Life After Full-time Work Blog

Learn about preparing for life after full-time work through posts from Don's upcoming book.

#84 Sometimes Bad Things Happen For Long Periods

Of course we hope for good outcomes when we invest. But we must consider the possibility that outcomes will be bad, perhaps even over long periods. That’s what risk means. Let’s take a look at history again, this time looking at bad news.


Post #83 ( explained sequence-of-returns risk: the chance that, in decumulation, low returns will occur early in the drawdown process and permanently deplete the pension pot to a much greater extent than if average returns had been experienced. I was (pleasantly) surprised by the number of your responses, both on this website and after my identical LinkedIn post, as well as in personal emails I received.

Among the comments were thoughts on how long a number of years a “safety bucket” might provide for; and in turn, of course, that would be influenced by how long a period low returns might persist.

So it’s clear that some historical analysis would be useful (even though, of course, it’s only the future that counts, not history). And in this post I’ll expand on the numbers provided in Post #36 ( that examined aspects of historical returns, this time focusing on the bad times.


In Post #14 ( we saw that, when the odds are in your favor, the longer you play the game, the more likely you are to experience a favorable outcome. But even a good strategy (playing the investment game for a long time) can have a bad outcome. In this post, let’s focus a little more closely on outcomes that are disappointing even over the long term.

When we took a look at longevity, we noted that our own future lifespan is uncertain. If we want to play it safe, we probably ought to plan to make our money last longer than we expect to live – just in case we outlive our life expectancy (see Post #12 For example, suppose people of our age are expected to live, on average, to age 85. (I’m picking a number out of the air here, just as an example.) That means that roughly half of us will live longer than age 85, and half won’t live as long as age 85. Perhaps we might be one of the people living longer than average. If we’re at least in average health, it would be wise to plan to make our money last beyond age 85, just to be on the safe side.

How much beyond age 85? I mentioned two safety margins (though there’s nothing magic about either of them). We can ask: if 85 is the age that half of us will live beyond, what’s the age that only a quarter of us will live beyond? Maybe that’s 89. And what’s the age that only one-tenth of us will live beyond? Maybe that’s 95. (Again, I’m making up the numbers.)

OK, then, we’ll be playing it a bit safe if we plan to make our money last to age 89. Because then, if the future is like the past, there’s only a 25% chance (a quarter) that we’ll outlive our money. And we’ll be playing it even safer if we plan to make the money last to age 95 – because then (if the future is like the past) there’s only a 10% chance (one in ten) that we’ll outlive our money.

Again, there’s nothing magic about “a quarter” and “one in ten” – they’re just different measures of playing it safe.


The same idea can be applied when we project investment returns into the future.

We saw in Post #36 what average returns have been in the past. Those are the equivalent of “average life expectancy” figures. If we project our money into the future using those average numbers, we’ll have a 50/50 chance of doing better than we project (if the future is like the past) and a 50/50 chance of doing worse. So, if we want to play it safe, we ought to project lower returns in the future.

How much lower?

I ran the numbers underlying the tables in Post #36 over 10-year periods. (I chose 10 years as a reasonably long term for future projections – again, no magic to 10 years.) First I calculated the average returns over 10-year periods. (I didn’t show them in those tables in Post #36 – never mind – you’d guess that they would be pretty close to the averages over 1-year periods.) Half of the ten-year periods had returns higher than those averages, and half of the ten-year periods had returns lower than those averages. (That’s what “average” means, essentially.)

So now I asked a different question. What was the return that only a quarter of the ten-year periods fell below? (Those of you who are familiar with statistics will recognize that I sought the “lower quartile break” of the distribution.) How much lower than the average return was this “a quarter” return? If we want to play it a bit safe when we project future returns, we can use the lower number instead of our best estimate of the future.

Here are the answers.

For equities, we should subtract 4.4% from the average annual return, if we want to project with a 75% (rather than a 50%) probability of doing better than our projection (assuming the future distribution of outcomes is the same as in the past). For Treasury bills, we should subtract 1.7% from the average annual return.

And then I also looked at the “one in ten” numbers (the lowest decile break, for you techies).

For equities, we should subtract 8.3% from the average annual return, if we want to project with a 90% (rather than a 50%) probability of doing better than our projection (assuming the future distribution of outcomes is the same as in the past). For Treasury bills, we should subtract 4.2% from the average annual return.


Whoa, that means we will be projecting returns that are much, much worse than the historical averages! Yes indeed, that’s a good interpretation.

In fact, that’s a reasonable way to look at the risk we’re taking. If the future turns out to be worse than the past, we’ll earn lower returns than the historical averages. Will that happen? We simply don’t know. We can’t know, until the future has happened! That uncertainty is one aspect of the risk we take when we invest. If we want to guess how we’ll feel if the risk that we take results, not in the reward we expect for risk-taking, but in a much worse outcome, one way is to project that “much worse” outcome and then try to guess how we’ll react.


And here’s another “in fact.”

In fact, if we don’t do something like that, we’re kidding ourselves. Because then we’ll just believe that investing in equities, with their higher average long-term return, is the answer to all our problems. “Need a higher long-term return? Sure, invest a greater amount in equities. The more you invest in equities, the better the outcomes will look.”

Not so. It’s true that the average projected outcome will look better; but it’s also true that there’s a chance that the actual outcome (which, remember, is necessarily uncertain) will be much worse. And unless we get some kind of estimate about how bad the risk can be, we’re looking only at the good outcomes, not also at the bad outcomes.

If there’s one big lesson to carry into making projections about the future, it’s that we should look at possible bad outcomes as well as possible good outcomes before making decisions. I can’t stress that strongly enough.


Another approach to dealing with investment risk in decumulation is to set up a “safety bucket,” separate from the rest of the pot that seeks long-term growth. The safety bucket is where we’ll draw money from in the short term, if growth is poor. We’re giving ourselves (really, the equity market) time to recover. Of course, the danger is that one day there’s nothing left in our safety bucket and the market still hasn’t recovered. There’s always that risk.

That approach has been referred to, more than once, in readers’ comments on Post #83. I’ll draft a separate blog post to discuss that approach.



To try to guess how we’ll feel if the future is like a bad outcome from the past, rather than just the average outcome, we should also consider much lower investment returns than historical averages.


I have written about retirement planning before and some of that material also relates to topics or issues that are being discussed here. Where relevant I draw on material from three sources: The Retirement Plan Solution (co-authored with Bob Collie and Matt Smith, published by John Wiley & Sons, Inc., 2009), my foreword to Someday Rich (by Timothy Noonan and Matt Smith, also published by Wiley, 2012), and my occasional column The Art of Investment in the FT Money supplement of The Financial Times, published in the UK. I am grateful to the other authors and to The Financial Times for permission to use the material here.

6 Responses to “#84 Sometimes Bad Things Happen For Long Periods”

  1. Ted Harris says:

    Referring to this and the previous blog, I would like to set aside 5 years of income too. Other factors we’ve considered are asset mix, including the characteristics of the components of each asset class – our own glide path, if you will. Additionally, we are fortunate in that we can easily modify our discretionary spending.

  2. mark says:

    Very informative post Don – Thanks
    Best regards

  3. Sorry to be long winded here, but I wish there was a computer simulation to tell us the odds of the expected returns for a successful thirty year retirement using only the historical data from the prior periods in which financial markets corrected after 10 years of prolonged government intervention in the free markets with global negative interest rates and trillions of global QE. If you could run that one for us… but of course no one can, so it’s one of the conundrums the users of asset class “optimizers” gloss over. Predictive reliability of past data assumes everything that happened in the past extrapolates to reliably predicting everything that might happen in the future. If the failure of credit default swaps were in the data in 2008, would fewer seniors who retired in 2007 been hurt by stress-related illness brought on as they watched their risk of running out of money suddenly get real? We equate it to the feeling you’d get from your boat sailing away on auto-pilot without you after you’ve accidentally gone into the water — it’s one thing to imagine when its never happened to you, and quite another to actually experience during.

    • Don Ezra says:

      Absolutely, right, Mark. Every time is different, in some way. The future will also be different, except that we don’t know how — and we can’t know how. So, you’re spot on: history is only for studying, not for predicting. It’s still useful to study it, to do “what if …?” simulations — just not to totally rely on it.

Leave your question or comment