Safe Withdrawal Rates – ERN #7

Safe Withdrawal Rates – ERN #7

Today’s post is our seventh visit to the mammoth series of posts on Safe Withdrawal Rates at the ERN blog.

Simple maths

In article 22, ERN looks at whether the “simple math behind early retirement” can actually make retirement more difficult.

  • He’s referring to Mr. Money Mustache’s (MMM’s) famous post which claims that the time is takes you to retire is a simple function of your savings rate – the proportion or your income that you are able to save each year.

The simple math assumes that the 4% rule holds, and so you need to build up a pot of 25 times your annual expenditure.

  • The more you save each year, the quicker you get there.

As we’ve seen in the earlier posts in this series, 3.25% is a more realistic Safe Withdrawal Rate (SWR).

  • So we would target a pot of 30 to 31 times annual spend.

When we looked at MMM’s post, we found that – for UK investors in particular – it was an oversimplification.

In response, we came up with a four pot solution that covered the same ground:

  1. ISAs, at FIRE age
  2. private pensions (SIPPs), at age 55
  3. company pensions, at age 60 (usually)
  4. state pension, at age 66 (and moving upwards).
Target pots

ERN’s post is not so much a look at the impact of MMM’s “simple maths” as a look at the impact of having a target pot size that triggers retirement.

  • The problem is that you are most likely to hit your target at the end of a bull market in stocks.
  • Which of course is just before a bear market in stocks.

So many people would retire at the worst possible time from a sequencing risk perspective, and would have lower SWRs than otherwise.

Endogenous retirement dates

Endogenous retirement dates

The chart shows monthly cohorts from 1871 saving 50% of their income into a 100% stock portfolio, towards a retirement target of 25 times annual spend.

  • The endogenous retirement rates are marked in red on the underlying plot of the S&P 500 cumulative real return (in blue).
  • They cluster around bull markets, with big gaps around bear markets.

Neighbouring cohorts

Neighbouring cohorts

This chart compares cohorts from October and November 1992.

  • They start one month apart, but retire more than four years apart.

That’s because the first cohort retired in the 2000 peak, whereas the second fell just short, and was hit by the summer 2000 crash.

  • Whatever multiple of spend you choose, some cohort will always just fall short at a market peak.

Neighbouring cohorts

Neighbouring cohorts

This chart shows the S&P returns (top) against the SWRs to exhaust the portfolio over 60 years (bottom).

  • The red dots in the bottom panel are mostly at the lower SWRs, and never at the peaks.

That’s because the peak SWRs are always in bear markets when nobody’s pot looks large enough to retire.

Of course, many of the endogenous SWRs are still more that adequate.

  • What we really need to know is the impact of endogenous retirement timing on the failure rates of the SWRs that we care about.

Neighbouring cohorts

Neighbouring cohorts

ERN looked at the 4% Rule, and at our preferred SWR of 3.25%.

  • He plotted failure rates over 60 years against equity allocation in retirement.
  • Note that the retirement dates were driven by a 100% equity portfolio during the accumulation phase.

The solid lines show the failure rates using all possible retirement dates (the regular cohort baseline).

  • The dotted lines show the failure rates using endogenous retirement dates.

The good news is that endogenous dates have more of an impact at 4% SWR than at 3.25% SWR.

  • Even better, a 3.25% SWR with 75% equity allocation (our best practice result so far) shows no impact at all.

In the real world, there will likely be less impact than this, since most investors will have a less than 100% equity allocation during accumulation.

  • And couples will be able to stagger their retirement by using two separate pot size targets, which should be hit at different times.

And using (rising) equity glide paths offers further protection against retiring right before the next bear market.

So a target pot size does increase failure rates because you are more likely to retire when stock markets are high.

  • But if you use a cautious SWR (3.25%) and a high equity allocation (75%) you should see no impact.
Side hustles

In Article 23, ERN looks at the options for dealing with market downturns in retirement.

  • We’ve already looked at lots of rules for flexible spending, and ERN has shown that CAPE-based rules work best.

ERN estimates that simple temporary cuts in withdrawals would have to be of the order or 20% to 25% to be effective.

  • That’s quite a jump, and those retirees who haven’t previously thought about flexible spending rules might not be able to implement such a cut.

Instead, ERN looks at what the FIRE community calls “side hustles” – a job (or income-producing hobby) that is worked during retirement.

I’m not keen on this for a variety of reasons:

  1. Tax complications and the impact on pension withdrawals.
  2. The need to potentially work this hustle for several years until the market recovers, or even longer.
    • ERN reckons in some cases you might need more than a decade to catch up.
    • That means that you need to find an income stream that you really enjoy (I haven’t yet).
  3. For some retirement cohorts, the SWR might still be more than 3.25% pa (or even 4% pa) despite the downturn.
    • These people would have gone back to work for nothing.

My mandatory (minimum) withdrawal rate is around 1% pa, so it’s unlikely that I will need a side hustle, even in a downturn.

  • So I won’t cover ERN’s analysis is as much detail as normal, but instead I’ll focus on his conclusions.

The simulation assumes a retiree with $1M who withdraws $40K pa (4%).

  • If his pot falls to 70% of its target value, he “gets a job” (at minimum wage) until the portfolio recovers to 80% of its target value.



In the first example, ERN looks at a start date of Sep 1929 (the market peak before the Depression).

  • Here the SWR (50 years, 100% portfolio depletion) is only 3.26%.
  • So the $1,000 per month from the side hustle is needed (for a 4% SWR).

Side hustle 1929

Side hustle 1929

In the narrow sense – helping the retiree to survive the Depression – the side hustle worked.

  • But the retiree had to work for a total of 22.7 years, within an overall period of 24.7 years.

That’s not retirement.

The second example ERN looked at was 1966.

  • The results aren’t much better.
  • The retiree works for 16.3 years during the first 25.2 years of retirement.

The third example is 1972.

  • Here the retiree goes back to work but it turns out they didn’t need to (remember, we don’t know the SWR at the start of retirement).
  • They work for 11.3 years during the first 13.9 years.
  • But it turns out that their SWR was 4.16%.

ERN also makes the point that you can’t know in advance how much money you need to earn from the side hustle.

  • For the 1929 cohort, $750 per month rather than $1000 would have meant that the retiree still ran out of money.

And even a large side hustle (100% replacement ratio, or $40K pa / $3,333 per month) would still require 9.3 years of work during the first 17 years of retirement for the 1966 cohort.

It’s possible after this analysis that a side hustle still appeals to you as a fallback plan, but I for one have ruled it out.

Flexibility myths

In Article 24, ERN looks at some more flexibility myths.

First he calculates how much more money would be needed by the 1929 and 1966 cohorts for them to survive the whole 50 years on 4% pa withdrawals.

  • Is flexibility cheap?

It turns out that the 1929 cohort needs an extra $266K.

  • The 1966 cohort needs an extra $146K.

The reason that so little extra is needed is that the withdrawal in the second half of the 50-year retirement cost a lot less in terms of initial capital because of the time value of money (discount rate).

These are still not inconsiderable sums, but one way to accumulate them would be to fall victim to  “one more year” syndrome.

  • If you work one or two years beyond the point at which you hit your target pot (25 times expenses, or 31 times), you will build in a significant margin of safety.

ERN believes that flexibility pre-retirement is much easier than flexibility post-retirement.

  • This assumes of course that you can stand to work your final job for an extra year or two.

The second myth is that “flexibility” (in whatever form it takes) only needs to last as long as the downturn.

  • In practice it has to last longer, because the end of the downturn only signifies a recovery in nominal terms.

Retirees need a recovery in real (inflation-adjusted) terms, which takes significantly longer.

  • And of course, the withdrawals made during the downturn increase the amount of ground that needs to be mad up (this is the impact of sequencing risk – selling stocks when they are low).

This is the same logic that makes ERN not too keen on a cash buffer.

  • I agree with his analysis that cash doesn’t increase the SWR, but I think that it has psychological benefits.

In the second half of the article, ERN looks at a lot of flexible withdrawal rates.

  • This feels at first like a rehash of work we covered in earlier posts, when ERN showed that a CAPE-based rule was best.

But in fact, looking at the unluckiest cohorts produces slightly different results.

1929 cohort flexibility

1929 cohort flexibility

All the rules produce low withdrawals for the first 25 years (of a 50-year retirement).

  • The Constant 4%, Bogleheads VPW and Guyton-Klinger rules fall more but then recover better.

On balance, I think I still prefer the better of the CAPE rules.

  • But this involves a VPW increase to allow full capital depletion, which I don’t want.

So the 1929 cohort gives pause for thought, but no perfect solution.

1966 cohort flexibility

1966 cohort flexibility

It’s a similar, though slightly more positive, story with the 1966 cohort.

Cash Bucket

In Article 25, ERN looks at a couple more variations on flexibility that were suggested by his readers.

First up is an initial cash bucket outside the $1M base portfolio.

  • This is used for withdrawals if the main portfolio is underwater by 20% or more, but is never replenished.
  • It’s a simple protection against Sequencing Risk in the first few years of retirement.

It turns out that the cash bucket needs to be around $100K to $116K for the two worst-case scenarios (1929 and 1966).

  • This is smaller than the “extra portfolio” needed to protect these cohorts (see above).

ERN is surprised by how well this works, and interprets the result as the use of a “mini-glidepath” (rising allocation to equities) in the early years.

  • He calculates that you need to save 27.5 times expenses (it’s actually 28 times) to include the cash bucket, rather than 25 times.
  • This translates to an SWR of 3.6% pa, which is pretty good.
Upward mobility

The second suggestion was to start with a failsafe SWR of 3.25% pa, then to “ratchet up” when the bad early returns don’t materialise.

  • ERN calls this Upward Mobility or one-sided Guyton-Klinger.

The rule he used was to increase consumption by 1% of base whenever the effective SWR fell below 99% of the initial 3.25%.

It does help in many cases, but not as often as you might think:

  • 1929 would have never moved up the withdrawals. Makes sense!
  • 1966 would have been stuck with the 3.25% SWR for a long time and only toward the end would have seen an increase by few percents.
  • 1972, which is a true type 2 error: a 4% SWR would have worked but we would have withdrawn a lot less for about 25 years! Then noticing the mistake we’d have almost doubled the withdrawals in the second half of the retirement!
  • The 2000 cohort would still be withdrawing the same initial amount! Even though the market recovered, the portfolio ex-withdrawals did not. Ouch!
  • The 2007 cohort has only recently started to move up their withdrawals. But we’re still below a 4% SWR even after the long rally! A bit of a disappointment!
7% SWR

ERN also looks at Jim Collins’ suggestion (he says it’s in the Trinity Study) that you can start with a7% SWR if you are flexible enough.

  • That’s higher than the long-run return from equities, and will rarely work.
  • It certainly won’t work when the CAPE is high.

A few more flexibility myths discussed at the end of the article, but they are all things that we’ve come across before.


Nothing fundamental has changed today, but we have a couple of tweaks to our strategy.

  • In particular, we can be reassured that if we stick with a 3.25% pa SWR and a 75% equity allocation, we don’t need to worry about the tendency of a pot size target to dump you out at the top of an equity bull market.

I now have more respect for the Constant Percentage, Bogleheads and Guyton-Klinger flexible withdrawal rules.

  • Under certain scenarios (extreme bad sequencing, 50 year retirement with pot depletion) they perform as well – and differently to – our previously preferred CAPE rules.
  • So I think it might be worth calculating a hybrid withdrawal percentage.

And it seems that a cash pot of around 3 years’ spending can boost your effective SWR by 0.35% pa, to 3.6% pa.

  • That’s great news, as I’m already using this approach.

We’ve now completed ERN’s series of 25 articles using seven of our own, and I expected this to be the end (appart from a summary article, of course).

But along the way I’ve discovered two more relevant articles:

  1. Ten lessons from 10 SWR Case Studies that ERN has written up.
  2. ERN’s own Early Retirement Capital Preservation plan.

They must be worth a look, so I’ll be back in a couple of weeks with the final post in this series (apart from the summary).

Until next time.

Share this with , Google+, Pinterest, LinkedIn, Tumblr, Reddit and StumbleUpon.

Mike Rawson

Mike is the owner of 7 Circles, and a private investor living in London. He has been managing his own money for 35 years, with some success.

Article credit to:

Leave a Reply

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