"Is sequence risk really a big deal for retirees?"

Vincenzo Corleone

Full time employment: Posting here.
Joined
Jul 20, 2005
Messages
617
I came across this on MarketWatch, "Is sequence risk really a big deal for retirees?"

One thing that I did find interesting:

"According to Estrada, a 4% withdrawal strategy actually failed just 4.4% of the time since 1900. And when analyzing the particular sequence of bad annual returns that led to those failures, he found that a different sequence of those same returns would have prevented those failures more than 90% of the time. So the odds of sequence risk sabotaging your retirement are quite low."

I was hoping to get others' reaction to this article.
 
4.4% failure rate is in line with Bengen/Trinity studies of a 5% failure rate.
As to using a different sequence to prevent 90% of those failures, isn't one then just changing the historical actual sequencing of returns on which the 4% guidance is based on and substituting it with a theoretical sequencing of returns?

The concept of Monte Carlo simulations is effectively discussed in the article as to using the worst years ever in consecutive years at the start of retirement, which as acknowledged in the article is not realistic, although theoretically possible.
 
The conclusion is intuitively attractive, but I think the methodology is highly questionable. Normally a study like this would use rolling 30year sequences of actual data. The 30 years starting in 1900, the 30 years starting in 1901, etc. Instead, the author did this:
"... imagine that the stock market’s return in each year of your retirement will be equal to what it actually was in any calendar year picked at random from all years since 1900."
Why do it this way? This approach eliminates the momentum that market results have. Real-world results in a given year are not totally independent of the results in a prior year.

My suspicion is that the author did do his study the conventional way, with rolling 30 year sequences, but that he did not like the results. IOW he did not get the answer he wanted -- something he could publish. So he went into the mode of "torturing the data until it confesses." The rule on this is that if you torture it long enough, you can get it to confess to anything.

Short of actually doing the work using 30 year rolling samples, there's no way to know. But IMO the way the author did it will for sure produce many crazy and impossible 30-year sequences.
 
"Is sequence risk really a big deal for retirees?"

I am concerned appropriately about SORR, as I am lots of risks, but I find all these 4% SWR over 30 years from an inflation-adjusted portfolio with a 50/50 domestic static allocation to be very, very particular to those inputs. In the real world versus the academic one, there is Social Security, Medicare, home equity and unlimited spending adjustments that a retiree will make well before someone wakes up one day after 30 years and says, “Oh, CRAP! I’m penniless!”
 
Last edited:
As Dtail implicitly notes, the fact that, when they changed the sequence of returns in the cases that failed, the change saved them from failing, well, that proves that SORR is real!

The larger point the article seemed to be making was that the SORR is smaller than we fear. Who knows?

The other point was to assert that the SORR could be blunted by relatively modest withdrawal changes, i.e., a dynamic withdrawal strategy. However, in some of studies I have read, changes in your withdrawal (according to whatever the dynamic rules are) come too slowly to save you from SORR.

Finally, I found this quote curious:
It’s also worth emphasizing that a flexible withdrawal strategy can also work to your benefit. It allows you to take out more than originally planned following years in which the market has done particularly well, for example. You could always put that extra amount in a rainy-day fund to support your retirement following years of poor market returns. (Emphasis added.)

Umm, you mean, if you DON'T increase your withdrawal when you can, you lower your risks? Who woulda thunk it?

(I think it is silly in this context to make a distinction between "spending" and "withdrawal." When, in a SWR study, one speaks of a withdrawal from the portfolio, it is assumed that that withdrawal is NOT available for spending in the future.)
 
Last edited:
.... In the real world versus the academic one, there is Social Security, Medicare, home equity and unlimited spending adjustments that a retiree will make well before someone wakes up one day after 30 years and says, “Oh, CRAP! I’m penniless!”

While I totally agree there are many ways to adjust to SORR, including spending adjustments, they are not unlimited, and even more limited the older one gets.

There are lots of stories of people choosing between prescriptions and paying the rent or food.

Unless of course one considers not eating or taking medicine as a spending adjustment :(
 
My issue with SORR is that it doesn't seem to factor in early retirement. For a 30 year plan of a 65 yo retiree, it might make sense, but not so much for 40 or 50 years of an early retiree.

Take two people, both 45. Person A retires at 45, and sets their AA conservatively due to SORR risk, with a glide path of 10 years. After 10 years of low to moderate gains, they feel they have avoided SORR and go with their planned 50/50 or 60/40 AA.

Person B is now 55 as well, and just starting ER. They have the same amount of investments as Person A, and the same expenses. Since they are just starting, they reduce their equity investment due to SORR.

So both A and B have the same future in front of them, but A thinks they are clear of SORR, while B thinks they are exposed. How does that make sense?

Personally, I don't think it matters to me anymore. I retired nearly 10 years ago, and this has been a great run to give me a lot of buffer. Even though I'm 58 with hopefully plenty of years to go, I'm not worried about SORR. I've got enough non-equities to get me through a few years of a down market, and my VPW plan will throttle me back through the down years. I don't recall SORR being aware of SORR in 2011. I can't say for sure what I would've done.
 
SORR is my biggest fear when RE at this stage of a market cycle. Especially because our spend is front end loaded. We’ve built in a buffer, but I won’t be completely comfortable until our assets have grown significantly while drawing down.
 
The conclusion is intuitively attractive, but I think the methodology is highly questionable. Normally a study like this would use rolling 30year sequences of actual data. The 30 years starting in 1900, the 30 years starting in 1901, etc. Instead, the author did this:
"... imagine that the stock market’s return in each year of your retirement will be equal to what it actually was in any calendar year picked at random from all years since 1900."
Why do it this way? This approach eliminates the momentum that market results have. Real-world results in a given year are not totally independent of the results in a prior year.

My suspicion is that the author did do his study the conventional way, with rolling 30 year sequences, but that he did not like the results. IOW he did not get the answer he wanted -- something he could publish. So he went into the mode of "torturing the data until it confesses." The rule on this is that if you torture it long enough, you can get it to confess to anything.

Short of actually doing the work using 30 year rolling samples, there's no way to know. But IMO the way the author did it will for sure produce many crazy and impossible 30-year sequences.

Yes.

Annual returns are not independently random from year to year. There's still the thing called business cycle. Timing it is of course not at all easy, but let's not get into that here. :)

Hence, I agree that the better way of looking at past data is to use rolling 30-year periods, and that's what FIRECalc does. And here on this forum, we have forever discussed the two bad periods in history, namely the Great Depression and the high-inflation period from about 1960 to 1980.
 
While I totally agree there are many ways to adjust to SORR, including spending adjustments, they are not unlimited, and even more limited the older one gets.

There are lots of stories of people choosing between prescriptions and paying the rent or food.

Unless of course one considers not eating or taking medicine as a spending adjustment :(

If my plan was so delicate that a few historically bad years would have me choosing between prescription drugs vs rent/food, #1 DW wouldn't let us RE and #2, I would plan harder-ie.. work longer. We are blessed with gov pensions and healthcare (tricare) so maybe that is why I don't seem to worry as much.

My analogy with these tight plans is a 100 mile trip with a car that gets 20 MPG. Do you put exactly 5 gallons of gas in the tank because historically you make that trip 99.5% of the time? What if there is an accident at the 50 mile mark which causes traffic to stop and idle? Then you find your self out of gas at the 97 mile mark. Bottom line for me is to build into all re plans lots of cushion. Plan A, B, C, D, etc... Make course corrections early and often if anything starts looking gloomy.
 
If my plan was so delicate that a few historically bad years would have me choosing between prescription drugs vs rent/food, #1 DW wouldn't let us RE and #2, I would plan harder-ie.. work longer. We are blessed with gov pensions and healthcare (tricare) so maybe that is why I don't seem to worry as much.

My analogy with these tight plans is a 100 mile trip with a car that gets 20 MPG. Do you put exactly 5 gallons of gas in the tank because historically you make that trip 99.5% of the time? What if there is an accident at the 50 mile mark which causes traffic to stop and idle? Then you find your self out of gas at the 97 mile mark. Bottom line for me is to build into all re plans lots of cushion. Plan A, B, C, D, etc... Make course corrections early and often if anything starts looking gloomy.


Yes.

You want more than 5 gallons of gas. You may want to turn off the engine and suffer the heat with no AC if you are caught in an accident traffic jam. You may want to slow down to improve the MPG if your tank runs low.

Still, something unexpected may just happen. This analogy reminds me of a trip I made from Phoenix to the LA area many years ago. My car usually could get me to Long Beach, where I would fill up. On this particular trip, I was running low early and had to stop to refill in Indio. I wondered if I had some unknown engine problems that caused the suddenly poor gas mileage.

I found the answer after I stopped at the pump and tried to exit the car. My car was still pointing west, and I had problems opening the door against the headwind. Holy cow! The wind was extremely strong. Of course I put the transmission in Park, but my car rocked backwards when I let my foot off the brake pedal. And I had to lean 30 degrees into the wind while pumping gas.

If we ran across some severe unexpected problems like that with our finance during retirement, it would be OK to suffer along with millions of other people.

Misery loves company. I would not be alone, and in fact would still do better than most of the people who are not as well prepared.
 
Last edited:
If we ran across some severe unexpected problems like that with our finance during retirement, it would be OK to suffer along with millions of other people.

Misery loves company. I would not be alone, and in fact would still do better than most of the people who are not as well prepared.

Yes. Our sequence had the finite non COLA pensions first and SS last. Seemed like a no-brainer at the time for longevity insurance. But now there are tangible threats to SS. But it's too late to redo the sequence now. So we will suffer with all the other SS recipients.
 
"Is sequence risk really a big deal for retirees?"

If my plan was so delicate that a few historically bad years would have me choosing between prescription drugs vs rent/food, #1 DW wouldn't let us RE and #2, I would plan harder-ie.. work longer. We are blessed with gov pensions and healthcare (tricare) so maybe that is why I don't seem to worry as much.



My analogy with these tight plans is a 100 mile trip with a car that gets 20 MPG. Do you put exactly 5 gallons of gas in the tank because historically you make that trip 99.5% of the time? What if there is an accident at the 50 mile mark which causes traffic to stop and idle? Then you find your self out of gas at the 97 mile mark. Bottom line for me is to build into all re plans lots of cushion. Plan A, B, C, D, etc... Make course corrections early and often if anything starts looking gloomy.



Yes, as the authors of “Your Money or Your Life” said, to paraphrase from my memory, “Have enough, plus a little more, before you retire.”

Having a margin of extra gas in the tank is what I was getting at above, too, which is that many of us aiming to voluntarily RE should have a lot more levers to pull to adjust course early if we find that we’re on one of the rare, poorly performing Monte Carlo lines. I’m not talking about the majority of Americans who don’t save much and take SS at 62, or other unfortunate scenarios that can land them in a pickle later. No one is immune from terrible SORR or other calamities but having a diversified plan with contingencies helps a lot.
 
Last edited:
Speaking of the analogy of running out of gas, please indulge me to tell another story.

This true story was told by my coworker, and it happened in the late 70s. He was driving an old car sold to him by his sister, and going from LA to Phoenix. His sister repeatedly assured him that the car would make it to Blythe before he needed to refuel again. Said she made several trips herself, and that was her experience. Blythe is on the border of California and Arizona.

Well, he ran out of gas something like 30 miles still west of Blythe. There was no gas station perhaps 50 or more miles east or west of Blythe. So, he had to hitchhike into Blythe, bought several 1-gal gas cans because that was all they had at the convenience store, then hitched a ride back to his car. Why several cans? Because back then, American cars with a V8 engine might get only 12 mpg, when everything worked well. As he did not know what the car gas mileage really was, he had to be sure this time, and wanted to bring back as many gallons of gas as he could.

He said he felt like such a fool, standing by the side of the freeway, holding a tray that he found in order to carry these gas cans. Said he did not speak to his sister for months afterwards. :LOL:


PS. He was driving the same stretch as I did in the story I told earlier, except that he was traveling eastward while I went westward. The prevailing wind has always been from west to east, hence the wind would have been a tail wind and helpful to him. Now, perhaps his sister was always helped with that tailwind, but the wind changed that time for his unlucky trip. I recall that he did not say that he found anything really wrong with the car after the trip.
 
Last edited:
The running out of gas story made me think of a story from my past, as well. I used to have a 1968 Dodge Dart that had a rebuilt 318 under the hood. It was pretty quick, but also pretty thirsty. Usually got 12-13 mpg in local driving, maybe 15-16 on the highway...17 if I was gentle.

Well, in 1995 I drove it out to Oklahoma, to see some friends. At one point of the trip, going across Arkansas, I think, I hit a long patch of highway with construction, where they had actually closed down an exit or two. And, gas stations were a lot more scarce out there than what I thought they'd be. Once the fuel gauge started dropping low, and I realized there wasn't a gas station every 50 feet, I started "Old-ladying" it until I found a gas station. Best MPG I ever got out of a tank with that car...17.8!
 
I came across this on MarketWatch, "Is sequence risk really a big deal for retirees?"

One thing that I did find interesting:

"According to Estrada, a 4% withdrawal strategy actually failed just 4.4% of the time since 1900. And when analyzing the particular sequence of bad annual returns that led to those failures, he found that a different sequence of those same returns would have prevented those failures more than 90% of the time. So the odds of sequence risk sabotaging your retirement are quite low."

I was hoping to get others' reaction to this article.

I think SORR is adequately contemplated within the 4% SWR. That rate is in fact quite conservative. I think anyone experiencing very negative market returns in the first few years of retirement will also tend to mitigate that by reducing spending. Retirees' tendency to reduce spending in the middle years of retirement is another mitigating factor.

Of course one can further mitigate it by using a 3.5% withdrawal rate 3% 2.5% or whatever. Of course in order to do so you have to reduce lifestyle or work/save for more years.

So to me it is something to be aware of, but not something to spend much time worrying about.
 
Last edited:
I don't worry too much about SORR, but I do try to plan and do things to minimize any negative effects.
Meaning I'll take SS at 70 or at least the higher earner will, as we have no pension to speak of.
I also consider keeping some $ in a 401K instead of all in IRA in case we run over a group of kindergarten children crossing the road as the lawsuit would be expensive beyond insurance.

I think the chance of SORR is tiny.
However perhaps that is simply a result of recent historical bias, and I'm too relaxed.

The last 60 yrs have been quite calm (some recessions, tiny conflicts, bunch of posturing like Cuba missile crisis) compared to the previous 60 yr block (WWI, the Depression, The Dust Bowl, WWII).

In the end, we can only prepare or be as smart as we are able, and the rest is up to fate.
 
When I started to contemplate ER, I thought about SORR a lot.

I am 64 now, and with 8 years of ER under my belt, I don't think about SORR anymore. I think more about running out of time.


Heh heh heh... Of course, YMMV.
 
What is "failure" defined as in the failure rate?

My personal definition is having less then about half of our current portfolio after inflation adjusting.

Many people seem to use near zero as a "failure". That is totally unrealistic.
 
I am concerned appropriately about SORR, as I am lots of risks, but I find all these 4% SWR over 30 years from an inflation-adjusted portfolio with a 50/50 domestic static allocation to be very, very particular to those inputs. In the real world versus the academic one, there is Social Security, Medicare, home equity and unlimited spending adjustments that a retiree will make well before someone wakes up one day after 30 years and says, “Oh, CRAP! I’m penniless!”


+1
 
Terrible article.
When I started to contemplate ER, I thought about SORR a lot.

I am 64 now, and with 8 years of ER under my belt, I don't think about SORR anymore. I think more about running out of time.


Heh heh heh... Of course, YMMV.
62 and 7 years in I think I'm subject to SORR. Not as worried as I'm sure I can adjust and will live well, but I've become accustomed to being in the top 10%
What is "failure" defined as in the failure rate?

My personal definition is having less then about half of our current portfolio after inflation adjusting.

Many people seem to use near zero as a "failure". That is totally unrealistic.
That's one of the great things of this game - each of us can decide what perimeters constitute a win.
Personally, if I hit 70 then start SS, own my home, and have $250k of today dollars stashed I'll consider it game won.
 
Another digression from the SORR discussion here, but the anecdotes about making it to Blythe, CA on a tank of gas remind me of an anecdote from my dim past 40 years ago. While still in college I was driving from Phoenix back to LA at the end of spring break. Being young and foolish the empty desert of western Arizona encouraged me to put my foot down and see how fast my crappy car would go. So I got it up to maybe 120, glanced down and saw the temperature gauge buried in the red. Yikes. Shut down the engine immediately and was able to coast for a couple of miles to the little town of Quartzite, AZ and its one service station. I got a mechanic there to take a look at my car. After doing so he came over to me and said the four most terrifying words in the English language:
"You'll like our town"
 
What is "failure" defined as in the failure rate?

My personal definition is having less then about half of our current portfolio after inflation adjusting.

Many people seem to use near zero as a "failure". That is totally unrealistic.

I was dismissive of this work upthread, although this was based solely on the MarketWatch description of it; I have not found the paper.

However, I wanted to point out that I first became familiar with this very author, Javier Estrada, for his work on trying to formulate a better measurement of "failure." As Lsbcal points out, reaching zero = "failure" is a poor metric. Running $10 short in your 30th year is a lot different than running out of money entirely in your 20th year.

Estrada introduces a "coverage ratio" to rank a withdrawal strategy. It takes into account the fraction of planned retirement spending achieved, and also weights it by a given utility function.

The paper can be found here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3135125
 
What is "failure" defined as in the failure rate?

My personal definition is having less then about half of our current portfolio after inflation adjusting.

Many people seem to use near zero as a "failure". That is totally unrealistic.

I actually define failure as portfolio value dropping below 100% of inflation adjusted portfolio value! This is due to the fact that I have a special need child and I am planning for his retirement as well. Thankfully I can "run" this scenario using FireCalc. I simply pay attention to the lowest portfolio value in the FireCalc results page and compare that with the starting portfolio value. I am "safe" as long as they are same!
 
GravitySucks makes the point that each of us can decide what constitutes a good outcome.

We should just make sure the outcome of our strategy would have worked with past data. If a minimum is reached during the years we are retired that would be unacceptable then consider changing your strategy now.

I usually look at the 1966 start as that reaches a minimum in 1982 before heading up again. Turns out the stock/bond ratio is not that big a factor for that start date to 1982.
 
Back
Top Bottom