So how risky is Amortization Based Withdrawal (ABW)

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills.
Post Reply
Topic Author
redmaw
Posts: 355
Joined: Mon Apr 22, 2019 7:20 am

So how risky is Amortization Based Withdrawal (ABW)

Post by redmaw »

So after a year or two I finally made it through the ABW thread here: viewtopic.php?f=2&t=274243, it left me with the topic question.

I hear you, it can never run out of money before the end date, how can it be risky? Well easy, it computes a withdrawal amount that is smaller than my necessary expenses. I consider this a failure even if its more mild than being broke. This question led me to make my own ABW spreadsheet with a very simple scenario similar to willthrill's in the above linked thread.

I came to a few conclusions:

1) ABW is more of a tool to be incorporated into a larger plan than a withdrawal method. The plan can be as conservative or aggressive as you make it.
2) "You will never run out of money" brushes the effects of sequence risk under the rug. (For fun, go put 100% expected return into the wiki ABW calculator.)
3) ABW is not very helpful when trying to determine if you have enough to retire because of point 2.
4) Aggressive plans/poor sequences can lead to wild swings in allowed withdraws.

Here is my methodology/assumptions:
Starting Age 45 (isn't actually used, but supports some of the other assumptions)
Starting balance: 1M
AA = 75/25
starting n = 600 (calculated monthly, this is 50 years)
g = 0 (no planned withdraw growth)
FV = 0
expected stock return = 1/CAPE
expected bond returns = 1% ( I can't support this, but I didn't have anything better readily available)
actual monthly stock and bond returns (real returns) were shamelessly stolen from the toolkit spreadsheet at earlyretirementnow.com. These are S&P500 and 10 year treasuries.
No future cash flows

I built the spreadsheet to spit out a series of withdraws based on a starting year, then just started playing with the years. I also threw in my favorite variable rate withdrawal rule for comparison ((.5*1/Cape +.01)*balance). I stepped through each decade and generally was left with the sense that the cape rule was more conservative, less variable, but broadly similar, though it often left a large remaining balance versus always ending at 0. I tabulated some results to try to draw some fixed conclusions and interesting trends jumped out. First the cape rule always started lower, I'm sure there is some simple mathematical explanation, but it wasn’t clear to me this had to be true. Interestingly the minimum withdrawal amount started out always favoring CAPE, but after 1960 that trend reversed and it has favored ABW since. I am not sure if this is even a real trend, but I can't find an underlying cause here. Next maximum withdrawals always favored ABW, not surprising since a feature is that this method always uses all funds, while that never occurred for the cape rule. Next the ABW method was significantly more variable, with the standard deviations usually much higher than the cape rule, and the only time it was lower was 1930, when the SDs were $0.50 apart, that is essentially the same. What was really surprising to me was that even though there was a lot of variability, and CAPE left money on the table, the average withdraws prior to 1980 were really very similar. I will also note, that the later, and in particular the 2000 and 2010 max and SD need to be taken with a grain of salt, they are largely the product of steady return with a lower steady expected return for multiple decades. Account balances and withdrawals ballooned as a result.

And here is a summary table (remember these are monthly withdrawal, the 4% rule comparison is $3,333.33):

Image

Some of those lower bound results are a bit shocking. 5 of the 12 start dates there have a minimum withdraw at least 1/3rd lower than the 4% rule (I'm not advocating its use), and a couple more are pretty close. 3 of the 12 have a minimum less than 1/2 of the starting withdraw. I think that requires some action beyond just "I'll be flexible".

Here is my main point, its pretty easy to pull down the calculators on the wiki and have it tell you what you can spend. If that's all you do you might be in for a wild ride. Its much more work to put together a full plan using value of money equations, tweaking the starting assumptions (age of death, FV, g, etc), and testing until you are comfortable with the potential spending that results. Unfortunately the results above indicate to me that you need to do exactly that if you plan on using ABW. One more visual for emphasis, this is the 1900 retiree, orange is ABW, Blue is cape rule:

Image
User avatar
vanbogle59
Posts: 1314
Joined: Wed Mar 10, 2021 7:30 pm

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by vanbogle59 »

redmaw wrote: Wed Jan 26, 2022 2:26 pm it computes a withdrawal amount that is smaller than my necessary expenses. I consider this a failure even if its more mild than being broke.
A failure of what, exactly?
Are you saying you're portfolio has failed? Your budget? ABW?
User avatar
Ben Mathew
Posts: 2720
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

redmaw wrote: Wed Jan 26, 2022 2:26 pm I hear you, it can never run out of money before the end date, how can it be risky? Well easy, it computes a withdrawal amount that is smaller than my necessary expenses.
Yes, the risk with ABW is that spending will fall too low. So it is important to consider the odds of spending falling too low. If ABW says you can take out $40,000 in the first year or retirement, and that's just about enough for your retirement, then you can't retire yet. You have to build a large enough cushion so that spending even after adjusting is not likely to become unacceptably low. To evaluate whether you have enough of a cushion and are ready to retire, you can use the ABW monte carlo simulator available on the wiki:

Image

Also, living off a 75/25 portfolio will be risky and will require large spending adjustments. Take a look at the simulation results from TPAW (which uses ABW) for a 35/65 portfolio: Historical simulations 1881-2021. It still requires significant spending adjustments, but less:

Image

ABW adjusts to the current portfolio value, so spending will be variable. You can always find a better-looking strategy through backtesting. The risk is overfitting. The backtested strategy knows that it's coming out quickly enough from a particular historical crash. So it tells us that we don't have to adjust much. That looks smart in the historical data. But we can't do that in real life when we don't know if and when a recovery is coming.
Total Portfolio Allocation and Withdrawal (TPAW)
User avatar
Ben Mathew
Posts: 2720
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

redmaw wrote: Wed Jan 26, 2022 2:26 pm 2) "You will never run out of money" brushes the effects of sequence risk under the rug.
ABW combined with a fixed asset allocation does really well with sequence of return risk. Spending tracks the portfolio. A market crash reduces spending. But if and when the market recovers, spending also recovers in full -- percent for percent. There is no permanent damage to the portfolio from a temporary crash--only a temporary decline in spending during the crash. This is because it is fully adjusting to portfolio changes.
Total Portfolio Allocation and Withdrawal (TPAW)
SilverSmurfer
Posts: 163
Joined: Sun Jan 16, 2022 7:41 pm

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by SilverSmurfer »

I ponder: what percentage of Boggleheads who debate SWR, VPW, ABW, SORR and on:

have all of their necessary spending covered by SS/pension/similar; meaning their portfolio returns are purely discretionary and legacy funds

I'd wager that it is quite a high amount and that these are great mental exercises to debate on, yet very, very rarely get tested in the real world.
slicendice
Posts: 564
Joined: Tue Sep 22, 2020 12:08 am

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by slicendice »

SilverSmurfer wrote: Wed Jan 26, 2022 4:45 pm I ponder: what percentage of Boggleheads who debate SWR, VPW, ABW, SORR and on:

have all of their necessary spending covered by SS/pension/similar; meaning their portfolio returns are purely discretionary and legacy funds

I'd wager that it is quite a high amount and that these are great mental exercises to debate on, yet very, very rarely get tested in the real world.
In terms of spending on bare bones necessities (minimum accommodations, decent but not extravagant food & clothing) pensions/SS I would bet cover a significant chunk, if not all of this for most Americans (assuming at least 20-25 years of paying into these programs) let alone Bogleheads. For most Bogleheads I am asssuming, the portfolio withdrawals exist to maintain a lifestyle above the basics provided by SS, to supplement an annuity/pension whose payments may decline in real terms, to enable retirement before age 70 when the maximum SS benefit can be claimed. It is hard to generalize too much about Bogleheads financial situation. There are people on this site with 8 figure portfolios at 50, barely 6 figure portfolios at 65, & 4 figure portfolios at 25 and everything in between and beyond. I think it is one of the compelling features of the site.

I think ABW is probably the most elegant solution to the problem of how do you ensure your portfolio survives for a certain time, enabling one to make annual withdrawals without fear of prematurely running out of money, that explicitly into account changes in annual market performance as well as changes in future expected returns. The annual withdrawal number should be taken as a guideline of the maximum withdrawal allowed in a given year to satisfy the math. As has been pointed out above, this number should be in excess of what you actually need to pull from the portfolio.
User avatar
vanbogle59
Posts: 1314
Joined: Wed Mar 10, 2021 7:30 pm

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by vanbogle59 »

Ben Mathew wrote: Wed Jan 26, 2022 3:52 pm To evaluate whether you have enough of a cushion and are ready to retire, you can use the ABW monte carlo simulator available on the wiki:
Where? I can't find it. :confused
User avatar
vanbogle59
Posts: 1314
Joined: Wed Mar 10, 2021 7:30 pm

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by vanbogle59 »

vanbogle59 wrote: Thu Jan 27, 2022 1:39 pm
Ben Mathew wrote: Wed Jan 26, 2022 3:52 pm To evaluate whether you have enough of a cushion and are ready to retire, you can use the ABW monte carlo simulator available on the wiki:
Where? I can't find it. :confused
found it
https://www.bogleheads.org/wiki/Amortiz ... alculators
User avatar
Ben Mathew
Posts: 2720
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

vanbogle59 wrote: Thu Jan 27, 2022 1:56 pm
vanbogle59 wrote: Thu Jan 27, 2022 1:39 pm
Ben Mathew wrote: Wed Jan 26, 2022 3:52 pm To evaluate whether you have enough of a cushion and are ready to retire, you can use the ABW monte carlo simulator available on the wiki:
Where? I can't find it. :confused
found it
https://www.bogleheads.org/wiki/Amortiz ... alculators
That's the link to the "Advanced Calculators" section. The ABW monte carlo simulator is located just above in the "Basic Calculators" section:

https://www.bogleheads.org/wiki/Amortiz ... calculator
Total Portfolio Allocation and Withdrawal (TPAW)
User avatar
vanbogle59
Posts: 1314
Joined: Wed Mar 10, 2021 7:30 pm

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by vanbogle59 »

Ben Mathew wrote: Thu Jan 27, 2022 2:28 pm
vanbogle59 wrote: Thu Jan 27, 2022 1:56 pm
vanbogle59 wrote: Thu Jan 27, 2022 1:39 pm
Ben Mathew wrote: Wed Jan 26, 2022 3:52 pm To evaluate whether you have enough of a cushion and are ready to retire, you can use the ABW monte carlo simulator available on the wiki:
Where? I can't find it. :confused
found it
https://www.bogleheads.org/wiki/Amortiz ... alculators
That's the link to the "Advanced Calculators" section. The ABW monte carlo simulator is located just above in the "Basic Calculators" section:

https://www.bogleheads.org/wiki/Amortiz ... calculator
Yeah, I fat fingered that.
I see it. TY
Charon
Posts: 631
Joined: Thu May 03, 2018 12:08 pm

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by Charon »

On cFIREsim you can run with VPW and set a spending floor (with a couple knobs to turn to make it more flexible than just VPW). Then you can see how often it fails in back tests with various spending floors.

The method is best combined with some form of fixed income - Social Security, a bond bridge to Social Security, or a SPIA (as it says in the wiki).

I assume these points were made somewhere in the 781-comment thread you mentioned, but maybe not.
JackoC
Posts: 4714
Joined: Sun Aug 12, 2018 11:14 am

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by JackoC »

redmaw wrote: Wed Jan 26, 2022 2:26 pm I came to a few conclusions:

1) ABW is more of a tool to be incorporated into a larger plan than a withdrawal method. The plan can be as conservative or aggressive as you make it.
2) "You will never run out of money" brushes the effects of sequence risk under the rug. (For fun, go put 100% expected return into the wiki ABW calculator.)
3) ABW is not very helpful when trying to determine if you have enough to retire because of point 2.
4) Aggressive plans/poor sequences can lead to wild swings in allowed withdraws.
1. There's no one solution to the whole issue I agree. However I'd say ABW is more of a tactical withdrawal method than a basis of a plan before you retire.
2. Yeah there's no way a particular method of reaching a decision under uncertainty eliminates the uncertainty. Most people here are committed to still investing (sometimes heavily) in risk assets in retirement. But they are called risk assets for a reason.
3. I agree ABW is not a good way to determine if you have enough to retire but not so much because it can't eliminate the risk of very poor risk asset returns. Again no method of calculating can do that. The flaw I see in ABW to determine how much you need to *start* retirement is the tendency to use an expected type return as the 'r', when you should really be using a quite low %-tile 'r'. And IMO simulations using past returns are worse than useless now. They are based on a distribution of realized returns with mean much higher than the expected return now. Unless the variance from now is also much lower than the historical variance, a past '95%-tile worst case' of the past could be say an only 80%-tile worst case looking from now. Although you *can* choose a pre-retirement 'r' that's low enough. Everybody is unlikely to agree what it would be though.
4. I agree this is true just by the math, at least wild swings in a big down/up by the stock market like say spring 2020. But the basic dilemma of risk asset investing is you never know if a violent down will be followed by a bounce back. Many people here seem really focused on the recent past in one country leading to a pseudo-law that 'it always comes right back'. But stock markets can crash and not come back for decades. There's again no way to address that with a different way of calculating, only by having more low risk/low return assets (implying working longer and/or living on less in retirement than turns out to be necessary in most cases) or expanding your (assumed) ability to be satisfied with life after being forced into big spending reductions if things go badly (which goes even beyond 'depends on the person'; it's 'now you' estimating how 'future you' will feel).
User avatar
Ben Mathew
Posts: 2720
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

JackoC wrote: Fri Jan 28, 2022 8:56 am The flaw I see in ABW to determine how much you need to *start* retirement is the tendency to use an expected type return as the 'r', when you should really be using a quite low %-tile 'r'. And IMO simulations using past returns are worse than useless now. They are based on a distribution of realized returns with mean much higher than the expected return now. Unless the variance from now is also much lower than the historical variance, a past '95%-tile worst case' of the past could be say an only 80%-tile worst case looking from now. Although you *can* choose a pre-retirement 'r' that's low enough. Everybody is unlikely to agree what it would be though.
You're right that arriving at a reasonable forward looking distribution of returns is challenging. People certainly won't agree on what the right distribution to use is.

But any reasonable decision about when to retire will depend--implicitly or explicitly--on that distribution. I don't see this as a special problem for ABW. There is just no way around this.
Total Portfolio Allocation and Withdrawal (TPAW)
milktoast
Posts: 626
Joined: Wed Jul 10, 2019 8:17 pm

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by milktoast »

redmaw wrote: Wed Jan 26, 2022 2:26 pm expected bond returns = 1% ( I can't support this, but I didn't have anything better readily available)
In my spreadsheet, I use the 10yr TIPS yield. And cap the r at 2%. Choosing an r near 1.25% gives you an initial withdrawal of 4% for a 30 year horizon.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by willthrill81 »

milktoast wrote: Fri Jan 28, 2022 11:16 am
redmaw wrote: Wed Jan 26, 2022 2:26 pm expected bond returns = 1% ( I can't support this, but I didn't have anything better readily available)
In my spreadsheet, I use the 10yr TIPS yield. And cap the r at 2%. Choosing an r near 1.25% gives you an initial withdrawal of 4% for a 30 year horizon.
Right, that's close to the real return, assuming no volatility, needed for the '4% rule' to work. But with ABW, both the portfolio balance and the r will change over time, resulting in the amount withdrawn also changing.
The Sensible Steward
JackoC
Posts: 4714
Joined: Sun Aug 12, 2018 11:14 am

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by JackoC »

Ben Mathew wrote: Fri Jan 28, 2022 10:29 am
JackoC wrote: Fri Jan 28, 2022 8:56 am The flaw I see in ABW to determine how much you need to *start* retirement is the tendency to use an expected type return as the 'r', when you should really be using a quite low %-tile 'r'. And IMO simulations using past returns are worse than useless now. They are based on a distribution of realized returns with mean much higher than the expected return now. Unless the variance from now is also much lower than the historical variance, a past '95%-tile worst case' of the past could be say an only 80%-tile worst case looking from now. Although you *can* choose a pre-retirement 'r' that's low enough. Everybody is unlikely to agree what it would be though.
You're right that arriving at a reasonable forward looking distribution of returns is challenging. People certainly won't agree on what the right distribution to use is.

But any reasonable decision about when to retire will depend--implicitly or explicitly--on that distribution. I don't see this as a special problem for ABW. There is just no way around this.
Fair point, and relates to mine elsewhere in that post that no calculation method can remove the underlying risk of relying on stocks (or other risk assets) to support a major part of your retirement. There's no way to know the realized outcome and while you might be able to estimate the forward looking mean, and guesstimate the forward looking variance of the distribution from which the realized result will be 'selected' I've learned it's seldom possible to convince somebody else with a much different idea that they're wrong. :happy And ABW does kind of leave that question more open; 'SWR from a study by authorities in the past' is a little more binary, one tends to either believe it's still applicable or it's not (I'd say not, basically, so again I grant that puts me in the position of estimating the forward distribution with no way around it besides lowering my reliance on risk assets).
Topic Author
redmaw
Posts: 355
Joined: Mon Apr 22, 2019 7:20 am

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by redmaw »

vanbogle59 wrote: Wed Jan 26, 2022 2:32 pm
redmaw wrote: Wed Jan 26, 2022 2:26 pm it computes a withdrawal amount that is smaller than my necessary expenses. I consider this a failure even if its more mild than being broke.
A failure of what, exactly?
Are you saying you're portfolio has failed? Your budget? ABW?
I would say its a failure of my plan. If I intend to follow a plan that uses this method to calculate withdrawals and end up in a place where I can't cover the basics, I'm left with poor options. I would probably need to make uncomfortable decisions about what "need" actually means. In the end I would conclude my planning was wrong and I need a new plan...hence calling it failed.
Ben Mathew wrote: Wed Jan 26, 2022 3:52 pm
redmaw wrote: Wed Jan 26, 2022 2:26 pm I hear you, it can never run out of money before the end date, how can it be risky? Well easy, it computes a withdrawal amount that is smaller than my necessary expenses.
Yes, the risk with ABW is that spending will fall too low. So it is important to consider the odds of spending falling too low. If ABW says you can take out $40,000 in the first year or retirement, and that's just about enough for your retirement, then you can't retire yet. You have to build a large enough cushion so that spending even after adjusting is not likely to become unacceptably low.
We are largely in agreement here. I just haven't gotten the sense that the magnitude of the buffer or the need to test an individuals specifics has been emphasized enough when I have read about ABW on here before.
Ben Mathew wrote: Wed Jan 26, 2022 3:52 pm Also, living off a 75/25 portfolio will be risky and will require large spending adjustments.
I knew this part would get a bit of criticism here, and maybe rightly so. That is why I included the bit about having to plan to support a long retirement period. A 35/65 portfolio would certainly be less volatile. The trade off from reducing sequence risk with a low stock allocation is introducing a risk that average returns drop to too low to sustain the longer term. Over the standard 30 year window with 25x savings, growth only has to cover 5 years of spending, at 50 years growth has to cover 25...it does change the risk picture a bit. This is one of the factors I would need to tweak and test a bit to long before I settled on retiring.
Ben Mathew wrote: Wed Jan 26, 2022 3:52 pm Take a look at the simulation results from TPAW (which uses ABW) for a 35/65 portfolio: Historical simulations 1881-2021. It still requires significant spending adjustments, but less:
I've added that thread to me reading list, at this point I can't really understand the significance of the various columns there, or how the simulation was performed. I gather this was from the split of thread to discuss your specific flavor of ABW after you and siamond couldn't agree on discount rates.
Ben Mathew wrote: Wed Jan 26, 2022 4:04 pm
redmaw wrote: Wed Jan 26, 2022 2:26 pm 2) "You will never run out of money" brushes the effects of sequence risk under the rug.
ABW combined with a fixed asset allocation does really well with sequence of return risk. Spending tracks the portfolio. A market crash reduces spending. But if and when the market recovers, spending also recovers in full -- percent for percent. There is no permanent damage to the portfolio from a temporary crash--only a temporary decline in spending during the crash. This is because it is fully adjusting to portfolio changes.
Maybe you mean something different by "fixed asset allocation" The simulation I posted in the OP for 1900 retiree was 75/25 (technically this was rebalanced monthly based on how I calculated it. To me that is as "fixed" as an asset allocation can be...but the second half of the withdraw period bounced around between 75% and 50% of the original withdrawal amount, it never recovered. I guess I'm unclear on what point is being made here.
Topic Author
redmaw
Posts: 355
Joined: Mon Apr 22, 2019 7:20 am

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by redmaw »

SilverSmurfer wrote: Wed Jan 26, 2022 4:45 pm I ponder: what percentage of Boggleheads who debate SWR, VPW, ABW, SORR and on:

have all of their necessary spending covered by SS/pension/similar; meaning their portfolio returns are purely discretionary and legacy funds

I'd wager that it is quite a high amount and that these are great mental exercises to debate on, yet very, very rarely get tested in the real world.
I think this is a more important point than many realize, though probably not as you meant it. There is a very significant difference in the rigor required for planning if your needs are taken care of without touching your portfolio. i.e. If you will be fine regardless, winging it is fine. Personally I could be in that position if I worked to a more normal retirement age. I intend to retire early at a point when relying on pensions/SS is unwise, and therefore need to be fairly sure about how large a portfolio I need, and how I will manage it. This point translates directly into how acceptable variations in spending are. If its all fun and games, a 50% cut is fine...not so much if that means deciding between paying the electric bill or gas bill this month.

In the end though I am all but certain to do this and other analysis in a very conservative manner, and realize about 3 years into retirement I will never get close to the limits and stop tacking them.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by willthrill81 »

redmaw wrote: Sat Jan 29, 2022 12:41 pm
vanbogle59 wrote: Wed Jan 26, 2022 2:32 pm
redmaw wrote: Wed Jan 26, 2022 2:26 pm it computes a withdrawal amount that is smaller than my necessary expenses. I consider this a failure even if its more mild than being broke.
A failure of what, exactly?
Are you saying you're portfolio has failed? Your budget? ABW?
I would say its a failure of my plan. If I intend to follow a plan that uses this method to calculate withdrawals and end up in a place where I can't cover the basics, I'm left with poor options. I would probably need to make uncomfortable decisions about what "need" actually means. In the end I would conclude my planning was wrong and I need a new plan...hence calling it failed.
Unfortunately, no withdrawal plan can compensate for poor portfolio performance. If your portfolio does poorly and you were to use something like the '4% rule', your withdrawals are perfectly stable in inflation-adjusted dollars until your portfolio is depleted and the withdrawals cease entirely (nobody actually does that though). The risk of premature portfolio depletion due to excessive withdrawals can be eliminated with a method like ABW, but then the risk is that the withdrawals make shrink to the point that they are inadequate to cover your needs.

You can have your cake, or you can eat it, but you cannot do both.

That said, there are many angles to the portfolio income issue, and one's withdrawal strategy is only part of that. AA is another part of it. There are many portfolios that would have historically supported significantly higher potential withdrawal rates than the 3-fund portfolio.
The Sensible Steward
Topic Author
redmaw
Posts: 355
Joined: Mon Apr 22, 2019 7:20 am

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by redmaw »

milktoast wrote: Fri Jan 28, 2022 11:16 am
redmaw wrote: Wed Jan 26, 2022 2:26 pm expected bond returns = 1% ( I can't support this, but I didn't have anything better readily available)
In my spreadsheet, I use the 10yr TIPS yield. And cap the r at 2%. Choosing an r near 1.25% gives you an initial withdrawal of 4% for a 30 year horizon.
This value is mostly used in conjunction with historical periods where tips didn't exist is order to determine what the total portfolio expected return is. yield minus inflation would be fine then, but my date source only listed actual real returns that month, not yield. I'm sure this parameter could be improved, i'm not sure how much it would effect the outcomes, I suspect it would not change my general conclusions.
User avatar
Ben Mathew
Posts: 2720
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

redmaw wrote: Sat Jan 29, 2022 12:41 pm
Ben Mathew wrote: Wed Jan 26, 2022 4:04 pm
redmaw wrote: Wed Jan 26, 2022 2:26 pm 2) "You will never run out of money" brushes the effects of sequence risk under the rug.
ABW combined with a fixed asset allocation does really well with sequence of return risk. Spending tracks the portfolio. A market crash reduces spending. But if and when the market recovers, spending also recovers in full -- percent for percent. There is no permanent damage to the portfolio from a temporary crash--only a temporary decline in spending during the crash. This is because it is fully adjusting to portfolio changes.
Maybe you mean something different by "fixed asset allocation" The simulation I posted in the OP for 1900 retiree was 75/25 (technically this was rebalanced monthly based on how I calculated it. To me that is as "fixed" as an asset allocation can be...but the second half of the withdraw period bounced around between 75% and 50% of the original withdrawal amount, it never recovered. I guess I'm unclear on what point is being made here.
We mean the same thing by "fixed asset allocation."

See these posts for how ABW with a fixed asset allocation deals with sequence of return risk (SORR):

How retirement income responds to a poor sequence of returns (temporary crash)

Why retirement income recovers fully after a temporary crash.

ABW should have done really well with SORR in your simulations. For any age the following should be true:

- If the average portfolio return till that age equals the expected return, withdrawal will be equal to what had been scheduled at the start. The sequence of returns till that age does not matter.

- If average portfolio return till that age is x% more/less than expected return, withdrawal will also be x% more/less than starting withdrawal. The sequence of returns till that age does not matter.

So if the withdrawal did not recover in the second half of a simulation, that must be because the market did not recover relative to the expected return. In other words, it's a total return problem. Not a sequence of return problem.
Total Portfolio Allocation and Withdrawal (TPAW)
Topic Author
redmaw
Posts: 355
Joined: Mon Apr 22, 2019 7:20 am

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by redmaw »

Ben Mathew wrote: Sat Jan 29, 2022 4:38 pm
redmaw wrote: Sat Jan 29, 2022 12:41 pm
Ben Mathew wrote: Wed Jan 26, 2022 4:04 pm
redmaw wrote: Wed Jan 26, 2022 2:26 pm 2) "You will never run out of money" brushes the effects of sequence risk under the rug.
ABW combined with a fixed asset allocation does really well with sequence of return risk. Spending tracks the portfolio. A market crash reduces spending. But if and when the market recovers, spending also recovers in full -- percent for percent. There is no permanent damage to the portfolio from a temporary crash--only a temporary decline in spending during the crash. This is because it is fully adjusting to portfolio changes.
Maybe you mean something different by "fixed asset allocation" The simulation I posted in the OP for 1900 retiree was 75/25 (technically this was rebalanced monthly based on how I calculated it. To me that is as "fixed" as an asset allocation can be...but the second half of the withdraw period bounced around between 75% and 50% of the original withdrawal amount, it never recovered. I guess I'm unclear on what point is being made here.
We mean the same thing by "fixed asset allocation."

See these posts for how ABW with a fixed asset allocation deals with sequence of return risk (SORR):

How retirement income responds to a poor sequence of returns (temporary crash)

Why retirement income recovers fully after a temporary crash.

ABW should have done really well with SORR in your simulations. For any age the following should be true:

- If the average portfolio return till that age equals the expected return, withdrawal will be equal to what had been scheduled at the start. The sequence of returns till that age does not matter.

- If average portfolio return till that age is x% more/less than expected return, withdrawal will also be x% more/less than starting withdrawal. The sequence of returns till that age does not matter.

So if the withdrawal did not recover in the second half of a simulation, that must be because the market did not recover relative to the expected return. In other words, it's a total return problem. Not a sequence of return problem.
Again this is contradicted by the OP. I listed achieved cagr over the entire withdraw period. For 1900 retiree I refer to above, cagr was 4.51%, initial expected return was 4.27%. initial withdraw was just below 4000, final withdraw was 3000, and it was near there for 20 years. Returns exceeded expectations, but withdraws never recovered.

I also see no intuitive reason this has to be true. Unless possibly if you are referring to money weighted returns, but I don't see how that would be valid. The opposite seems intuitive to me, i.e. that sequence of returns still matters.

Edit: after a bit more thought I think I can get there if expected returns are held constant. The trade off then if even more volatile withdraws.
User avatar
Ben Mathew
Posts: 2720
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle

Re: So how risky is Amortization Based Withdrawal (ABW)

Post by Ben Mathew »

redmaw wrote: Sun Jan 30, 2022 10:01 am
Ben Mathew wrote: Sat Jan 29, 2022 4:38 pm
redmaw wrote: Sat Jan 29, 2022 12:41 pm
Ben Mathew wrote: Wed Jan 26, 2022 4:04 pm
redmaw wrote: Wed Jan 26, 2022 2:26 pm 2) "You will never run out of money" brushes the effects of sequence risk under the rug.
ABW combined with a fixed asset allocation does really well with sequence of return risk. Spending tracks the portfolio. A market crash reduces spending. But if and when the market recovers, spending also recovers in full -- percent for percent. There is no permanent damage to the portfolio from a temporary crash--only a temporary decline in spending during the crash. This is because it is fully adjusting to portfolio changes.
Maybe you mean something different by "fixed asset allocation" The simulation I posted in the OP for 1900 retiree was 75/25 (technically this was rebalanced monthly based on how I calculated it. To me that is as "fixed" as an asset allocation can be...but the second half of the withdraw period bounced around between 75% and 50% of the original withdrawal amount, it never recovered. I guess I'm unclear on what point is being made here.
We mean the same thing by "fixed asset allocation."

See these posts for how ABW with a fixed asset allocation deals with sequence of return risk (SORR):

How retirement income responds to a poor sequence of returns (temporary crash)

Why retirement income recovers fully after a temporary crash.

ABW should have done really well with SORR in your simulations. For any age the following should be true:

- If the average portfolio return till that age equals the expected return, withdrawal will be equal to what had been scheduled at the start. The sequence of returns till that age does not matter.

- If average portfolio return till that age is x% more/less than expected return, withdrawal will also be x% more/less than starting withdrawal. The sequence of returns till that age does not matter.

So if the withdrawal did not recover in the second half of a simulation, that must be because the market did not recover relative to the expected return. In other words, it's a total return problem. Not a sequence of return problem.
Again this is contradicted by the OP. I listed achieved cagr over the entire withdraw period. For 1900 retiree I refer to above, cagr was 4.51%, initial expected return was 4.27%. initial withdraw was just below 4000, final withdraw was 3000, and it was near there for 20 years. Returns exceeded expectations, but withdraws never recovered.

I also see no intuitive reason this has to be true. Unless possibly if you are referring to money weighted returns, but I don't see how that would be valid. The opposite seems intuitive to me, i.e. that sequence of returns still matters.

Edit: after a bit more thought I think I can get there if expected returns are held constant. The trade off then if even more volatile withdraws.
Here's the ABW simulator with four columns added at the end to show how sequence of returns plays out in ABW. The last two columns in red show that the withdrawal at any age matches the total cumulative excess return (i.e. excess over expected return) up to that point. So the withdrawal at any age is only a function of total return till that age. It does not depend on the order of returns till that age.

Image

Spreadsheet here.

This spreadsheet assumes g=0, but the result is valid for any g. We would just need to compare against the scheduled withdrawal rather than the starting withdrawal.

You are right that this result assumes no change in expected returns. What happens when expected returns change would depend on the expected return model being used and how it interacts with realized returns. If the expected return increases when markets do badly, then we are effectively moving funds from later years to earlier years (justified by the increase in expected return). Later on, if the market does well and expected return decreases, we are moving back the funds from early years to later years (necessitated by the decrease in expected return). But since we already consumed the extra funds for early years during this time, we won't be able to make the later years whole. So a poor return will not be fully offset by a good return in later years, leading to sequence of return dependence.
Total Portfolio Allocation and Withdrawal (TPAW)
Post Reply