When I wrote Nest Egg Care (NEC), I used data for stock and bond returns starting from 1926. I displayed FIRECalc results for 72 23-year sequences in Graph 2-4. I highlighted the most harmful sequence – a blue line on that graph – as the sequence that depleted a portfolio in the fewest number of years for a given spending rate, investing cost, and mix of stocks and bonds. The 4.40% spending rate that I displayed is the Safe Spending Rate (SSR%) with Zero Chance of depleting in 19 years . The implication is that you have about a 1 in 75 chance of experiencing a return as bad as THE Most Harmful sequence of return. That’s not correct. Your chances are about 1 in 150, and I could argue that it’s 1 in 1,800 chances. The purpose of this post is to explain why the Safe Spending Rates (SSR%s) I provide in NEC are based on a sequence of returns that has half the chances of occurring than you may have thought.
== Shiller data from 1871 ==
When I wrote Nest Egg Care, I did not use the full data set for stock and bond returns that FIRECalc uses. I set FIRECalc to use the annual return data from 1926, since I was trying use similar years of data that Vanguard uses in its Monte Carlo simulations to provide its data on how long a portfolio lasts. As I describe most recently in this post, that is a pointless effort. I conclude the results from a Monte Carlo simulation are not credible: they lead you to far too low of safe spending or withdrawal rate. Count on the results from FIRECalc.
I conclude it is fair to use the same data that FIRECalc uses (from Shiller): the sequence of stock returns (S&P 500 stocks) and bonds returns (10-year US government bonds) to find the most harmful sequence in history. Once FIRECalc updates its data to include 2021 (The most recent update was a year ago.), the data set is 151 annual returns from 1871 through 2021. (I could use or FIRECalc could use the monthly data from Shiller for 1,812 – 151 times 12 – 12-month return periods, but let’s stick with 151 annual returns.)
Let’s assume we want to examine all 20-year sequences of stock and bond returns. FIRECalc will display that as 132 complete sequences of return: 1871 though 1890, 1872 through 1891, …. 2002 through 2021. I first found in this post that the most harmful sequence of returns is the one that started in 1969: that will deplete a portfolio the deepest and fastest for all periods of return greater than ten years. (The nest complete 20-year sequence to be added in FIRECalc’s update – 2002 through 2021 – is not nearly as harmful as the 1969 sequence.)
For your planning, you pick the number of years you want for Zero Chance of depleting a portfolio, and you then derive your Safe Spending Rate (SSR%) by testing that Most Harmful 1969 sequence. I did that work for a wide range of years in Graph 2-7 and Appendix D. Portfolio values from all other sequences of return will not deplete and therefore are less harmful. Most are far, far better. You use your age appropriate SSR% times your Investment Portfolio value to obtain your annual Safe Spending Amount (SSA).
== Count the partial returns ==
FIRECalc will display 132 complete 20-year sequences ending with the start in 2002, but we know that NONE of the 19 partial sequences of return starting in 2003 are candidates for most harmful: none of those partial sequences can replace the 1969 sequence as THE most harmful in history. Those sequences run in length from 19 years (2003 through 2021) to just one year (2021).
Why do we know that? We know that because a candidate for the most harmful sequence 1) must result in a portfolio decline in the first year, and 2) subsequent returns cannot return a portfolio to its initial value. Let’s examine 2021 as the example. Could it have been the start of the most harmful sequence of return in history?
• Example 1: We’ll use the example of a retiree, Tom. Tom started his retirement plan in December 2020 with $1 million. His portfolio mix is 75% stocks and 25% bonds. He withdrew his SSA of $50,000 (5.0%) for his spending in 2021. He started January 1, 2021 with $950,000. If the return in 2021 did not earn back the $50,000 he withdrew, he would have less before his second withdrawal in December 2021 than he did before his first withdrawal. Tom could not calculate to a real increase in his SSA; he sticks with his $50,000 real withdrawal for the next year. Tom could conclude that 2021 was potentially the start of the Most Harmful sequence of returns.
• Example 2: Tom’s actual, real portfolio return for 2021 was 12.7%. He earned back more than the $50,000 that he withdrew: lot’s more – 12.7% times $950,000 equals ~$121,000. He has ~$71,000 more in real spending power than he started with before his first withdrawal. That means 2021 was not the first year of a most harmful return sequence.
Let’s assume Tom sticks with his original withdrawal rate of 5.0%. Tom can increase his withdrawal by 7% in real spending power. He does that and starts anew on January 1. His 5% withdrawal rate assumes 2022 will be the first year of the most harmful sequence of return for 20 years.
(At the end of 2021, Tom could also apply the greater SSR% appropriate for his shorter life expectancy. He’d calculate to an even greater pay increase.)
== The last 19 years ==
Using the example of a 5.0% withdrawal rate, we can see what would happen to Tom’s portfolio if he started his retirement plan in any of those years: 2003 to 2021. If Tom’s real portfolio return was greater than 5.3% in any year ($50,000/$950,000) he would have more portfolio value than he did before his first withdrawal, and that year cannot be the start of a most harmful sequence of return
The annual return rates of the 19 past years show that 13 years (highlighted in green) had real portfolio returns >5.3% and are not possible candidates as the first year of a most harmful sequence.
Six years highlighted in gold with less than 5.3% return are candidates for most harmful sequence, but only one is a serious candidate.
• 2011, 2015, and 2018 are not candidates, since the returns the next year or two are obviously great enough to earn back more than the prior withdrawals. Tom would have more than $1 million real portfolio value before a withdrawal and has to calculate to a greater Safe Spending Amount.
• Of the three years 2005, 2007, and 2008, only 2005 is a potential candidate for most harmful sequence. We have to calculate to find that. The 1.1% return in 2005 meant a portfolio declined; the withdrawal lowered it before the start of 2006; the 9.5% return in 2006 just fell a bit short of earning back to more than it was before the initial withdrawal in late 2004. That means 2005 is a worse start than all subsequent years.
After 16 years, a portfolio exceeds its initial value. Tom would have to calculate to a greater SSA. This sequence is not a candidate for Most Harmful. (Tom’s age appropriate SSR% would be much greater than the initial 5.0%; he’d calculate to a real increase in his SSA well before 16 years.)
==2005 is far better than the 1969 sequence ==
Is 2005 a competitor to 1969 for THE Most Harmful Sequence? Nope. Not close. It’s a below average sequence. It declines to its low point – by 30% in real spending power at the end of 2008, but that low is not close to a tipping point – roughly 50% decline in value. We see that the 1969 sequence had depleted by more than 50% in six years and spirals toward depletion.
If Tom rode the 2005 sequence, he would have more than he started with at the end of the 16th year and more again at the end of the 17th. For comparison, after 17 years of the 1969 sequence, portfolio value was just 4% of the its initial value.
Conclusion: One can find the most harmful sequence of return in history and use that sequence to find the Safe Spending Rate – the spending rate you know results in zero chance for depleting a portfolio. Some argue that using sequences of return in the historical order than they occurred in history doesn’t give you enough sequences to test. This post concludes we have 151 actual and potential sequences of return since 1871 to test and just one, the sequence of return that started in 1969, is truly the most harmful sequence of return. You have 1 in 151 chances of experiencing a return like that if you use the Safe Spending Rates (SSR%s) that I provide in Nest Egg Care.