"Sequence of returns risk"

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
tennisplyr
Posts: 1746
Joined: Tue Jan 28, 2014 1:53 pm
Location: Sarasota, FL

"Sequence of returns risk"

Post by tennisplyr » Mon Nov 28, 2016 5:06 pm

I'm in my mid sixties and have been retired for about 6 years. Since these early years of my retirement were in a bull market, does that mean, all things being equal, I stand a better than average chance of my portfolio lasting for my lifetime. My portfolio currently is larger than it was when I retired. I realize there are no guarantees.
Those who move forward with a happy spirit will find that things always work out.

User avatar
Taylor Larimore
Advisory Board
Posts: 27300
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

Re: "Sequence of returns risk"

Post by Taylor Larimore » Mon Nov 28, 2016 5:20 pm

tennisplyr wrote:I'm in my mid sixties and have been retired for about 6 years. Since these early years of my retirement were in a bull market, does that mean, all things being equal, I stand a better than average chance of my portfolio lasting for my lifetime. My portfolio currently is larger than it was when I retired.
tennisplyr:

The answer is "yes." You are fortunate to have retired in a bull market. This quote is from Investopedia:
If you happen to retire immediately before a prolonged bull market, there really isn't anything to worry about. However, if you end up retiring prior to a bear market, your retirement dreams could crumble if your portfolio is unprepared.
http://www.investopedia.com/articles/re ... rement.asp

Best wishes
Taylor
"Simplicity is the master key to financial success." -- Jack Bogle

lack_ey
Posts: 6603
Joined: Wed Nov 19, 2014 11:55 pm

Re: "Sequence of returns risk"

Post by lack_ey » Mon Nov 28, 2016 5:30 pm

Well, having more money doesn't hurt, does it? By all means this indicates your chances have improved, yes, but the answer is maybe a bit more nuanced depending on how you think of probability and returns.

It's not clear in which way you really mean that and which things you think are kept equal. I think the canonical formulation for discussing sequence of returns risk is to imagine different N-year sequences of returns ending in X% annualized return. For a given X, your money remaining for a given level of withdrawals is path dependent, heavily influenced by the particular sequence of returns. For a retiree it's better to have the relatively good returns frontloaded and happening earlier.

However, in real life we don't know what X is going to be for your N years. Or even which N matters to you, to be honest. Your actual position now is even better than in the previous paragraph's formulation because there's no guarantee at all that your good returns the last 6 years are taking away from future returns. If 6 years ago you thought that returns of Y% annualized were most likely over the next N years, presumably your estimate today of future returns for the remaining N-6 years is higher than the result that would produce Y% over the full period.

Also, it's not as if most sequences of returns are all that scary for a retiree. You generally don't plan on just barely squeaking by if the average outcome happens. What you're scared of is getting unlucky, with returns lower than expected. If nothing else you know now that you didn't start off by getting unlucky, hitting a 5 percentile outcome for example.

Dandy
Posts: 5284
Joined: Sun Apr 25, 2010 7:42 pm

Re: "Sequence of returns risk"

Post by Dandy » Mon Nov 28, 2016 5:34 pm

Better than 6 years ago for sure. But with 20+ years likely to go a lot can happen. I'm almost 69 and starting to feel even better -- at least asset/withdrawal wise. I'm sure there will be a substantial dip ahead after a long bull market run. But, the appreciation in assets especially during the wait to collect SS at age 70 has been appreciated.

A major decline at the outset of retirement would make the financial pain to fund the wait to collect SS at 70 hard to do.

Random Walker
Posts: 2979
Joined: Fri Feb 23, 2007 8:21 pm

Re: "Sequence of returns risk"

Post by Random Walker » Mon Nov 28, 2016 7:53 pm

I agree, the answer to your question is yes.
I believe those first few years of retirement are critical. People frequently say things like "I'm retired, I can't afford to wait a long time for a tilted portfolio to outperform" or "I can't risk a long period of underperformance by tilting". I think these people have it backwards. The shorter the time period, perhaps the more important it is to be diversified across potential sources of return. In any given single year, there is no telling which factors will do well or poorly. I think sequence of returns risk is one of the strongest arguments for heavily diversifying the equity portion of the portfolio across factors and having a substantial dose of safe bonds.

Dave

User avatar
FIREchief
Posts: 2625
Joined: Fri Aug 19, 2016 6:40 pm

Re: "Sequence of returns risk"

Post by FIREchief » Mon Nov 28, 2016 8:03 pm

Random Walker wrote: I think sequence of returns risk is one of the strongest arguments for heavily diversifying the equity portion of the portfolio across factors and having a substantial dose of safe bonds.

Dave
In other words, a simple "three fund" portfolio???
I am not a lawyer, accountant or financial advisor. Any advice or suggestions that I may provide shall be considered for entertainment purposes only.

AlohaJoe
Posts: 3668
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: "Sequence of returns risk"

Post by AlohaJoe » Mon Nov 28, 2016 8:10 pm

Without knowing how much you are withdrawing it is hard to offer a meaningful statement.

lack_ey
Posts: 6603
Joined: Wed Nov 19, 2014 11:55 pm

Re: "Sequence of returns risk"

Post by lack_ey » Mon Nov 28, 2016 8:31 pm

FIREchief wrote:
Random Walker wrote: I think sequence of returns risk is one of the strongest arguments for heavily diversifying the equity portion of the portfolio across factors and having a substantial dose of safe bonds.

Dave
In other words, a simple "three fund" portfolio???
If that was not meant facetiously—I can never tell and have been burned both ways assuming the wrong thing on similar statements—no. In case somebody is reading this and needs the explanation...

That was an argument for a smaller equity portfolio tilted to be riskier and encompassing more purported sources of return, paired with a higher percentage of bonds, rather than using total market funds. The premise is to jack up the risk and hopefully the return of the equity component relative to the total market, which allows you to use less of it. This effectively reduces both positive and negative tails, making outlier outcomes less likely. Probably. Some imagine that this probably reduces (expected) returns as well, but others don't. Usually this argument comes from the pro-factor investing crowd who might view diversification in terms of sources of returns (factors). Regardless of whether or not you subscribe to those kinds of assumptions and interpretations about the market, the total market funds in the traditional three-fund portfolio have zero factor loadings outside of the market.

Random Walker
Posts: 2979
Joined: Fri Feb 23, 2007 8:21 pm

Re: "Sequence of returns risk"

Post by Random Walker » Mon Nov 28, 2016 8:33 pm

Fire chief,
By three fund portfolio you mean Total Stock Market, Total International, Total Bond? No I definitely do not mean that. Total market portfolios definitely have wide exposure to equities, but not net exposure to size and value. On the equity side, they are effectively exposed to a only a single factor, market beta. There are other drivers of returns which are significant and have low correlations to market beta: size, value, profitability, momentum. Diversifying across all these uncorrelated factors on the equity side minimizes the risk of getting slammed by a single year where market beta does very poorly. In fact, since size and value provide additional expected returns above market beta, one can increase the safe bond component of a portfolio and keep expected return the same as a TSM portfolio. This keeps expected return the same but lessens the risk of getting hammered in any single year. For example a tilted 40/60 portfolio may have the same expected return as a TSM 60/40 portfolio with a lot less risk of suffering one really bad year at a time when the investor can least afford it. Take a look at Larry's book Reducing The Risk Of Black Swans.

Dave

Random Walker
Posts: 2979
Joined: Fri Feb 23, 2007 8:21 pm

Re: "Sequence of returns risk"

Post by Random Walker » Mon Nov 28, 2016 8:37 pm

Lack_ey,
Summarized my thoughts pretty well. It's not just the higher expected returns of small and value though, it's also the lack of correlation between the factors.

Dave

AlohaJoe
Posts: 3668
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: "Sequence of returns risk"

Post by AlohaJoe » Mon Nov 28, 2016 9:36 pm

AlohaJoe wrote:Without knowing how much you are withdrawing it is hard to offer a meaningful statement.
To elaborate on what I meant, with a reasonable portfolio you could have withdrawn almost 9% a year over the past 6-years and still ended up with a bigger portfolio than when you started. If you are withdrawing 9% a year you are "safer" than you were before but certainly not safe in any meaningful sense over the next two decades.

On the other hand, if you were withdrawing 1% a year over the past 6 years, you are "safer" but that's almost meaningless because you went from a 0.00000001% chance of failure to a 0.0000000000000001% chance of failure.

feh
Posts: 1229
Joined: Sat Dec 15, 2012 11:39 am

Re: "Sequence of returns risk"

Post by feh » Tue Nov 29, 2016 1:22 pm

tennisplyr wrote:I'm in my mid sixties and have been retired for about 6 years. Since these early years of my retirement were in a bull market, does that mean, all things being equal, I stand a better than average chance of my portfolio lasting for my lifetime. My portfolio currently is larger than it was when I retired. I realize there are no guarantees.
What happened in the past is not relevant. Pretend you were retiring today - what does firecalc say about your time horizon and current portfolio value?

rbaldini
Posts: 900
Joined: Mon Mar 23, 2015 3:20 pm

Re: "Sequence of returns risk"

Post by rbaldini » Tue Nov 29, 2016 1:42 pm

tennisplyr wrote:I'm in my mid sixties and have been retired for about 6 years. Since these early years of my retirement were in a bull market, does that mean, all things being equal, I stand a better than average chance of my portfolio lasting for my lifetime. My portfolio currently is larger than it was when I retired. I realize there are no guarantees.
Multiplication is commutative. In the long run, it doesn't matter what order the returns come in. In the short term, I'd say it's probably better to lose x% of your money when you have a lot of it, so you can still afford to eat and pay the rent.

On the other hand, to answer the question, "was it good that my investments increased at a rate higher than inflation?": of course, yes. Or in other words, gaining money is better than losing money. So the bull market was good for you in the same way that it was good for everyone who invested in it.

feh
Posts: 1229
Joined: Sat Dec 15, 2012 11:39 am

Re: "Sequence of returns risk"

Post by feh » Tue Nov 29, 2016 1:45 pm

rbaldini wrote: Multiplication is commutative. In the long run, it doesn't matter what order the returns come in.
Absolutely false: http://www.investopedia.com/terms/s/sequence-risk.asp

User avatar
Dutch
Posts: 1277
Joined: Thu Jun 27, 2013 2:12 pm

Re: "Sequence of returns risk"

Post by Dutch » Tue Nov 29, 2016 1:45 pm

rbaldini wrote:Multiplication is commutative. In the long run, it doesn't matter what order the returns come in.
You may want to re-think that statement

Wakefield1
Posts: 814
Joined: Mon Nov 14, 2016 10:10 pm

Re: "Sequence of returns risk"

Post by Wakefield1 » Tue Nov 29, 2016 1:50 pm

You have one $. It goes up 50%. You now have $1.50
Your $1.50 now goes down 50%. You now have $0.75 !! :confused

bigred77
Posts: 1997
Joined: Sat Jun 11, 2011 4:53 pm

Re: "Sequence of returns risk"

Post by bigred77 » Tue Nov 29, 2016 1:53 pm

tennisplyr wrote:I'm in my mid sixties and have been retired for about 6 years. Since these early years of my retirement were in a bull market, does that mean, all things being equal, I stand a better than average chance of my portfolio lasting for my lifetime. My portfolio currently is larger than it was when I retired. I realize there are no guarantees.
Assuming you took out roughly 4% of your portfolio balance in year 1, adjusted each year's withdrawal for inflation, had a reasonable portfolio for a retiree (I dunno, something between 1/3 and 2/3 equities), and don't plan to substantially increase withdrawals then yes, your likely, in tremendous shape.

I see sequence of return risk mostly discussed in terms of the "4% rule" but in reality it's always a factor in sustainability. If you started with a 6% withdrawal you might not be so safe. If you plan to substantially increase withdrawals now that your portfolio has increased quite a bit in value (something that might even make perfect sense in your situation) you might not be perfectly safe.

If you're as conservative as a the median or average Boglehead, I suspect you can go on worry free.
rbaldini wrote:
tennisplyr wrote:I'm in my mid sixties and have been retired for about 6 years. Since these early years of my retirement were in a bull market, does that mean, all things being equal, I stand a better than average chance of my portfolio lasting for my lifetime. My portfolio currently is larger than it was when I retired. I realize there are no guarantees.
Multiplication is commutative. In the long run, it doesn't matter what order the returns come in.

On the other hand, to answer the question, "was it good that my investments increased at a rate higher than inflation?": of course, yes. Or in other words, gaining money is better than losing money. So the bull market was good for you in the same way that it was good for everyone who invested in it.
I agree returns to a portfolio are commutative, as long as one is not making additions or subtractions from the portfolio in any given year. If a retiree is making withdrawals, then the sequence of returns can matter quite a bit.

rbaldini
Posts: 900
Joined: Mon Mar 23, 2015 3:20 pm

Re: "Sequence of returns risk"

Post by rbaldini » Tue Nov 29, 2016 1:55 pm

a*b = b*a

$1000*(1-.1)*(1+.1) = $1000*(1+.1)*(1-.1)

That's all I mean.

Now the sequence does obviously effect the path that you take to the final point, and that is surely of interest.

lack_ey
Posts: 6603
Joined: Wed Nov 19, 2014 11:55 pm

Re: "Sequence of returns risk"

Post by lack_ey » Tue Nov 29, 2016 1:56 pm

rbaldini wrote:Multiplication is commutative. In the long run, it doesn't matter what order the returns come in. In the short term, I'd say it's probably better to lose x% of your money when you have a lot of it, so you can still afford to eat and pay the rent.

On the other hand, to answer the question, "was it good that my investments increased at a rate higher than inflation?": of course, yes. Or in other words, gaining money is better than losing money. So the bull market was good for you in the same way that it was good for everyone who invested in it.
We're talking about somebody withdrawing from the allocation, so order matters significantly.

If your portfolio loses 90% early on and you need to withdraw 3% of the initial amount annually (around 33% of the current) for expenses, it doesn't matter if five years later returns are 50% a year for a couple decades. You won't have any money left.

It also clearly matters if making contributions (for a given long-term return, you'd prefer the reverse order, with the better returns later and worse returns earlier).

rbaldini
Posts: 900
Joined: Mon Mar 23, 2015 3:20 pm

Re: "Sequence of returns risk"

Post by rbaldini » Tue Nov 29, 2016 2:01 pm

bigred77 wrote: I agree returns to a portfolio are commutative, as long as one is not making additions or subtractions from the portfolio in any given year. If a retiree is making withdrawals, then the sequence of returns can matter quite a bit.
So I think what you're saying can be boiled down to
(a-c)*b =/= (b-c)*a
That's true: if you're going to withdraw fixed amounts from your account (i.e., that do not depend on the size of the account), then sequence matters.

On the other hand, if you only withdraw at fixed *percentages* (so it does depend on size of account) (say, 3% of the fund's value every year), then sequence doesn't matter:
a*(1-c)*b = b*(1-c)*a

EDIT: I imagined that the fixed percentage removed is applied to the current size of fund. If, as lack_ey put it, your percentage depends on the fund's initial size, then it's a fixed amount withdrawal, so order does matter.

User avatar
Dutch
Posts: 1277
Joined: Thu Jun 27, 2013 2:12 pm

Re: "Sequence of returns risk"

Post by Dutch » Tue Nov 29, 2016 2:13 pm

rbaldini wrote:On the other hand, if you only withdraw at fixed *percentages* (so it does depend on size of account) (say, 3% of the fund's value every year), then sequence doesn't matter:
a*b*(1-c)*d = a*d*(1-c)*b

EDIT: I imagined that the fixed percentage removed is applied to the current size of fund. If, as lack_ey put it, your percentage depends on the fund's initial size, then it's a fixed amount withdrawal, so order does matter.
Right ... and if all your expenses go up and down proportionally with the stock market, that would be nice as well. But the real world doesn't work that way.

rbaldini
Posts: 900
Joined: Mon Mar 23, 2015 3:20 pm

Re: "Sequence of returns risk"

Post by rbaldini » Tue Nov 29, 2016 2:15 pm

Dutch wrote: Right ... and if all your expenses go up and down proportionally with the stock market, that would be nice as well. But the real world doesn't work that way.
Fair enough. I imagine reality is between the two extremes: when going gets tough, you probably do spend less, but not exactly in proportion to your decline in value. So, I'll concede: generally, order matters.

lack_ey
Posts: 6603
Joined: Wed Nov 19, 2014 11:55 pm

Re: "Sequence of returns risk"

Post by lack_ey » Tue Nov 29, 2016 2:21 pm

rbaldini wrote:
Dutch wrote: Right ... and if all your expenses go up and down proportionally with the stock market, that would be nice as well. But the real world doesn't work that way.
Fair enough. I imagine reality is between the two extremes: when going gets tough, you probably do spend less, but not exactly in proportion to your decline in value. So, I'll concede: generally, order matters.
Yes, exactly, you wouldn't actually withdraw amounts independent of current value but spending cannot scale like that.

And in any case for modeling purposes, you need to penalize a worse outcome. If you're withdrawing less because that's prudent based on the market returns, that in of itself is already a negative.

2retire
Posts: 369
Joined: Wed Jun 13, 2012 9:00 am

Re: "Sequence of returns risk"

Post by 2retire » Tue Nov 29, 2016 2:23 pm

Yes. As a matter of fact, if you are up more than 50%, you may even be able to ratchet up your withdraw percent. See this article my Kitces.

https://www.kitces.com/blog/the-ratchet ... he-4-rule/

longinvest
Posts: 2907
Joined: Sat Aug 11, 2012 8:44 am

Re: "Sequence of returns risk"

Post by longinvest » Tue Nov 29, 2016 2:25 pm

rbaldini wrote: EDIT: I imagined that the fixed percentage removed is applied to the current size of fund. If, as lack_ey put it, your percentage depends on the fund's initial size, then it's a fixed amount withdrawal, so order does matter.
Many Bogleheads know by now (given past forum debates) that taking a fixed inflation-adjusted amount as yearly withdrawal (the famous SWR method) is foolish. It lets most of its adopters die as the richest people in the graveyard while bankrupting most of the rest. Anyway, no rational person would be able to stay the course with it during a severe bear market.

Many Bogleheads know how to use a flexible withdrawal method, instead. One can search the forum for "Taylor Larimore withdrawal method" or "VPW".

There used to be an SWR thread of the week (or two, or three, or more), caused by people suddenly realizing that SWR is dumb and foolish, and thinking that it was the only way to take money out of a portfolio, thanks to all the noise in the financial press.
Last edited by longinvest on Tue Nov 29, 2016 5:48 pm, edited 3 times in total.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VLB/ZRR

MIretired
Posts: 733
Joined: Fri Sep 06, 2013 12:35 pm

Re: "Sequence of returns risk"

Post by MIretired » Tue Nov 29, 2016 2:36 pm

I view it simplified to two things happening because of the sequence of returns you get starting at the withdrawal period.
The yield you take from your portfolio year by year, and your dwindling time of life remaining for the remaining portfolio to cover, which makes the remainder of retirement funding more certain.

If you weren't either adding or withdrawing from your portfolio, there wouldn't be any "sequence" of returns risk.
When adding, a lower portfolio value gives you a higher percentage of portfolio added for a constant dollar added.
When withdrawing, a lower portfolio value gives you a higher percentage of portfolio withdrawn for a constant dollar taken.

With no additions or subtractions, your portfolio can take any path, and with the same CAGR, will end in the same place.

User avatar
Dutch
Posts: 1277
Joined: Thu Jun 27, 2013 2:12 pm

Re: "Sequence of returns risk"

Post by Dutch » Tue Nov 29, 2016 3:26 pm

rbaldini wrote:On the other hand, if you only withdraw at fixed *percentages* (so it does depend on size of account) (say, 3% of the fund's value every year), then sequence doesn't matter:
a*(1-c)*b = b*(1-c)*a
On second thought, even in this fictional example, I believe sequence matters greatly.

I hope somebody with better math skills can back me up on this. But I believe that the sum of c (withdrawals) is subject to sequence of returns.

Maybe the terminal portfolio balance is the same. But how important is that?

User avatar
Phineas J. Whoopee
Posts: 7362
Joined: Sun Dec 18, 2011 6:18 pm

Re: "Sequence of returns risk"

Post by Phineas J. Whoopee » Tue Nov 29, 2016 3:28 pm

Wakefield1 wrote:You have one $. It goes up 50%. You now have $1.50
Your $1.50 now goes down 50%. You now have $0.75 !! :confused
You have $1.00. It goes up by $0.50. You now have $1.50.

You have $1.50. It goes down by $0.50. You now have $1.00.

$0.50 is 50% of $1.00, and 33% of $1.50.

$0.75 is 75% of $1.00, and 50% of $1.50.

If your holding goes up by 50% of $1.00, it went up by $0.50. If it goes down by 50% of $1.00, it goes down by $0.50.

If your holding goes up by 50% of $1.50, it went up by $0.75. If it goes down by 50% of $1.50, it goes down by $0.75.


Here's the anti-confusion bit:

If your holding goes up by 50% of $1.00, it went up by $0.50. If it subsequently goes down by 50% of $1.50, it goes down by $0.75.

Percentage is a ratio between two numbers. In your example the number being compared to is 1.00 to begin with, and 1.50 later. Of course the ratio between two numbers will be different if one of the numbers in the comparison is different.

Helpful?

PJW

lack_ey
Posts: 6603
Joined: Wed Nov 19, 2014 11:55 pm

Re: "Sequence of returns risk"

Post by lack_ey » Tue Nov 29, 2016 3:47 pm

Dutch wrote:
rbaldini wrote:On the other hand, if you only withdraw at fixed *percentages* (so it does depend on size of account) (say, 3% of the fund's value every year), then sequence doesn't matter:
a*(1-c)*b = b*(1-c)*a
On second thought, even in this fictional example, I believe sequence matters greatly.

I hope somebody with better math skills can back me up on this. But I believe that the sum of c (withdrawals) is subject to sequence of returns.

Maybe the terminal portfolio balance is the same. But how important is that?
Maybe I misunderstand where you're going with this, but summing c over different years doesn't really say much of anything. In that formulation, c is a fraction like 1/30. It is not withdrawals in dollar amounts, which I think is the more meaningful measure that most people assume for starters or care about.

I think you might be imagining summing over total cumulative withdrawals (measured by dollars), which would be something else.

Wakefield1 wrote:You have one $. It goes up 50%. You now have $1.50
Your $1.50 now goes down 50%. You now have $0.75 !! :confused
For 50% up and 50% down, that is 1.5 and 0.5. The geometric mean between 1.5 and 0.5 is sqrt(1.5*0.5) = 0.866, which is not 1.

Point is that 1.5 * 0.5 = 0.5 * 1.5. Order in multiplication doesn't matter.

MIretired
Posts: 733
Joined: Fri Sep 06, 2013 12:35 pm

Re: "Sequence of returns risk"

Post by MIretired » Tue Nov 29, 2016 3:54 pm

Dutch wrote:
rbaldini wrote:On the other hand, if you only withdraw at fixed *percentages* (so it does depend on size of account) (say, 3% of the fund's value every year), then sequence doesn't matter:
a*(1-c)*b = b*(1-c)*a
On second thought, even in this fictional example, I believe sequence matters greatly.

I hope somebody with better math skills can back me up on this. But I believe that the sum of c (withdrawals) is subject to sequence of returns.

Maybe the terminal portfolio balance is the same. But how important is that?
How about: ?
I would say that the sum of "c" would be the same if taking a constant percentage each year. If CAGR is same regardless of path.
If every year you take out a (1+c) rate. Then port: [(1+r1)(1+r2)...(1+rn)]/(1+c) = (1+CAGR)^n/(1+c)

And the sum of c's would be c *(1+CAGR) * (init. port. value) * n. ??

Edit 3x

Alright. The /(1+c) doesn't work. It's the same situation as computing returns while not reinvesting dividends. They are point events and not continual compounding.
Last edited by MIretired on Tue Nov 29, 2016 6:47 pm, edited 4 times in total.

User avatar
Dutch
Posts: 1277
Joined: Thu Jun 27, 2013 2:12 pm

Re: "Sequence of returns risk"

Post by Dutch » Tue Nov 29, 2016 4:21 pm

lack_ey wrote: Maybe I misunderstand where you're going with this, but summing c over different years doesn't really say much of anything. In that formulation, c is a fraction like 1/30. It is not withdrawals in dollar amounts, which I think is the more meaningful measure that most people assume for starters or care about.

I think you might be imagining summing over total cumulative withdrawals (measured by dollars), which would be something else.
Right, I just ran two simple scenarios in a spreadsheet where annual withdrawals are a percentage of the balance at the beginning of the year. From those examples, it's clear that the portfolio ending balance (after 10 years in my simple example) in that particular case, is independent of sequence of returns.

But the sum of the annual withdrawals is very much dependent on sequence of returns.

longinvest
Posts: 2907
Joined: Sat Aug 11, 2012 8:44 am

Re: "Sequence of returns risk"

Post by longinvest » Tue Nov 29, 2016 4:32 pm

This post by Skjoldur explains the difference between "sequence of returns" risk and "market returns" risk:

viewtopic.php?f=2&t=190721&start=50#p2900256
skjoldur wrote:Maybe I can help illustrate the difference between what longinvest refers to as 'market risk' and sequence of returns risk. I have made three graphs using made up numbers.

I made 10 random returns, and then scrambled those 10 returns into a variety of different sequences. The first picture shows a $1M portfolio following each of these different paths. Note that the final portfolio value is identical in each case. Holding a portfolio without adding or withdrawing is not subject to sequence of returns risk. The sequence does not matter.

Image

The next image shows the same portfolio with a 4% annual withdrawal. This is not a constant dollar withdrawal, it is a constant percentage withdrawal. Note, that the end values of all the paths are the same. Constant percentage withdrawal is not subject to sequences of returns risk. The portfolios are subject to market risk, in that the ending value could be high or low but the sequence does not matter.

Image

The final image shows the same portfolio with a $40K withdrawal, this is the dreaded constant dollar withdrawal. Note that in this case, the final portfolio values are all different. That difference is sequence of returns risk. In this case, poor returns early combined with constant withdrawals results in varied outcomes.

Image

It turns out (and this surprised me mathematically, but longinvest demonstrated it another thread) that a sequence of varied percentage withdrawals has the same property as a constant percentage withdrawal. In other words, VPW is also mathematically immune to sequence of returns risk.

So here is a bonus picture. In this one, the portfolios are all subject to the same sequence of varied percentage withdrawals. You can see that the end portfolio values are all the same once again.

Image

So VPW is immune to sequence of returns risk with regard to final portfolio value.

That does not mean that VPW is not 'risky' or that it magically fixes the fact that the markets themselves are 'risky.' But (along with similar percentage based withdrawal methods) it does not suffer from sequence of returns risk. How cool is that!
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VLB/ZRR

MIretired
Posts: 733
Joined: Fri Sep 06, 2013 12:35 pm

Re: "Sequence of returns risk"

Post by MIretired » Tue Nov 29, 2016 4:40 pm

^ Yea. That's it A constant withdrawal yield. I thought VPW also used rules based WD. Maybe not.
Although it's front loaded by expected returns, but converges to actual returns, I believe.

rbaldini
Posts: 900
Joined: Mon Mar 23, 2015 3:20 pm

Re: "Sequence of returns risk"

Post by rbaldini » Tue Nov 29, 2016 7:29 pm

Dutch wrote:
rbaldini wrote:On the other hand, if you only withdraw at fixed *percentages* (so it does depend on size of account) (say, 3% of the fund's value every year), then sequence doesn't matter:
a*(1-c)*b = b*(1-c)*a
On second thought, even in this fictional example, I believe sequence matters greatly.

I hope somebody with better math skills can back me up on this. But I believe that the sum of c (withdrawals) is subject to sequence of returns.

Maybe the terminal portfolio balance is the same. But how important is that?
Right - the terminal portfolio balance is the same.

And yes, we can prove that the total amount withdrawn does depend on sequence, even with the above example: in the first case, you withdraw a*c; in the second case, you withdraw b*c. I can see why this is important.

EDIT: A more complete demonstration, I suppose, is to add one more withdrawal at the end. If a precedes b, then we have total withdrawal of a*c + a*(1-c)*b*c. If b precedes a, then it's b*c + a*(1-c)*b*c. Withdrawal at the end is necessarily the same, but intervening amount can vary.

(I'll admit that, being 28 years old, I think more about accumulation than withdrawal. That's probably why I took us on this tangent about terminal portfolio balance in the first place - I misunderstood the point of all this.)

rbaldini
Posts: 900
Joined: Mon Mar 23, 2015 3:20 pm

Re: "Sequence of returns risk"

Post by rbaldini » Tue Nov 29, 2016 8:02 pm

BTW, sure - sequence of returns effects your total income in retirement (assuming you don't do a fixed amount withdrawal). But apart from being an interesting mathematical fact, why should we much care about this? How is it an actionable piece of information? I don't see how it is, unless we could magically reorder future returns.

lack_ey
Posts: 6603
Joined: Wed Nov 19, 2014 11:55 pm

Re: "Sequence of returns risk"

Post by lack_ey » Tue Nov 29, 2016 8:28 pm

rbaldini wrote:BTW, sure - sequence of returns effects your total income in retirement (assuming you don't do a fixed amount withdrawal). But apart from being an interesting mathematical fact, why should we much care about this? How is it an actionable piece of information? I don't see how it is, unless we could magically reorder future returns.
It's usually brought up from a planning perspective to warn against making poor assumptions about future returns and the resulting impact on potential spending ability.

Financial literacy and understanding of probability aren't particularly high in the general populace (actually, among financial professionals too, all too often). Many are given or come up with a number like X% returns and conceptualize this wrongly as X% every year and run calculations and spreadsheets based on that. In reality this is too optimistic, even if returns are indeed X% over the full period, as spending power with reasonable assumptions about withdrawals depends on the sample path and many of them will be less friendly to the retiree.

In that way it is an instructional concept, one that suggests a little more caution in setting up withdrawal strategies and rates than intuition might otherwise suggest. In a cynical way it should also be a popular idea among the advisory community to emphasize, as many are paid directly or indirectly based on AUM, and anything to convince a client to spend less money means more AUM remaining and thus more money for the advisor.

bigred77
Posts: 1997
Joined: Sat Jun 11, 2011 4:53 pm

Re: "Sequence of returns risk"

Post by bigred77 » Tue Nov 29, 2016 10:59 pm

rbaldini wrote:BTW, sure - sequence of returns effects your total income in retirement (assuming you don't do a fixed amount withdrawal). But apart from being an interesting mathematical fact, why should we much care about this? How is it an actionable piece of information? I don't see how it is, unless we could magically reorder future returns.
I think it serves as a great reminder to derisk ones portfolio once withdrawals start. Knowing that 80/20 portfolio has higher expected returns than a 50/50 portfolio may lead you to select a higher equity portfolio. Knowing that drastic plunges in portfolio value hurt much worse when withdrawals are being taken help encourage a more conservative AA.

Random Walker
Posts: 2979
Joined: Fri Feb 23, 2007 8:21 pm

Re: "Sequence of returns risk"

Post by Random Walker » Wed Nov 30, 2016 12:44 am

Rbaldini,
Sequence of returns is obviously a huge issue for someone in early retirement; the withdrawn money cannot recover after a bear. But there is something I believe one can do to minimize SOR risk. If one could construct a portfolio with the same mean expected return, but a narrower dispersion of potential returns and lest subject to fat tails, then that would blunt the damage of a couple bad years at just the wrong time. Moreover, by dampening portfolio volatility, this would result in greater terminal value by lessening the drag of volatility on the CAGR. This can be achieved by diversifying across weakly correlated, uncorrelated, and possibly even negatively correlated sources of return. Maybe market beta gets killed at the start of ones retirement, but a bigger dose of bonds and perhaps momentum, size, or value help blunt the losses.

Dave

itstoomuch
Posts: 5343
Joined: Mon Dec 15, 2014 12:17 pm
Location: midValley OR

Re: "Sequence of returns risk"

Post by itstoomuch » Wed Nov 30, 2016 12:54 am

+1 Random Walker.
The sacrifice, as you mention, is that you give up the wider dispersion and fatter head/tail for safety in the narrow dispersion and thinner head/tail. To compensate, I keep a Discretionary account (now about 20% or retirement assets) to do moderate swings.
YMMV
Rev012718; 4 Incm stream buckets: SS+pension; dfr'd GLWB VA & FI anntys, by time & $$ laddered; Discretionary; Rentals. LTCi. Own, not asset. Tax TBT%. Early SS. FundRatio (FR) >1.1 67/70yo

rbaldini
Posts: 900
Joined: Mon Mar 23, 2015 3:20 pm

Re: "Sequence of returns risk"

Post by rbaldini » Wed Nov 30, 2016 9:46 am

bigred77 wrote: I think it serves as a great reminder to derisk ones portfolio once withdrawals start. Knowing that 80/20 portfolio has higher expected returns than a 50/50 portfolio may lead you to select a higher equity portfolio. Knowing that drastic plunges in portfolio value hurt much worse when withdrawals are being taken help encourage a more conservative AA.
That "drastic plunges in portfolio value hurt much worse when withdrawals are being taken" isn't the point of sequence of returns risk. Sequence of returns risk is about the order in which the plunges occur during withdrawal. Your point is already clear if you know about market volatility. Sequence of returns doesn't really add anything here.
Random Walker wrote: If one could construct a portfolio with the same mean expected return, but a narrower dispersion of potential returns and lest subject to fat tails, then that would blunt the damage of a couple bad years at just the wrong time. Moreover, by dampening portfolio volatility, this would result in greater terminal value by lessening the drag of volatility on the CAGR. This can be achieved by diversifying across weakly correlated, uncorrelated, and possibly even negatively correlated sources of return.
Again, I don't see how sequence of returns adds anything here. This is all about reducing your risk by investing more conservatively. You don't need to know about sequence of returns to know this, and it doesn't really add anything. You pick an investment strategy prior to retirement that allows you to withdraw for a long time, even if times are hard. You update this strategy as you go along and conditions change.
itstoomuch wrote: The sacrifice, as you mention, is that you give up the wider dispersion and fatter head/tail for safety in the narrow dispersion and thinner head/tail.
Again, all market volatility. Nothing about sequence of returns.

I suppose one actionable thing you could do is pick an investment + withdrawal strategy that is more conservative at the beginning of retirement and more aggressive later. This would guard against the problem of losing lots of money at the beginning of your retirement. But this isn't really cogent, because every year is the new "now": you're always at risk of the future sequence of returns risk, regardless of how far you are into retirement. A much more logical approach just ignores sequence of returns risk (or, rather, hides it and only addresses it implicitly): simulate many possible futures of market returns and pick a withdrawal rate that makes sense. Simple.

Random Walker
Posts: 2979
Joined: Fri Feb 23, 2007 8:21 pm

Re: "Sequence of returns risk"

Post by Random Walker » Wed Nov 30, 2016 10:20 am

Rbaldini,
I disagree with your response to me above. If two portfolios have the same expected return, but different Sharpe ratios, I wouldn't call one portfolio more conservative. I'd call it more efficient. It's just more diversified across potential sources of return and types of risk.

Dave

2015
Posts: 1827
Joined: Mon Feb 10, 2014 2:32 pm

Re: "Sequence of returns risk"

Post by 2015 » Wed Nov 30, 2016 2:02 pm

Might not another method of "derisking" be to reduce withdrawals (i.e., discretionary expenses) during a bad "sequence" thereby reducing the damage?

lack_ey
Posts: 6603
Joined: Wed Nov 19, 2014 11:55 pm

Re: "Sequence of returns risk"

Post by lack_ey » Wed Nov 30, 2016 2:40 pm

rbaldini wrote:I suppose one actionable thing you could do is pick an investment + withdrawal strategy that is more conservative at the beginning of retirement and more aggressive later. This would guard against the problem of losing lots of money at the beginning of your retirement. But this isn't really cogent, because every year is the new "now": you're always at risk of the future sequence of returns risk, regardless of how far you are into retirement.
In fact the concept of a rising equity glidepath (start off less aggressive, increase stock allocation further into retirement) has been given some play, though this is based on data and simulations with certain assumptions about mean reversion. But as you say, each year is the new now, and in practice you alter your strategy based on those conditions. Doesn't make sense to plan out a path that's better on average with some assumptions, including the one that you can't adjust your plans in the future based on the actual conditions of the future, which you will know then.
rbaldini wrote:A much more logical approach just ignores sequence of returns risk (or, rather, hides it and only addresses it implicitly): simulate many possible futures of market returns and pick a withdrawal rate that makes sense. Simple.
Yes, it's just a concept for intuition. It's something that would come out of the results if you simulated different scenarios and doesn't need further consideration if you're already accounting for it in that way.

2015 wrote:Might not another method of "derisking" be to reduce withdrawals (i.e., discretionary expenses) during a bad "sequence" thereby reducing the damage?
As a matter of strategy this is the obvious thing to do. Of course you adapt based on what the market gives you and what you know at the time.

Post Reply