. What’s the point?
      I’ve stopped reading blogs of such cheerleaders. I prefer more conservative/realistic as this one and a couple more. So, I’d say I visit 2-3 blogs maximum per week nowadays.


      A great article! As always it’s a bit math heavy for my head, but I got the message OK, I think.
      Like others, I’m also curious whether if one starts at 3-3.25%, are they “safe” for 50 years?
      What if I ‘worry’ not to leave too much $ behind, can I be more aggressive and start at say 3.5-3.75%? Of course I don’t wish to die destitute either.
      Why did you choose to base your analysis on 80/20?
      What would happen if your analyzed cohorts had 70/30 or 60/40, the latter being Bogleheads’ favorite?

      Thanks for a great analysis, and comments were interesting too.

      1. Thanks!
        In most cases, where the (unknown) initial SWR was higher than 3.25% would have over-accumulated assets with the initial 3.25% SWR. So, resetting the withdrawals every once in a while if they slip too low would be a sensible approach. That resetting and ratcheting up withdrawals may happen…
        1) never (in 1929)
        2) only very late in the retirement (1966)
        3) or still many years into retirement in many other years.
        Only if there’s a very strong bull market would you adjust the withdrawals pretty quickly…

  3. The gist is you could have maintained retirement through the Great Depression and the Great Stagflation with about a 3.2% WR, right?

    Perhaps another interesting take would be the success rates of algorithms where you RE on a frugal budget (e.g. a 3% WR) and then decide 5 or 10 years later whether to continue with frugality or increase consumption (i.e. travel the world, upgrade housing, etc.). If markets/inflation have gone well during that time, the danger of running out of money might be greatly reduced by then. E.g. things looked sketchy for the 2010 cohort at the time, but not now. If markets/inflation have gone badly in those critical early years, you confidently continue with the WR that would have survived the Great Depression or Stagflation. Call this “upside flexibility”.

    1. What about starting with a 3% WR “floor”, and increasing it whenever it’s less than 3% of the portfolio? I’m having a hard time wrapping my head around the math, but it seems to me that this would be equivalent to “re-retiring” at a new 3% “constant” rate, and since 3% has never failed thus far, it’s safe to increase it to that amount, right?

      You could even figure out how long it typically takes to raise the rate to 4% of the original principal, and frame it as “tightening your belt” for a limited time at the start of retirement. If you’re retiring in 1929, then it’s good you kept to 3%; if you’re luckier, you only have to wait a little while before you’re spending at 4% or greater.

      1. That’s a good approach. Notice that 3% is below the failsafe rate. So it should work in both 1929 and 1966. My suspicion is that it would take a long time before you walk it up again because even at 3% you’d temporary deplete your portfolio by quite a bit. I will take a look at that approach and simulate it… Thanks!

    2. An analysis along these lines would be great to see. The earlier posts in this series are convincing enough that planning on a 4% rate is not so safe, at least (or especially) under some circumstances, such as the current valuation of the market. But if safety is the goal, having some kind of system to judge when it is likely to be safe to raise an initially low WR would be nice. The “Tighten Belt” baseline might be a starting point for thinking about this, but it’s not set up to be a particularly viable option in and of itself.

      The post itself was very helpful as I am more or less facing the one more year decision right now. Thanks!

      1. Thanks! Yes, and again the tighten belt approach is just to show how much you have to tighten the budget for a pretty long period (5Y). Do anything less than that and it’s lights out after 45 years. I’d rather do the 3% initially and then ratchet it up once it’s safe again. 🙂

      1. Karsten, I very much look forward to seeing you dive into this thread in more detail 🙂

  4. What about the strategy of holding on to 2 (or 3?) years worth of expenses in cash or a high-interest savings account? When the markets are good, you live off the stock/bond portfolio as usual. When the markets are poor, you switch to spending the cash reserves to avoid selling shares at the worst time during the bottom of a recession. Ideally, the worst of the market crash is over before your cash reserves run out, and your stocks are mostly able to recover before you start withdrawing from them again.

  5. Hey Big ERN!

    Thanks for including the 50/60 guardrail approach in this post! It’s great to see how this strategy stacks up to the competition. I understand your intuition that there are no free lunches, and it generally seems that if you want to increase your withdrawal rate to 4% or greater (from the failsafe 3.25%) you need to either accept some risk of portfolio failure or accept some risk side hustle/decreased withdrawals in the future.

    Once intuition that we all must share is that by designing for the tail-end scenarios of September 1929 and January 1966, we loose sight of the fact that there is a very small probability of either representing us as we retire today or in the future. One of my favorite posts is “The Ultimate Guide to Safe Withdrawal Rates – Part 22” where you point out that people working towards a specific retirement balance are more likely to choose a retirement date that exposes them to SORR than someone choosing a date at random (due to stock run-ups that get people to that number, followed by corrections around the corner).

    How can we build a model that captures this key fact? My situation is that my wife and I will be financially independent sometime in the next 10-15 years, and while we have high-earning jobs at the moment, we would like to have the flexibility to earn significantly less between now and then. A calculator that combines your side-hustle math with your probability of failure math would be the ultimate tool for us! Is this possible, and how would you suggest using both in conjunction to figure out our side-hustle parameters assuming say an acceptable 5% probability of portfolio failure? Consider this a shameless plug for one more case-study 😉

    1. Thanks! THat’s a hard problem. Ideally, one would model this as a dynamic withdrawal percentage. So, your net worth target is now a moving target equal to consumption divided by the CAPE-based SWR (see part 18). That would alleviate some of the risk of retiring at the peak.

  6. Your G-K algorithm is just one possibility in a whole framework, that includes the idea of putting a floor on withdrawal spending, so it does *not* necessarily suffer the problem of simpler variable withdrawal stategies, where your portfolio survives but only by reduces withdrawals by some very large amount. When I do historical simulations, it looks like 4% works out just fine in 1999 if you’re willing to drop your withdrawals by 3-4%/year and total 15-20% from original when your current WR gets too high. Yes, in 1929, and 1966 you end up dropping to 80% of your original spending and basically never get it back, while also reaching scary portfolio levels that get bailed out by high returns. But you make it.

    And in 80% of scenarios, you end up cutting less, or only for a few years. You end up raising your withdrawals as often as dropping them (assuming you build in the increase possibility).

    Note: the default parameters aren’t really very good. A key to making it work well is not raising your withdrawals until you’re seriously in the clear and even then doing it fairly slowly.

    In any case, if you’ve chosen a spending level that is 15-20% above what you *could* do, 4% with good G-K parameters will work just as well as a flat 3.25%. Admittedly that’s partly because 3.25% is about 80% of 4%, but with the G-K you only go down that far when you *need* to (plus a bit more).

    The other thing — you’re stating quickly that it’s always just a matter of one or two more years to get to a lower withdrawal rate, but that’s not always or even mostly the case. If you’re talking about the difference between 4% and 3.2%, that’s a 27.5% increase in your portfolio. That could certainly happen in one or two years, but unless you are saving a very high percentage of income (7-10 year FIRE plan) it’s going to require a pretty good market performance to get it done in 2 years, and a fantastic one to get it done in one.

    Also note, if you are actually in the 1929 scenario where you hit a 4% WR FI in 1929, Black Tuesday happens, and you have to keep working for quite some time before it comes back even to 25x again — assuming you kept your job during that mess. It’s not quite as ridiculous in the Dec. 1965 scenario, but it’s not pretty. You certainly don’t get to 32x spending in 1-2 years, unless you’re pounding away an extraordinary % of income.

    Finally, when I do some random trials of going back to work if your portfolio hits <X, even if you make 1/2 the money you were making before you retired, it's pretty uncommon that you need to work more years to get back to your original WR than you would have to go from 4% to 3% in the first place.

    I think you're making some blithe assumptions about what happens when you OMY that aren't really warranted. Don't get me wrong, I think if you like your job ok, there's no reason to super optimize your time being retired rather than just work to ultra-safety and probable old age generational wealth. But if you really want to get out, the balance of possibilities suggest that 4% is quite reasonable and 4.5-5% may involve an acceptable level of risk for a lot of people. If the goal is to maximize expected years not working for money, the answer is probably a bit *higher* than 4%, around 4.5%.

    1. Gnomeozurich, I like the framework you’re coming at this with. If the goal is to have complete financial independence, then you should focus on the worst-case scenario. But if your goal is to work less and let your money do the work instead, minimizing working years seems to be a better framework for analyzing strategies and withdrawal rates. After all, most who are interested in FIRE do in fact have the ability to work more, it’s just that they would prefer not to have to.

      The side-hustle worksheet that Big ERN has provided in Part 23 comes closest to capturing this goal, but it seems that there is so much more that could be done to optimize the approach. I love Big ERN’s requirement of a rules-based approach that doesn’t leave anything up to subjectivity.

    2. GK would not have performed very well in the 2000s. 1999 might have been a slightly less awful starting date, but the tail event of the Jan 2000 cohort would have slashed its consumption by 50%.

      1. I feel like you’ve ruled out G-K based on an incomplete look at its possibilities. The original paper discussed having a floor for withdrawal amount as part of the framework, yet you consistently dismiss GK and other strategies based on slashing consumption too far. Does your model have you running out of money fairly often if you put in a floor of 20% below original spending (which is my normal preference for a G-K model)? Mine does not, but I haven’t done the work of imputing returns to 30+ year periods starting before 1988 and I know 1999/2000 are very tough years.

        That said, they are also very tough years (like 1929) to not retire because you are pushing for a lower withdrawal rate. In 1929, 2000 and 2007, it takes a lot more than 1-2 years to go from 25x to 32x spending even if you are saving at a very high rate. At more moderate rates like 30%, 1929 doesn’t get where you want until ~1945. Instead, it gets oh so close in early 1937 and then crashes again. If you have a high enough savings rate, you do hit the mark in early 1937, but then the subsequent crash is so bad, I wonder if you’d really have the stomach to wait it out and not go back to work if you can, even though we know post-hoc that you make it. Anyway, my whole point is that it’s a cost benefit analysis between how long you have to continue working to hit that 3.2% vs. your chances of problems if you pull the trigger at 4%. And if you look carefully at the actual historical scenarios where 4% fails or comes really close, it’s generally no picnic getting to 3.2%, it’s a LONG wait.

        1. Gnomeo, you hit the nail on the head! One of the flaws in the “2-more-year strategy” to get from 4% to 3% SWR is that when you really need it, it doesn’t deliver (precisely because of the SORR)! Ostensibly working more and saving more doesn’t hurt your financial situation, but it doesn’t necessarily help that much if you’re already at 4% SWR.

          All this healthy discussion leads me to conclude two things (both of which may be incorrect!):

          1) If you do retire, and your portfolio takes a 50-90% hit, you shouldn’t pull your remaining money out, but you should definitely get back to work, period! It seems easy to overlook this when looking at historical data because we have designed our portfolio to withstand the Great Depression, for goodness sake! But imagine yourself in that situation, portfolio 75% in the hole, and deciding whether to keep going with your withdrawals. Maybe you should just not touch that portfolio for the time being…just in case it gets slightly worse. Would anyone argue for continued withdrawals and if so, on what grounds?

          2) Karsten’s approach to SWR rate is bar-none, the most comprehensive I’ve seen so far, but there is definitely something intriguing to the utility function approach that Gnomeozurich is hinting at. Can we use the same data to minimize our working years at the j*b, while taking into account similar withdrawal strategies that have been discussed here (G-K, CAPE, etc.) Maybe with a withdrawal floor of X% of the original SWR, and a back-to work trigger? Again, not something that guarantees you’ll make it through the 1929 and 1966 retirement dates, just something that gets your working years down as much as possible assuming X savings per year and a FI time horizon of Y-more years.

  7. A thought provoking post as always Karsten.

    I think McNeill’s comment raised an interesting point about if and when to increase the SWR.

    The great ERN SWR series arrives at a SWR of around 3.25% for 40+ year horizons assuming 80%+ allocation to equities. This would have survived the worst historical sequence of returns. But what about when we don’t get the very worst SOR (i.e. which is every other time!) when retirees could be safely withdrawing 3.5% or even 4% but we won’t know that when we start drawdown!

    Karsten, does the historical data in your SWR series answer whether if you start at the failsafe 3.25% for all cohorts, after how many years and/or at what portfolio value identifies the cohorts that were ‘out of the (SOR) woods’ and could then have safely nudged up to a new inflation adjusted SWR maintained for the remainder of the time horizon? I hope the question makes sense.

    1. Using the analysis done here — basically ensuring that no historical period would have failed (including simulations only for non-finished periods), a strategy of using 3.25% ratcheting up every year possible but never ratcheting down. (i.e. every year you withdraw *either* 3.25% of the new portfolio *or* an inflation adjustment from what you withdrew last year, whichever is *greater* will still pass — because each step up is just a new point from which we already know that 3.25% never historically fails.

      That said, such a strategy won’t actually be as safe as making this decision once and sticking to that inflation adjusted expense level. The problem is that if you do this, you are always going to face the most difficult series possible over your 50-60 years, because you’re essentially “re-retiring” every year. So a retirement in the 1980s or early 90s generally sails by at 3.25% or even 4-5%, but if you ratchet every year you can then you’re *also* effectively re-retiring in 2000, which is looking a bit dicey even for 3.5-4% at the moment and still could potentially fail for 3.25%. This is basically an extreme version of the problem discussed in Part 22 — where you’re far more likely to retire at times when a higher WR wouldn’t work and *most* likely to retire at exactly the times when you need a really low WR to survive. You’re always going to end up doing your last ratchet at the worst possible time.

      A monte carlo will also demonstrate that difference, although you give up the automatic yty correlations inherent in using only historical series, so the true answer to just how risky this will be is unclear.

      It’s probably safe *enough* to just ratchet straight away as long as you’re starting with a low rate like 3-3.5%, but then I’m of the opinion that withdrawal rates close to 4% are “safe enough” for most people, even though they are nowhere near certain because people *can* lower spending or go back to work even if they don’t really want to. If you want to be safer, pick a somewhat lower % like 3 or 2.75% as your ratchet trigger and that will probably be almost as safe. IOW, start at 3.25%, and every year take either an inflation adjustment *or* go to 2.75% of the new portfolio, whichever is bigger.

      This is likely to be very little more risky than a plain 3.25% rate, but will allow for very likely increasing expenses above inflation over time.

      I would love to see a model of this.

  8. BTW, I *love* this series and the work you’re doing. I read my initial comment and am worried I may come across as a hater, because I’ve had that criticism since I read most of this drawdown series a few months ago, but *long* after it didn’t make sense to comment on the original post about variable withdrawal plans.

    In the spirit of doing some research myself instead of insisting that you should do simulation that I think are worthwhile — I have struggled hard to find good data with simple google searches. I finally got pointed to the monthly series of Shiller at yale after a fair bit of following links more useful for high end quants doing back testing but requiring way more work to wrangle. Where does your bond data come from for before the mid 70s? Shiller has 10y treasury *yield* information, and I could use that to generate an expected total return from month to month just based on standard bond pricing models, but that would just be looking at 10y treasury and not a diversified bond portfolio the way the S&P 500 (or whatever proxy is being used before 1957) represents a diversified large cap portfolio for stocks.

    What are you doing in your simulations? Are you calculating probable returns based on the yields? Are you pro-rating the annual series from Shiller? Or are you using some other data source for bond returns?

    1. THanks! No offense taken! 🙂
      I once got a long snapshot of 10-year benchmark returns. These are rebalanced every few months, so a true picture of a 10-year Treasury ETF with returns consisting of both interest and duration effect (and roll-down effect, for the purists). Got the returns until 2013, then splice with IEF ETF returns. 🙂
      Feel free to use the returns (see Google sheet from Part 7).

  9. Gnomeozurich

    “start at 3.25%, and every year take either an inflation adjustment *or* go to 2.75% of the new portfolio, whichever is bigger.”

    A very interesting approach to a dynamic withdrawal rate. With both historical sequence data and MC analysis at sensible static WRs many cohorts finish with a much larger portfolio than they started with as a consequence rather than by design.

    “I would love to see a model of this.”

    Me too.

  10. Great post Karsten. An interesting follow up to your earlier post in the series.

    The issue with ‘flexibility’ is that, when you add it to the plan, it becomes the rule. Then you need to think about having flexibility on that flexibility. That leaves you with either: potentially, catering for a very worst case scenario (which you can never be certain won’t happen) or being tempted to break your rules and ultimately adopt a sub-optimal strategy.

    That said, I think (as I’ve mentioned in my post on the subject), flexibility can be a helpful tool if used the right way. That is through resilience and protection.

    By resilience, I mean adopting a lifestyle and outlook that can adapt to adverse environments. That is using a kind of variable withdrawal strategy, being open to changes in consumption or perhaps things such as go-arbitrage.

    By protection, I mean things such as insurance, guaranteed income (index linked annuities or such like) or an investment strategy with downside protection (like the one you explored in your recent post). I think that, generally speaking, the FI community is heavily geared towards self-insurance. But there are a number of very helpful insurance products and plans that can not only give great downside protection but also peace of mind. (I’m thinking of the case where you ‘side hussle’ for life!)

    1. THanks! But this is about semantics. Rules can be very flexible! I’m thinking here in terms of dynamic programming, Bellman Equations, etc.: that’s a whole field of engineering that was developed for that purpose that’s now used in finance and economics. So, we’re on the same page: flexibility as in optimally adjusting actions in response to a changing environment. That’s why I like the CAPE-based rules a lot! 🙂

      1. Thanks Karsten. We’re on the same page.

        Talking about a dynamic or flexible withdrawal rate: I’m sure that it is the right way to go. It makes the most sense conceptually – you make the withdrawal decision that maximises your future income stream. That ‘axiom’ is violated with a fixed withdrawal rate.

        I’ve always thought that there is some way of calculating a withdrawal rate using a Markov Decision Process. But I haven’t done any optimisation theory since university (besides I don’t think I’m half smart enough to do it!).

        I’m very thankful for the work you’ve done on the subject. Not only because it’s the best analysis out there. But also because it gets my brain biscuits whirring.

        Thanks again.

  11. Super interesting post (as always!). I agree with others that you’ve convinced many of us to start with a lower SWR than 4%. I’d love to see this analysis done with a moderately lower SWR, perhaps 3.5%. Thanks for the great work here!!

    1. Thanks! The problem with the 3.5% SWR is: Do you still apply the flexibility rules? Even under the 3.5% rule you’d have gone back to work in the guardrail to work scenario. Or you’d have reduced withdrawals with Guyton-Klinger. THat’s the scary part: All of the flexibility rules still create false alarms even if the initial SWR would have worked in the end!

  12. Thank you for another great post! I read some of the other FIRE blogs and can’t help but cringe at some of the comments from readers. It seems that flexibility is often the answer to an inconvenient fact. It would be great if some of those blog writers would address the vulnerability of the 4% rule especially when the time horizon is extended out past 30 years.
    My own plan is to retire in maybe 5 years and I have been favoring the bucket approach described in a book by James Cloonan. In short you remain invested in stocks with any cash not needed in the near term, say 5 years. For near term expenses you have a bucket with only safe investments. He thinks a reasonable approach is to have the safe bucket equal to 4 years of living expenses. Once annually you decide where to withdraw the next years expense, if the market is good you withdraw form stocks. If the market is down you withdraw from the safe bucket. He backtested this approach with good results including the Great Depression. It would go a long way to mitigate the Sequence of Return risk.
    Having only recently discovered this site I am loving the detailed analysis. Keep it up!

  13. ERN, thank you for continuing your series. Once you actually go on retirement, please don’t forget us.

    One of my big takeaways from this series is that a constant %-age (or maybe VPW) is actually likely to be the best choice for most married couples.

    It is pretty common for married couples to split responsibilities and interests. My wife does a lot of things, but she is not really mathematically inclined and while, naturally frugal, she isn’t finance-minded. So asking her to implement CAPE based withdrawals on her own is likely a recipe for failure; ditto dealing with calculating inflation, selecting which accounts to draw from, etc.

    I am 5 years older than my wife and men in my family mostly live into their 80s; but women in her family routinely live to their 90s. There is some risk (if we want to call it that) that better geriatric medicine will extend lifespan, so like you I am looking at potentially a very long horizon and thus tend to be quite conservative (we are targeting a 2.75% initial WR and climbing to 3% WR after the first 5Y; unlike you I do plan to implement a rising equity glidepath).

    Family history aside, sudden, unexpected deaths occur; as do medical issues leading to mental incapacity (e.g., a severe stroke or brain cancer). So people like me need a plan that my wife can step into and implement immediately if these things occur. A quarterly fixed %age (annualized 2.75-3%) or VPW seems like the best option for low-effort, low-training hand-off to a loved one.

    1. Mario, are you ramping your WR after 5 years to mitigate sequence of returns risk? Have you taken other steps along those lines like eliminating your mortgage? I’m also concerned about SRR and I’m considering starting with a low WR and increasing it slowly. But I haven’t performed enough analysis to be confident in how low to start my WR, how fast to ramp it up, and how high my final WR target should be.

      1. Pretty much. I expect a significant crash. That said, that’s not the only reason.

        The big issue is that my portfolio is almost entirely taxable and has very substantial gains since I went more than a decade without rebalancing it and it’s in a lot of bad choices. To rebalance now would involve taking a fairly huge tax hit due to California’s taxes (“temporarily” 13.3%). I don’t want to draw down a portfolio that’s as unbalanced as this one to the extent that I can.

        We plan to relocate to a state with much better capital gains tax treatment and lower cost of living, but it can take several years to get free of CA (the FTB believes they own you unless you can prove otherwise) even if you establish residency from the point of view of some other state. Because our portfolio is completely out of whack, we want to do a lower rate initially until we can rebalance. 0.25% may not seem significant, but it’s something.

        1. Wow! Good luck wrangling everything. I’m not expecting a crash (maybe several corrections / sideways weirdness), but I share the worries about a high CAPE.

    2. Interesting! A 2.75% SWR sounds like a very safe approach. Even with your longevity “problem”
      Chances are you’ll eventually walk that up to what would be a 3% SWR. Sounds like your way of being flexible is to keep the withdrawals at the lower level if things don’t work out. That’s the kind of flexibility I can definitely accept! 🙂

  14. Thanks again, ERN for all the great work! A question to you. Many in the investment community agree that both ST and LT returns going forward will be structurally lower than before (due to permanently lower interest rates, lower productivity growth, lower population growth, etc.) What do you think this means for the SWR? Eg. what if in this new world a person retires into smth similar to the 1966 scenario but then the roaring 80s do not come, or are somewhat less roaring? It’d be great to hear your thoughts on this.

    1. Great question! I agree that returns will be a bit more lean for the next 10 years. But if they are there is also a chance that PE ratios normalize again and returns for years 11+ will look good again. Maybe not 7% real returns, but 6% real equity returns (to account for a 1% lower GDP trend growth rate). THat’s still pretty decent!

  15. So the solution is to essentially go to a 3.3% withdrawal rate ($40K/$1.2M) instead of 4% to ensure someone can cover for 50 years for the failed scenarios, which is always a reasonable option in any withdrawal scenario to extend the years to make the math 100%. But you know, sometimes I feel we get so wrapped up in the “worst case” scenario we fail to recognize that such a scenario has a very low probability of occurring at all. If these two were the only failures of the 4% rate over the historical timeline of say 70 iterations, that’s a 3% chance of failure, or a 97% chance of success. Those are pretty good odds to bet and are probably better odds than you or I will even live to age 70. Seems to me we are looking for guarantees in an imperfect world. This is what causes someone who is 50 years old with annual spending of $50K and $3M in liquid assets to be worried they are going to run out of money in old age. I feel the biggest myth though is that someone retiring early simply never ever does anything that generates earned income again, particularly if they feel financially threatened. Seriously, the problem with the job market isn’t that there aren’t jobs available, it’s that most people need full time jobs that can fully support them. If someone is in a position where they only need minimum wage part time or full time seasonal work and are anywhere close to physically fit, there is no shortage of work that can easily garner $5-10K equivalent a year and knock down the annual withdrawal to 3-3.5%. Also even if someone who had a pretty reasonable salary for a decade or more does decide to never earn a salaried income again, they’ll get something from SS. While many don’t want to “count on it”, the myth is that it won’t be available even in a reduced capacity. If it dropped your withdrawal rate just 0.5% in your later years, it would none the less be a significant impact for your remaining 20-30 years of withdrawals. While it can be wise to look at the possible failures so you can mitigate if you have to, there comes a point where you cross from being practical to just going for overkill.

    1. Some of us want to “never ever [do] anything that generates earned income again”. If that’s a requirement, then Karsten’s analysis is not overkill.

      1. a serious issue for professionals is that the probability of resuming a career after being out of the field for 2 or more years is very low. as a fairly high seniority hardware engineer I can straight up say that two years in the wilderness would be very toxic for resuming at my current level (or even one or two levels below me).

        1. Excellent point!!! Same here! I work in finance and would not get my job back. I would be completely between two chairs: jobs with my skill level would be out of reach with a gap in the resume. Lower-skilled jobs where I’m vastly over-qualified wouldn’t be an option either. The employer would think I just do this job for a few weeks while looking for a better-suited job.

    2. The analysis in Part 22, suggests we should be a litle less sanguine about that “97% probability of success”, *especially* if you are aggressively saving to retire early and do not have anything unrelated to portfolio size triggering your retirement (like pension age or social security eligibility, etc.). You are *far* more likely to reach your FI number at a Market valuation peak or plateau than in a down cycle, and will never reach it at the bottom of a big bear market or crash. So the set of historical times when you could have retired will always contain all the potential failure points (1929/1966/1999), but won’t contain any of the easiest rides (1974, 1932 1940 2009 etc.). If you look only at realistic times to retire, the failure rate of 4% is quite a bit higher than 3-5% probably more like 15-20% for 40-60 year retirements. That said, I’m still sanguine about ~4% for various reasons.

      1. The potential failure rate of the whole political and economic system of our country/culture over 40-60 years is probably at least 10%. The chance of you dying before 25-30 years has passed, even if you are healthy now and in your 30s/40s, is another ~5-10% and this number is higher the older you are when you FIRE. That means that any failure rate within the system of under about 5% is pretty much “good enough”, and the actual chance of “running out of money due to portfolio underperformance” considering all relevant possibilities does not increase by a high fraction of the modeled failure rate difference, until you’re in the 15-20% range.

      2. Most people will still make some money in retirement if they are retiring earlier than 55ish but few are actually planning on this in their retirement modeling.

      3. Most people who are capable of retiring early are on the frugal side and probably won’t increase their expenses at the published rate of inflation (which encodes a small amount of hedonic adaptation).

      4. If your budget is very lean, it’s easy to take care of a LOT of it by going back to work even at a part time low-paying job. If your budget very comfortable, unless most of the costs are hard-baked in, it’s usually not that hard to cut it by 10-20% if you start to realize you’re in a potential failure mode and 2 and 3 aren’t doing enough for you.

      5. Most people retiring before age 50+ are modeling without considering social security or similar old age pensions in other countries. This is fine, but, especially on a leaner budget, it will make a big impact on your failure plans for retirees in their 40s and even 30s. It’s a piece that, at age 70 will cover a big percentage of most budgets and possibly all of a lean budget. That means your portfolio doesn’t usually have to cover your whole budget for 60 years, and the 30y failure rate of 4% is a fair bit lower than the 50-60y.

      Obviously, people should model things individually. For some people, 2 and 3 are not major factors (they are not stoic masters and are dead set against working for pay in retirement) and they may already be considering 5 in their simulations. Also, the potential for failure needs to be weighed against the time/psych cost of staying in your job longer for a safer retirement. If you mostly like your job ok, that tradeoff is going to look very different than if you’re counting the days/dollars to your retirement like a prisoner marking the days to parole.

    3. First of all, the 3% failure rate is the unconditional probability. Failure rates are much higher when you condition on expensive equity valuations (see SWR Series, Part 3).
      Also, as I showed in Part 23 and here again: that $12,000/year job may sound OK if you do this for only a year. But not 20+ years. I’m not going to work into my 60s!

  16. None of this math takes into account the risk of spending a large chunk of one’s last healthy, active years doing things one does not want to do, only to not have SOR risk materialize after all. There is a point where it’s a rational gamble to say “F*** it, I’m spending the next 3 years of my life on my own terms at the risk of being vulnerable to a 8% chance of having to work again someday.” The downside is living like a normal person.

    There are also people who work longer because of fear, and who are destined to die young because of occupational hazards like being sitting in an office chair 7-8 hours per day.

    How can we model this risk?

    1. There’s rampant ageism in many industries, especially tech. Every year of working in tech is one year closer to being disregarded by many employers. Gaps in one’s resume make this far worse. Going back to work after a significant break from the industry is a legitimate fear.

      The occupational hazards risk is an interesting point, and could be used to model career moves for people not angling for FIRE but just wanting a better work / health balance.

      1. Robert: a mitigating factor here is that if you’re choosing a WR with a 5-15% failure rate like 4%, even in most of the “failure” or close failure scenarios, you don’t actually have to go back to work at anywhere near your prior income to come out ok. I’m looking to harness similar data sets to ERN to model this better, but some rough looks suggest that that even if you assume that you’ll only be able to make 1/2 of your prior salary, it’s nearly unheard of to have to work longer to protect your portfolio than you would have had to stay working (at your initial salary and savings rate) to go from 4% to 3.25%. Basically, 1966 (and potentially 2000) go worse for the 4%er unless they are overly prescient. All the other historical retirement dates, even if you do have to go back to work for lower pay, it’s for about the same or less time than you’d have stayed working to get to 3.25%.

        1. Gnomeozurich, those are great points. I did some similar analysis of the cost/benefit tradeoff of extreme austerity in the first several years of early retirement. Part 14 of Karsten’s SWR series shows that savers and retirees are in a zero sum game. For retirees with extreme budget flexibility or a big enough portfolio, they can avoid sequence of returns risk by being a saver instead of a spender during early years of retirement (e.g. reinvesting a portion of dividends and interest instead of spending income and drawing down capital).

          I wanted to know how long I would have to commit to an austere withdrawal rate (AWR) for my annual take-home income (in real dollars) to grow faster than inflation and surpass (in real dollars) what it would have been with the safe withdrawal rate. Using the SWR Toolbox, I computed my SWR for a 60 year period at 2.68% (for 0% failure with 100% preservation of capital). In the extreme tail case (0th percentile), an AWR that is 0.25% lower than the SWR could take 14 years before the smaller percentage AWR from a growing portfolio surpasses the larger percentage SWR from a smaller portfolio. In the 5th percentile, that duration drops to less than five years. Most retirement cohorts take even less time than that, but given our focus on safety and the currently high CAPE it seems best to focus on the painful tail cases.

          You can see the time ranges in this vertical gantt chart I created from my analysis for various WR reductions below the SWR:!/vizhome/SWRvsAWR_0/YearsuntilAWRpayoff

          To your point that it can take a long time of additional years of working to help reduce the SWR, this shows that it takes a long time of below-SWR austerity to grow one’s portfolio to the same spending power as the SWR. And that doesn’t account for the opportunity cost of years of missed income.

    2. Let me ask you back: Imagine the rental car company tells you it’s OK to return the car empty, say, because they didn’t refill it and gave it to you almost empty.
      How would you return the car? Run out of gas a mile before the airport and push it the rest of the way? I would not stress out it return it 1/8 or even 1/4 full. Sure, I give a free gift to Alamo, but it’s not worth the risk running out of gas. It’s an asymmetric risk!

  17. Congrats on setting your date!

    I think it’s a lot easier to shoot from the hip and say you’ll be flexible when you’re younger and a long way from FIRE, like jp6v said. Life changes so fast when you’re young – new jobs, new cities, new countries, marriage, divorce, kids, buying cars and houses, etc., so it’s not unreasonable to think you can easily change your spending in the future.

    But after reading this post, it’s obvious it’s much easier to work another year or two than try to drastically cut back your spending for years or decades if the worst case scenario hits you.

    It’s debates like these that make the whole FIRE blogging community worthwhile. A number of blogs talked about being flexible, then Karsten ran the numbers and presented his results logically and in an easy to understand way. Emotions aside, it’s clear now how difficult being flexible really is.

    I like to think most FIRE followers are smart enough to realize they read blogs to learn from the masters like Karsten, rather than to re-enforce their own uneducated opinions.

    This reminds me of a mentor I once had on the shop floor who used to say, “Measure twice, cut once.” The same goes for SWR – it’s all in the planning and you only get one chance. This blog is like a mentor for everyone else to learn from.

  18. Ahhh the Tacoma narrows bridge! Monte Carlo analysis was invented in 1946 by Ulam after being briefly considered by Fermi in the 1930’s. Soon enough Monte Carlo was adopted to bridge building to try and simulate out a Tacoma narrows kind of problem. Monte Carlo is a very common technique in communications, physics, physical chemistry, engineering etc. If you ever took chemistry, stochastics is a physical representation of monte carlo and the laws of large numbers namely 6.023 x 10^23

    I doubt anyone can really address the future especially a 50 year future by looking at 1929 or 1966. No one has the first clue “what it’s going to be like”. I look at portfolio protection in a slightly different light. When you buy a portfolio you are basically purchasing a product. The product is a “safe retirement” meant to last over a specified time at a specified withdrawal rate in some specified economic environment, with specified return and specified variance. In other words it’s a statistical prediction. I don’t really care how you arrive at your portfolio whether using FIREcalc or some other backward looking model, or by stress testing your portfolio by using monte carlo (my preferred method), or using some simple minded numeric like 4% x25 (which I find to be ridiculous). If you look at your portfolio like a lifeboat, you can do a few things to toughen up the boat.

    1 understand your risk and portfolio efficiency. I believe in the efficient frontier, some don’t. Clearly the bogelheads don’t and I think their boilerplate portfolios add too much risk. Too much risk over 50 years is pretty much a guarantee of failure, same as too much resonance in the Tacoma narrows was it’s death note (pun intended). So tune the portfolio to best risk. Also tune the portfolio to least tax drag. The government wants your money and they will take it especially using RMD so the old saw “fill up your pretax accounts first” may not be the best advice for a 50 year retirement.

    2. Understand your need. How you withdraw money does matter. When I retired I picked a monthly amount because I wasn’t exactly sure what living on that amount was going to be like. I went to my SS account and found for 50 years of taxforms, I added up my Medicare taxes and came up with a number that represented all the money I ever made, and I subtracted how much I had in the bank. That number was how much money I had spent on my life. I divided the money spent number by 50 years and this was my average spending. It came out to $79K I knew exactly how it was to live on 79K per year since its how I had been living. I’ve been through the 60’s 70’s 80’s 90’s 00’s 10’s and now I’m at 2018 and during that time I funded a portfolio bought a house put 2 kids through college etc etc. I knew exactly what that felt like, I understood my need.

    3 I took what I had in the bank and 3% ruled it, divided by 12 and used that as my monthly retirement starting point.

    4 I retired since I had more money than I spent in 50 years and I wasn’t going to live another 50.

    5. Once retired I decreased my monthly spending to see how low I could go as an experiment. I could go down 30% without batting an eye, and my wife didn’t complain. My subsistence no complaints level was 84K per year basically the same as my average 79K so there is reasonable agreement. so I live on $84K per year and I bank the 30% difference. If I want to do something I save up that 30% If I don’t I save it till I want to do something. So my spending is forward looking and somewhat flexible.

    6 I used some of the excess portfolio to purchase a second retirement (lifeboat to the lifeboat), in case the first fails. $500K grows pretty big in 30 years (or 28). So does $250K.

    7 I stress tested the hell out of my assumptions using monte carlo

    8 I included and un-included SS and tested the best strategy to take SS including my wife in the mix. It makes a difference how you take SS

    9 I tax loss harvested aggressively and built a post tax portfolio in stocks TLH + post tax stock = rich mans Roth.

    10 I am converting my IRA to Roth at the 24% bracket. Third best choice 22% second best choice somewhere between 22% and 24%. I am using some of that post tax +TLH cash to live on tax free while doing the conversion. The tax savings pay off and overcome the SORR in about 19 years. If taxes go up or I die they payoff is sooner because my wife will pay the much higher single rate without conversion. My end of life portfolio is 754K larger with Roth conversion vs no Roth conversion at year 32.
    11 I funded “one offs” like a new car or a new roof during my working years.
    12 I also use the concept of epics. Now I am in the Roth convert epic. When that ceases I’ll be in the residual RMD, SS, since I’m probably going to have some IRA left. A few years later my wife will be added to the SS residual RMD epic and so on. At each epic a cash flow re-evaluation will ensue.

    Have an efficient portfolio, a big pile, a small draw, a back up retirement, forward fund one offs, understand your need, budget, and a nice life.

  19. Thanks for all the contact Karsten… an engineer, I love the analysis. We are within 1-3 years of retiring in our early 50’s. With our longevity, the CAPE-based rules are very logical and easy to implement. In this part of your series, you mentioned the CAPE-based rule with a VPW adjustment. Sorry….but the explanation given for this rule was not clear to me. Could you please show an example of this equation and how the VPW is incorporated into the basic CAPE rule?

    1. The CAPE is done targeting capital preservation. Imagine you get 3.5% from the CAPE.
      Now assume the VPW (targeting capital depletion) recommends 4.2% at your age but 4.0% with a very long (infinite) horizon. So, 0.2% is due to your less than infinite horizon. I’d add the 0.2 percentage points to the 3.5 of the CAPE for a total of 3.7%.
      Hope this helps.

      1. We are targeting capital preservation in our plan in order to do bequests, so it sounds like CAPE (with no VPW adjustment) would be the best path.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.