Misunderstanding Sequence of Returns Risk

There is a widespread understanding that “sequence of returns risk” is a big threat to retirees and potential retirees. Here’s a recent blog by Kitces about it and ways to protect against it: https://www.kitces.com/blog/managing-portfolio-size-effect-with-bond-tent-in-retirement-red-zone/

I think most people misunderstand what a bad sequence of returns looks like in practice. I think most people think sequence of returns risk = big market crash right when I retire. But that’s not what happens!

(At least not in US history!)

First let’s take a look at the Maximum Safe Withdrawal Rate (MSWR) for each year of retirement from 1871 to the present.

We can see that the number varies substantially. In some retirements you can only withdraw 3.7% safely; in others you can withdraw 8% or even 9%.

Another way to look at things is the distribution of MSWRs. We can see that over two-thirds of the time we can withdraw at least 5%. The median MSWR is 5.6% — that is, 50% of the time you can withdraw at least 5.6%.

Distribution of MSWRs

But there are still some scenarios where we end up with a very low MSWR. 20% of the time the MSWR will be 4.5%. 10% of the time it will be 4.1%.

What separates the normal case from these tail risk cases? These are all the years when the MSWR was under 4.5%.

We know that the 1966 was a particularly bad year, so let’s look at that more closely first. Were the problems caused by a massive stock market crash? Not really.

If we look at the annual returns, things don’t look so bad. The average is over 10% a year, which seems pretty good. 1973 wasn’t great but its 18% drop isn’t enough to classify as a bear market. There are 4 years when the returns are over 20%. The first three years all feature pretty good returns.

Nominal annual equity returns

So what went wrong? Things look very different when we stop looking at nominal returns and add in inflation. Now the average annual return drops to 2.3%. 1977 goes from a small loss to a large loss; 1978 goes from a respectable gain to a small loss. All of the gains, especially in 1979 and 1980, take haircuts.

So it looks like the nominal stock market crash may not have been the problem; rather it appears that high & persistent inflation was the problem.


Maybe inflation is the real problem behind low MSWRs?

30 year average inflation

Every time we see MSWRs take a nosedive, we also see inflation spike. We can check the correlation (this time using the average inflation of the first 15 years).

The r-squared is 33%, which is okay but exactly great.

But it is a better guidance than “a big nominal drop in the first N years”?

The average equity return in the first 5 years of retirement only has an r-squared of 25%, which is quite a bit less than our 15-year inflation variable.

Geometric mean return in first 5 years of retirement

It looks like inflation should be more of a worry than (nominal) stock market crashes.


In the previous section I looked at average inflation over 15 years and average equity returns over 5 years. Those were chosen pretty much arbitrarily. Would different lengths of time work better or worse?

We can check that pretty easily. First let’s look at the r-squared values we get when looking at different lengths of time for equity returns.

different lengths of equity returns

It looks like it peaks around the 6 or 7 year mark. That is, the equity returns over the first 6 or 7 years are most correlated with your eventual lifetime MSWR.

different lengths of inflation

With inflation the effect is more gradual and seems to peak around 16 or 17 years. That is, the average inflation over the first 16 years of your retirement.

Notice again, the while it is not exactly a massive lead, inflation seems to always be better correlated than equity returns.

So while the link between inflation and MSWR isn’t as strong as I initially suspected, I stand by my initial belief that worrying about a big (nominal) crash in the first few years of retirement is the wrong way to think about sequence of returns risk. Inflation is the bigger risk by a fair amount.

Of course, the best thing is to look at both together: real equity returns.

real returns averaged over various periods of time

We can see that the r-squared for real returns over the first 8 years of retirement is 62%, which is the best we’ve found yet. But, again, being worried about returns in the first year or two of retirement seems to be taking too short of a view. We are again looking a situation where we should be most worried about nearly a decade of (on average) poor returns.

Like what you read? Give EREVN a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.