The Ultimate Guide to Safe Withdrawal Rates – Part 23: Flexibility and its Limitations

Talk to anyone in the FIRE community and ask how folks will deal with market volatility (especially downside volatility) during the withdrawal phase and everyone will mention “flexibility.” Of course, we’re all going to be flexible. Nobody will see their million dollar portfolio drop to $700k, $600k, $500k, $400k and so on and then keep withdrawing $40k every year no matter what. Rational and reasonable retirees would adjust their behavior along the way and nobody will really run out of money in retirement in the real world, as I noted in my ChooseFI podcast appearance. In other words, we’ll all be flexible. But is flexibility some magic wand we can swing to make all the worries about running out of money go away? Or is it BS? It’s a bit of both, of course. For example, I would put the following into the BS category:

  • I’ll do “something” with my asset allocation and recover the losses.   Good luck with that!
  • I will skip the Starbucks Lattes for two months until the market recovers! Ohhhh-Kaaayyy….?!
  • I will sit out one or two years of inflation adjustments.    Qualitatively, a good idea, but it won’t work quantitatively.
  • I will rely on Social Security.    That may work for middle-aged early retirees but not for 30-year-old early retirees!

But flexibility will work through significantly reducing spending. And again, let’s be realistic, foregoing a 2% inflation adjustment for a year is not enough. Flexibility would involve being prepared to cut spending by probably around 20-25%, maybe more. A different route and maybe a better solution might be the side hustle. Specifically, one reader, Jacob, emailed me with this proposal:

Your series is quickly covering a lot of financial acrobatics to discover and maximize safe withdrawal rates while working to reduce the risk of running out of money. However, so far the most tried-and-true solution to the “not enough money” problem has not been considered: Get-A-Job. I acknowledge that for most job-hating FIRE-aspiring people this is the nuclear option, but it’s still an option.

Great idea! Get a side hustle and solve the safe withdrawal rate worries and (hopefully) salvage the 4% Rule! But there are two very important limitations:

  1. The side hustle might last for longer than a few months or years. Withdrawals plus the market drop equals Sequence of Return Risk and might imply that the side hustle will last much longer than the S&P 500 equity index drawdown. How long? Try a decade or two, so if you want to go that route better make sure you pick a side hustle that’s fun!
  2. For some historical cohorts where the 4% Rule would have worked even without a side hustle, flexibility would have backfired; you would have gone back to work for years, maybe even a whole decade and afterward it turned out it wasn’t even necessary!

But enough talking, let’s do some simulations!

Being flexible… but in a quantifiable way

Jacob, the reader who contacted me via email even had a proposal for how to run this in practice:

Let’s take a “typical” early retiree (if there is such a thing) with [$1 million] in assets supporting a $40,000 per year spend at [a 4%] safe withdrawal rate. I would propose some sort of “Get A Job Guardrail plan.” If his portfolio hits 70% of its starting value, he commits to getting the equivalent of a full-time federal minimum wage job until his portfolio recovers to 80% of it’s starting value.

Wow! Amazing! That’s pretty much exactly how I would have set this up! I built a little Google Sheet to do exactly that, see link below:

Link to Side Hustle Google Sheet

(As always, please save your own copy first. You cannot make changes to my clean copy, for obvious reasons!)


  • Enter parameters in the orange-shaded fields. All else is computed by the sheet!
  • Stock/Bond/Cash portfolios. Stocks=S&P500, Bonds=10Y benchmark US Treasury, Cash=3M Tbills. I assume fixed portfolio weights.
  • We can also specify the horizon (in months) and the final portfolio value target (in % of initial value). Right now I use 600 months and a final value target of 0 (portfolio depletion).
  • As always, we can specify extrapolated return expectations for 2018 and onward, to accommodate more historical simulation windows. For example, we’d be able to calculate a historical simulation for a 1970 start date even though we don’t have the last couple of years of returns. The numerical results all hinge 99.99% on the first 48 years of returns and very little on the last two (think Sequence Risk!).
  • The main parameter tab looks very much the one used in the SWR Toolkit (see SWR Series Part 7):
Main parameters (similar to those in the Google Sheet from the SWR Series Part 7).
  • In the second tab, we specify more parameters and display the main results, see screenshot below.
  • In this first example, I use the peak before the Great Depression (Sep 1929), an initial portfolio value of $1,000,000, 4% withdrawal rate, guardrails of 70%/80% and $1,000/month of side hustle income.
  • Notice that in this case study the side hustle income was truly necessary because the actual SWR without a side hustle was only 3.2626%. In other words, without the extra income from the side hustle ($1,000/month), we would have depleted the portfolio for sure!
More parameter for this case study
  • The “target portfolio” used to derive the guardrails is the portfolio that we would have observed over time using a fixed (real) withdrawal and fixed (real) asset return to exactly match the final portfolio value, see chart below. By the way, if you think this looks like a mortgage amortization schedule, you’re right! That’s exactly what a risk-free withdrawal plan with a specific final value target would amount to!
  • Whenever the actual portfolio value drops below the lower guardrail, we go back to work in the side hustle, earning a real, inflation-adjusted wage.
  • Whenever the portfolio breaks through the upper guardrail we quit the side hustle. A quick sidenote: Why isn’t the upper guardrail always set to 100%? We could do that, too, but while playing around with the simulations I realized that with a 100% upper guardrail we’d work way too long into the stock market recovery and then subsequently over-accumulate. 80% seemed like a sweet spot for this parameter. Not sure if Jacob had done his own simulations already, but his initial intuition and guesstimate were spot on!!!
Target portfolio to exactly reach the final portfolio value target. Lower and upper guardrail below.

How do the results look for this case study? Here’s the time series of the portfolio values relative to the guardrails for the 1929 cohort:

Time series of portfolio values. The side hustle saves the retirement for the 1929 cohort!
  • With the side hustle and just $1,000 of monthly income, or 30% of the monthly budget we pretty handily navigated the Great Depression.
  • But the big drawback: We would have to work in this side hustle for an extended period! It looks like after 296 months (almost 25 years) we’re finally done hustling! For me personally, as a 44-year old early retiree the prospect of working until I’m 69 sound very unappealing.
Main results for the 1929 retirement cohort: Be prepared to find a side hustle for 2+ decades!

Also notice that even with the side hustle, retirement wasn’t a total cake walk. If we look at the bottom part of the table, the funding status of the actual portfolio as a percentage of the smooth target drawdown portfolio value (blue line in the chart), even with the side hustle we’d spend decades with a seriously underfunded retirement stash. Still better than running out of money in retirement but it seems a bit scary!

Another case study: 1966

A second disaster scenario I always like to study would be the cohorts around 1965 and 1966. They would have faced some extended poor portfolio returns due to the 1970s and early 1980s (recessions = bad for stocks) and rising bond yields, which was bad for bonds. So let’s see how the January 1966 cohort would have fared. We keep the parameters the same, just change the dates, see tables and chart below:

  • Very similar results: The side hustle only finishes 25+ years into the early retirement. You would have worked a total of more than 16 years.
  • The side hustle is still better than running out of money, but just like before you would have spent a significant portion of your early retirement with a seriously underfunded portfolio. Even though everything worked out with the side hustle, I don’t think this would have been a very comfortable and relaxed retirement: 70 months with a <50% funded portfolio and 10 years with a 50-75% funded portfolio!
The January 1966 cohort. Work for over 16 years until 25 years after starting retirement!
Time series of the 1966 retirement cohort portfolio values. The side hustle would have prevented running out of money. But at the cost of having to work a substantial portion of the early retirement portion.

Flexifailure: Going back to work when it wasn’t even necessary (a.k.a. Type 1 Error, a.k.a. False Alarm)

One of the drawbacks of this side hustle method: Because we don’t know the SWR of our own retirement cohort in advance there is the possibility that folks go to work but then it turns out, after the dust settles, it wasn’t even necessary. Here’s one example: the 1972 retirement cohort. (actually, there were many different cohorts both before and after 1965 and 1966 where the 4% Rule ended up working but this simple side hustle rule would have sent you back to work for multiple years, even decades!)

The 1972 cohort was hit hard by the 1973-75 recession and then again by the 1980 and 1982 recessions. But it turns out, despite all those adverse events, the portfolio recovered so rapidly post-1982 that your side hustle pre-1982 was utterly unnecessary. But since nobody knew about the roaring 1980s you would have wasted 11 out of your first 14 years in retirement pursuing a side hustle that wasn’t even necessary! See table/chart below!

The 1972 cohort. The actual SWR (without the side hustle) was >4%, so was no need to go back to work. But you would have wasted 11+ years in the side hustle!
Time series of portfolio values for the 1972 cohort. The side hustle was unnecessary. But you didn’t know that in real-time, so you would have spent the 11 out of the first 14 years back in the workforce!

So, trying to salvage the 4% Rule through flexibility and a side hustle is a bit like squeezing a balloon: You reduce the absolute disaster scenario (Type 2 Error = running out of money) but now blow up a failure of a different kind: You go back to work for years, even decades when that wasn’t even necessary. But who knew that in the 1970s? Reading about the “Death of Equities” didn’t really inspire a lot of confidence, right?!

The Death of Equities Magazine Cover

Other results not displayed today

OK, this is still work in progress. And the post is already getting too long! I can definitely see a part 2 about this topic in the future. The 70%/80% guardrails with the $1,000/month income (=30% replacement ratio) seemed to work quite nicely in the two prominent deep recessions. But I haven’t done any comprehensive simulations over all the possible retirement cohorts and different parameter values. While playing around with the Toolbox, though, I came up with some other interesting findings:

  • If the side hustle is too small (say $750/month or less) no guardrail would have saved your retirement in 1929. The 1929 cohort would have worked in that side hustle for the entire 50 years in retirement and still run out of money! Pretty scary! I’m picturing myself 94 years old and I have to look for a full-time job. Maybe I’ll become a personal trainer?!
  • Larger side hustle incomes will shorten the amount of time one has to work. But not by much. $2,000/month (or a 60% replacement ratio) would have still required a 13-year work history (1929 cohort). If the side hustle pays for all expenses ($3,333/month=100% replacement ratio) that would still require a 6.5-year work history within the first 14 years of retirement for the 1929 cohort and 9.25-year work history (!!!) during the first 17 years of early retirement for the 1966 cohort! not very appealing!


Flexibility through going back to work and pursuing a side hustle will certainly solve some of the problems of the 4% rule. But the side hustle flexibility is no panacea. A side hustle could potentially last so long, we might as well consider it a multi-decade-long extension of our corporate career.

In addition, the flexibility of working a side hustle raises the issue of what we call a “Type 1 Error” in statistics. We now create failures – of sorts – that would have been considered a success under the inflexible 4% Rule. Specifically, some of the historical cohorts would have gone back to work because they didn’t know in real-time that a strong equity market rally was around the corner (e.g., the 1970s!). So, I have a bit of a dim view on this whole “flexibility” and “side hustle” mantra in early retirement. I’m glad I worked a few extra years beyond 25x spending and accumulated more assets to lower my effective withdrawal rate for my 2018 retirement! That last year before retirement in a nice cushy corporate career with a window office on the 39th floor: it lowered the chance of working as a greeter at Walmart at age 69. Cheers to that!

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: Pixabay


102 thoughts on “The Ultimate Guide to Safe Withdrawal Rates – Part 23: Flexibility and its Limitations

  1. Is the only difference between P-Base and P-SideHustle in the “Portfolio Values vs. and Guardrails” the added sidehustle income? The reason I ask is because the comparative slopes between the two are often much different than I would intuitively expect. This very well could be just me, but I wanted to ask.

    For example, in the first graph you get a spike on the green line between approximately month 530-540 when the purple line is losing even more money (on an absolute basis).

    We also see this in the second graph between [approx] months 495-510. Also in the second graph, growth really picks up on the purple line around month 360 where the slope of the green line: 1. stays relatively constant; 2. is negative instead of positive (both of which perplex me a bit).

    The relative slope differences of green and purple lines in the third graph seem to be more consistent and [intuitively] reasonable to me.

    1. The only difference is that in P-Base you stubbornly withdraw $40k/year and in P-SideHustle you potentially and temporarily adjust the withdrawals (due to the side income and/or lower consumption).

      That said, because the two portfolio values potentially diverge so much over time, their slope will be different because a $1,000,000 portfolio will generate more portfolio return than a $500,000 portfolio.

      In any case, if any of the numbers don’t make sense, check the underlying calculations, they are all in the Google Sheet, all completely transparent. All simple arithmetic for people to convince themselves that everything is A-OK.

Leave a Reply

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