The myopia of failure rates

Whenever you read about safe withdrawal rates, they usually lie to you about how often they fail. What if I told you that 4% safe withdrawals don’t fail 5% of the time…they fail 53% of the time?

Safe withdrawal rate research achieves this sleight-of-hand by only looking at the end state…and ignoring all of the intermediate states along the way. If a portfolio lasted 30 years without dropping to $0 then success is declared.

But that’s not what real humans would consider success. Consider someone who retired in 1912 and used 4% withdrawals.

Image for post
Image for post

In this scenario 4% withdrawals “succeeded”. You actually last the whole 30 years. But notice how your portfolio has dropped to under $400,000 (despite starting at $1,000,000) within just a decade of retiring. And notice that at no point does it ever get back to $600,000 (60%) of where it started.

Does that seem like a worry-free retirement to you? Or do you spend 15–20 years stressing about your “safe” withdrawal rate?

Another way to look at the same retirement is to look at the actual, current withdrawal rate over time. Sure, it started out at 4%. But each year the withdrawal amount goes up (by inflation) and the portfolio…does whatever the market does.

Image for post
Image for post

Do you feel confident in your “safe” withdrawal rate when you are pulling 6%, 8%, or even 10% from the portfolio with decades to go?

Here’s the “current withdrawal rate” each year, along with how many years of retirement are left.

Image for post
Image for post

We know that 4% is “safe” for a 30-year retirement. But is 4.3% safe for a 29-year retirement? Is 4.6% safe for a 28-year retirement?

If I started out with something that was “safe” but end up withdrawing something that isn’t safe…was my initial rate really safe after all?

The best case scenario is you’re going to feel super-stressed for a few years and then the market has a strong recovery and your retirement squeaks across the finish line. But does that feel like “success”? Does it feel like something a real human being would be able to put up with?

Calculating a 53% failure rate

The traditional “safe withdrawal rate” simulation just runs a calculation a single time at the beginning of retirement. But real retirement is a process. Every single year we are going to look at our portfolio, look at our withdrawal rate, and decide whether we are on the path to success or failure.

What if we take the same “continuous evaluation” approach to calculating failure rates?

Here’s what we’ll do:

  • First we calculate the “safe withdrawal rate” (using historical data) for every length of retirement. We’ll know the SWR for a 30-year retirement and a 29-year retirement and a 28-year retirement and so on.
  • Each year we look at our current withdrawal rate and compare it to known SWR for the amount of time we have left.

Going back to our 1912 retiree, the process will look like this. (Skip down to the bottom of the chart to follow along.)

Image for post
Image for post
  • In year 6 we are withdrawing 6.9%. There are 24 years left. The SWR for a 24-year retirement is 4.1%. Failure.
  • In year 7 we are withdrawing 7.9%. There are 23 years left. The SWR for a 23-year retirement is 4.2%. Failure.
  • In year 8 we are withdrawing 9.0%. There are 22 years left. The SWR for a 22-year retirement is 4.3%. Failure.

You get the picture. 4% is not safe for a 1912 retiree.

We are essentially performing an inductive safe withdrawal rate check. If at any point our strategy left us in an unsafe position — a withdrawal rate that we know (based on the historical data) was unsafe — then our original rate was not safe.

If we apply this metric to all retirement periods we find that 53% of the time the “safe withdrawal rate” causes our retirement to fail by leaving us in an unsafe scenario for at least one year.

In fact, across all retirements 25% all years will be in an unsafe withdrawal rate. We started out “safe” became “unsafe” in 1 out of 4 years. That means on average a single 30-year retirement will have nearly 8 unsafe years.

53% of the time the “safe withdrawal rate” causes our retirement to fail by leaving us in an unsafe scenario.

If you’re going to accept 8 unsafe years…why the fixation on making sure that the first year (or the first handful of years) are safe?

Safe now, or safe later: intertemporal consistency?

Why not pick 4.5% as our starting withdrawal rate if we don’t mind having some years above what is considered “safe”? After all, even with a 4.5% starting rate, most of the time (61% of the time) our portfolio will grow faster than inflation and our later withdrawal rates will be safe.

Said another way, why is this sequence of withdrawals — starting at 4% but quickly going up — considered safe

Image for post
Image for post
The first 5 years for a 1912 retiree

But this one — starting at 4.5% but quickly going down — considered the height of recklessness

Image for post
Image for post
The first 5 years for a 1952 retiree

EarlyRetirementNow’s “safe” withdrawal rate

So far we’ve talking about standard 30-year retirements with the standard 4% safe withdrawal rate. But what if we take ERN’s 3.25% safe withdrawal rate and look at longer retirement periods? How unsafe is 3.25%?

  • For a 40 year retirement, 3.25% fails 58% of the time, with 12.7% of all years unsafe.
  • For a 50 year retirement, 3.25% fails 65% of the time, with 13.2% of all years unsafe.

It should be clear that even 3.25% isn’t especially safe when we pay attention to the intermediate states instead of fixating solely on the final state.

Safety is an illusion

One thing I’m trying to get across here is that absolute safety is an illusion. Even if you pick a very conservative, safe withdrawal rate you are going to find yourself with many years of unsafe withdrawals. If you think otherwise, then you are likely going to be unpleasantly surprised.

Our first thought might be “maybe we just need to pick an even lower safe withdrawal rate?”

If 3.25% resulted in 58% of 40-year retirements failing and 12.7% of all years being unsafe…what happens if we use a lower number?

First, what’s our goal? Let’s say our goal is to find a number such that no more than 5% of all years are unsafe. (Why 5%? I dunno, everyone always picks 5% for these kind of things.)

We need to go all the way down to 2.71% withdrawals to hit that mark.

But that still means that 33% of all retirements have at least one unsafe year! Holy smokes. One out of three seems really high, especially since our goal was safety. Lest you think “oh, I can handle a single bad year, I would definitely just tough it out because I’m cool as ice”. It usually isn’t “just one unsafe year”. Using 2.71% withdrawals:

  • 33% of retirements have at least 1 unsafe year
  • 26% of retirements have at least 3 unsafe years
  • 21% of retirements have at least 5 unsafe years
  • 6% of retirements have at least 10 unsafe years (and you thought 2.71% was safe….)

It should be clear we’re not talking about just one stressful year of unsafe withdrawals. We are talking a good chance (21%) of multiple years (at least 5).

If our goal is that under 5% of all retirements have no unsafe years, then we need to go all the way down to 2.01% withdrawals.

Again: if we’re willing to tolerate an unsafe withdrawal rate in year 16, why aren’t we willing to tolerate it in year 1 (and start out with a number higher than 4%)? How do we find any kind of logical consistency?

If our goal is that under 5% of all retirements have no unsafe years, then we need to go all the way down to 2.01% withdrawals.

Retirement is largely about understanding risks and making tradeoffs. But it should be clear that it is virtually impossible, especially for an early retiree, to ever find “safety”.

Written by

Learn how to enjoy early retirement in Vietnam. With charts and graphs.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store