Variable Percentage Withdrawal (VPW)

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

SnowBog wrote: Thu Aug 04, 2022 2:46 am
But it may be useful for those "concerned" about the "variability" of VPW, as apparently Rob Berger was.
The best way to show that is to see the impact of dampening on 1929 but I believe the backtest spreadsheet is no longer being maintained.

The very short forward test isn’t very compelling.

Using the Longinvest VPW spreadsheet may be simple but it’s hard to confirm the expected results independently.
dknightd
Posts: 3727
Joined: Wed Mar 07, 2018 10:57 am

Re: Variable Percentage Withdrawal (VPW)

Post by dknightd »

I think we can all agree there is no perfect withdrawal strategy! Since you can not predict the future.
I wonder if @longinvest follows the VPW withdrawal method. I suspect not, but I could be wrong! It is just another one of many useful tools. In my mind the VPW tables are just a more aggressive way of spending (or withdrawing your money from tax deferred) than the RMD requires. I'm more or less required to take RMD when I get to that age.

I'm a bucketeer. I have one bucket to see me through to SS. I have one bucket for buying SPIA (aka fixed pension). I have one bucket for potential long term care needs (that bucket could go to the kids if I do not need it - my gift to them is not to have them worry about my potential long term care expenses). The last bucket is hopefully discretionary. I plan to withdraw at least the RMD table, but hopefully not more than the VPW table.

edit: I'm not waiting till I'm 80 to buy an SPIA. I bought one when I retired, I'll buy another when I'm 65, then another when I claim SS at 70. Together SS and SPIA should cover our base expenses for as long as we live.
Retired 2019. So far, so good. I want to wake up every morning. But I want to die in my sleep. Just another conundrum. I think the solution might be afternoon naps ;)
User avatar
canadianbacon
Posts: 677
Joined: Sun Nov 10, 2019 9:04 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by canadianbacon »

Marseille07 wrote: Wed Aug 03, 2022 8:07 pm I wasn't aware the advice was N-20. You're correct that buying SPIAs at N-20 brings in a different set of pros and cons. I do not believe SPIAs are a panacea.
Unless something has changed since I learned about this, Will is right. The spreadsheet went to age 100, and the SPIA recommendation was at age 80. The exact SPIA age might change depending on actuarial tables and your gender. You want to wait until the rates make sense for you (balancing the rate offered and the remaining balance available to fund the annuity).
Bulls make money, bears make money, pigs get slaughtered.
HootingSloth
Posts: 1050
Joined: Mon Jan 28, 2019 2:38 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by HootingSloth »

willthrill81 wrote: Wed Aug 03, 2022 7:51 pm Retirees can certainly do a lot worse than VPW, which is itself only a single application of the amortization based withdrawal (ABW) method. VPW embeds historic stock and bond returns into the formula used to derive the annual withdrawal rates, something that not everyone is aware of.
Most variable methods that people talk about can be viewed as some special or limiting case of ABW. As you note, VPW is basically ABW with N=100 and a fixed value for the discount rate (a loosely historically-based "WAG"). Constant percentage withdrawal methods can also be seen as ABW with a fixed value for the discount rate in the limit where N goes to infinity.

Very roughly (and, yes, there is a lot more to it and longinvest has put a lot of thought into the details of the VPW method), you can think of VPW as ABW for people who don't want base their plan on changing estimates of expected returns (or use some of the other bells and whistles that can make ABW very flexible) and CPW as ABW for people who also think they (or their investments as held by their heirs) might live forever.
Global Market Portfolio + modest tilt towards volatility (80/20->60/40 as approach FI) + modest tilt away from exchange rate risk (80% global+20% U.S. stocks; currency-hedge bonds) + tax optimization
Marseille07
Posts: 16054
Joined: Fri Nov 06, 2020 12:41 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by Marseille07 »

SnowBog wrote: Thu Aug 04, 2022 2:46 am
Marseille07 wrote: Thu Aug 04, 2022 12:36 am We just have to see through what is being said and understand what the implications are.
The primary implications, as evidenced by the 3 years of the forward test, is the fluctuations of the withdrawals are dampened - as the name implies.

This may well be mental accounting, as mathematically the aggregate result of 6 months of averaged withdrawals versus 6 months of straight withdrawals should be fairly close (excluding slight difference in the first 6 months).

But it may be useful for those "concerned" about the "variability" of VPW, as apparently Rob Berger was.

And the dampening approach definitely isn't "required" for VPW. So those interested can use it, otherwise skip it.

In our case, we likely won't use a dampening approach (but TBD when we get there). For us, one of the main appeals of VPW is its simplicity. Towards that end, we'll likely make annual withdrawals and avoid the extra math required to maintain the dampening account, with the goal of keeping things as simple as possible. But then again, we aren't concerned [anymore] about the variability of VPW, we understand it to be a feature of the model.
Yeah, the dampening approach is needed *if* you're somehow spending near the limit year in and year out. Since VPW front-loads spending, most retirees would end up in a situation where VPW says they can spend 200K but they only spend 80K. In other words, smoothing out isn't necessary.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by willthrill81 »

HootingSloth wrote: Thu Aug 04, 2022 9:42 am
willthrill81 wrote: Wed Aug 03, 2022 7:51 pm Retirees can certainly do a lot worse than VPW, which is itself only a single application of the amortization based withdrawal (ABW) method. VPW embeds historic stock and bond returns into the formula used to derive the annual withdrawal rates, something that not everyone is aware of.
Most variable methods that people talk about can be viewed as some special or limiting case of ABW. As you note, VPW is basically ABW with N=100 and a fixed value for the discount rate (a loosely historically-based "WAG"). Constant percentage withdrawal methods can also be seen as ABW with a fixed value for the discount rate in the limit where N goes to infinity.

Very roughly (and, yes, there is a lot more to it and longinvest has put a lot of thought into the details of the VPW method), you can think of VPW as ABW for people who don't want base their plan on changing estimates of expected returns (or use some of the other bells and whistles that can make ABW very flexible) and CPW as ABW for people who also think they (or their investments as held by their heirs) might live forever.
Very well put. :thumbsup

I have long thought it interesting, to say the least, that longinvest has characterized most uses of dynamic discount rates rather than the static, historically derived discount rates used in VPW so negatively while implying that the latter is completely valid (e.g. "most VWR concepts rely on various metrics and guru prognostications to calibrate, annually, their withdrawal rate calculations"). First, this assumes that stock returns are completely unrelated to one another such that stocks would be just as likely to go drop in value by 30% immediately after a 50% decline as they were before the 50% decline, which almost nobody believes to be true. Second, this assumes that the bond portion of one's AA is going to return what it did in history, which is completely false as starting yields are far more logical and have been far more predictive.
The Sensible Steward
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by willthrill81 »

Marseille07 wrote: Thu Aug 04, 2022 9:51 am
SnowBog wrote: Thu Aug 04, 2022 2:46 am
Marseille07 wrote: Thu Aug 04, 2022 12:36 am We just have to see through what is being said and understand what the implications are.
The primary implications, as evidenced by the 3 years of the forward test, is the fluctuations of the withdrawals are dampened - as the name implies.

This may well be mental accounting, as mathematically the aggregate result of 6 months of averaged withdrawals versus 6 months of straight withdrawals should be fairly close (excluding slight difference in the first 6 months).

But it may be useful for those "concerned" about the "variability" of VPW, as apparently Rob Berger was.

And the dampening approach definitely isn't "required" for VPW. So those interested can use it, otherwise skip it.

In our case, we likely won't use a dampening approach (but TBD when we get there). For us, one of the main appeals of VPW is its simplicity. Towards that end, we'll likely make annual withdrawals and avoid the extra math required to maintain the dampening account, with the goal of keeping things as simple as possible. But then again, we aren't concerned [anymore] about the variability of VPW, we understand it to be a feature of the model.
Yeah, the dampening approach is needed *if* you're somehow spending near the limit year in and year out. Since VPW front-loads spending, most retirees would end up in a situation where VPW says they can spend 200K but they only spend 80K. In other words, smoothing out isn't necessary.
I'm not sure that most of those who use VPW will be withdrawing only 40% of what the method indicates that they can.
The Sensible Steward
Marseille07
Posts: 16054
Joined: Fri Nov 06, 2020 12:41 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by Marseille07 »

willthrill81 wrote: Thu Aug 04, 2022 9:54 am I'm not sure that most of those who use VPW will be withdrawing only 40% of what the method indicates that they can.
Yeah my example went a bit too far. A 2.2M portfolio & VPW saying 100K/year is probably realistic though. If you spend 80K/year, there isn't much to smooth out. I guess it's unclear why dampening is needed at all when your spending needs are met at all times.

If they argue that "50% flexibility" is necessary, shouldn't it mean exactly that - tough it out during a downturn?
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by willthrill81 »

Marseille07 wrote: Thu Aug 04, 2022 10:02 am
willthrill81 wrote: Thu Aug 04, 2022 9:54 am I'm not sure that most of those who use VPW will be withdrawing only 40% of what the method indicates that they can.
Yeah my example went a bit too far. A 2.2M portfolio & VPW saying 100K/year is probably realistic though. If you spend 80K/year, there isn't much to smooth out. I guess it's unclear why dampening is needed at all when your spending needs are met at all times.

If they argue that "50% flexibility" is necessary, shouldn't it mean exactly that - tough it out during a downturn?
If the withdrawal rate is so low that one can meet one's spending needs in all circumstances, then that's not much different than using a 'SWR method' with a sufficiently low WR.

The point of most variable withdrawal approaches is so that retirees can safely withdraw about as much as they can while minimizing the risk of premature portfolio depletion.
The Sensible Steward
Marseille07
Posts: 16054
Joined: Fri Nov 06, 2020 12:41 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by Marseille07 »

willthrill81 wrote: Thu Aug 04, 2022 10:05 am If the withdrawal rate is so low that one can meet one's spending needs in all circumstances, then that's not much different than using a 'SWR method' with a sufficiently low WR.

The point of most variable withdrawal approaches is so that retirees can safely withdraw about as much as they can while minimizing the risk of premature portfolio depletion.
I'm not sure about your situation but I've seen posts where the practitioners of VPW mention that it sets the limit so high (we're talking about 4.5% to start and goes higher) that they never go anywhere close.

Which makes sense because we're accustomed to accumulating at least 25x which is 4%, and VPW says you can spend 4.5% and goes higher.
HootingSloth
Posts: 1050
Joined: Mon Jan 28, 2019 2:38 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by HootingSloth »

willthrill81 wrote: Thu Aug 04, 2022 9:52 am
HootingSloth wrote: Thu Aug 04, 2022 9:42 am Most variable methods that people talk about can be viewed as some special or limiting case of ABW. As you note, VPW is basically ABW with N=100 and a fixed value for the discount rate (a loosely historically-based "WAG"). Constant percentage withdrawal methods can also be seen as ABW with a fixed value for the discount rate in the limit where N goes to infinity.

Very roughly (and, yes, there is a lot more to it and longinvest has put a lot of thought into the details of the VPW method), you can think of VPW as ABW for people who don't want base their plan on changing estimates of expected returns (or use some of the other bells and whistles that can make ABW very flexible) and CPW as ABW for people who also think they (or their investments as held by their heirs) might live forever.
Very well put. :thumbsup

I have long thought it interesting, to say the least, that longinvest has characterized most uses of dynamic discount rates rather than the static, historically derived discount rates used in VPW so negatively while implying that the latter is completely valid (e.g. "most VWR concepts rely on various metrics and guru prognostications to calibrate, annually, their withdrawal rate calculations"). First, this assumes that stock returns are completely unrelated to one another such that stocks would be just as likely to go drop in value by 30% immediately after a 50% decline as they were before the 50% decline, which almost nobody believes to be true. Second, this assumes that the bond portion of one's AA is going to return what it did in history, which is completely false as starting yields are far more logical and have been far more predictive.
For my part, I do see the fixed discount rate as a feature and not a bug. I am not personally inclined to try to use any of the existing methods for deriving changing estimates in the expected returns of stocks for three main reasons. First, I have doubts that existing methods for estimating future expected returns produce lower error than just always guessing a fixed estimate (which doubts have been discussed at length in other places). Second, I do not personally value greater "smoothing" of withdrawals from my risk portfolio because the necessary smoothing will be provided from other sources (namely, Social Security, a paid-off home, a COLA pension, and a TIPS ladder to bridge from retirement until age 70). Third, I value having a plan that has as few "tweakable" parameters as possible, and adding on a whole model that predicts future expected returns of stocks will add a lot of opportunity for tweaking. But, I don't really see doing this as an invalid approach, just not something that would make my plan a better fit for my beliefs, values, and circumstances.

The first and third reasons, above, do not actually translate to reasons to avoid using estimates of future expected returns of the fixed income portion of my risk portfolio. First, I do believe that current yields provide a better estimate, in the context of marketable fixed income assets, of future expected returns than a fixed historically-derived figure. Second, it does not seem like there would be much that would be "tweakable" about adding in current yields, since they are whatever they are. I might consider doing that, but I am not sure if the extra complexity would be worth it. I expect that my risk portfolio will likely be something like 90/10 or 80/20 (and the other sources outside the risk portfolio will likely be a significant chunk of the overall picture), so there may not be a lot of difference from adding extra precision only on the fixed income side of the risk portfolio.
Global Market Portfolio + modest tilt towards volatility (80/20->60/40 as approach FI) + modest tilt away from exchange rate risk (80% global+20% U.S. stocks; currency-hedge bonds) + tax optimization
User avatar
canadianbacon
Posts: 677
Joined: Sun Nov 10, 2019 9:04 pm

Re: Variable Percentage Withdrawal (VPW)

Post by canadianbacon »

The WAG is a conscious choice not to try to predict future returns; you have to put SOMETHING there or the calculation makes no sense. It really comes down to philosophy, do you want to try to predict or not? That can of worms has been opened in enough other threads.
Bulls make money, bears make money, pigs get slaughtered.
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

canadianbacon wrote: Thu Aug 04, 2022 11:05 am The WAG is a conscious choice not to try to predict future returns; you have to put SOMETHING there or the calculation makes no sense. It really comes down to philosophy, do you want to try to predict or not? That can of worms has been opened in enough other threads.
Well, if the WAG is based on backtesting or at least analysis it’s probably good enough if you update the WAG every so often.

As we can see with the International vs domestic wiki page another decade of data can change the graphs significantly…

Skimming the early pages again I don’t see the methodology from which the original VPW tables were developed. Maybe they are still good. Maybe they need to be re-examined.
Topic Author
longinvest
Posts: 5682
Joined: Sat Aug 11, 2012 8:44 am

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by longinvest »

willthrill81 wrote: Wed Aug 03, 2022 7:51 pm Retirees can certainly do a lot worse than VPW, which is itself only a single application of the amortization based withdrawal (ABW) method. VPW embeds historic stock and bond returns into the formula used to derive the annual withdrawal rates, something that not everyone is aware of.
(I added the emphasis).

Forum member Willthrill81 is mistaken.

For one thing, VPW was created (in July 2013) many years before the creation of the Amortization based withdrawal wiki page and ABW acronym (November 2020). VPW itself was initially inspired from Canadian Registered Retirement Income Fund (RRIF) minimum withdrawal rules (above age 70).

For another, the principles on which VPW is based are quite different from those of ABW as the later is presented in our Wiki. In particular, unlike ABW, VPW doesn't use an expected return in the hope of predicting future returns. Instead, VPW uses a growth trend, which is a wild ass guess (WAG) discount rate, to distinguish between low returns and high returns. While early in this thread, the "expected returns" term was used, that was before I learned that expected returns were just guesses by gurus, differing depending on which guru you were listening to! In other words, I've since realized that expected returns are pure garbage used to impress gullible clients of advisory firms and consumers of financial media. Obviously, the creators of ABW profoundly disagree with my assessment of expected returns (their whole withdrawal method is based on them). We don't have to agree; we're working on distinct portfolio withdrawal methods.

VPW doesn't use return predictions. Instead, VPW provides tools such as a backtesting spreadsheet (old tool) and a -50% stocks loss test within a retirement worksheet (newer and better tool) to concretely illustrate the uncertainty of future returns and help the retiree adequately prepare for variable withdrawals.

Forum member Siamond, one of the creators of ABW, does agree with me:
longinvest wrote: Sun Nov 15, 2020 5:08 pm I think that one problem is that ABW tries to be things it isn't. It isn't VPW, even though both take into account the time-value of money. And, it isn't RMD as RMD doesn't involve the time-value of money; it simply involves division by a moving "time left until life expectancy for the current age group".

ABW wants to model both income and future expenses; it gets involved into what VPW considers "budgeting" concerns. VPW, instead, is only concerned with income, much like an employer is concerned with paying a salary and doesn't get involved into the personal budget of employees. A salary isn't matched to increase when lump sum expenses happen. That's why people use low-interest debt to spread the cost of big items over time, or they save and buy later, when they can afford to wait.

ABW and VPW have differing philosophies and they don't use the same mathematical approach. If I'm right, ABW brings everything back into a present value calculation. The VPW accumulation and retirement approach, instead, uses various mathematical formulas, including time-value of money ones, to appropriately model each part of its simple overall plan while integrating pensions and annuities as fundamental parts of the plan to address longevity concerns, eliminating any attempt at guessing how long the retiree will live.

I suggest to stop pretending that ABW is some sort of generalization of VPW or RMD, or claim that VPW is a specific application of ABW; VPW is sufficiently different in its overall philosophy to be kept separate and fully distinct from ABW.

Just call it ABW; don't claim it's a sort of VPW. You'll be free to use ABW-specific terms, like "expected returns" to describe how it works, It will be much simpler and cleaner that way.
siamond wrote: Sun Nov 15, 2020 5:10 pm
longinvest wrote: Sun Nov 15, 2020 5:08 pmI suggest to stop pretending that ABW is some sort of generalization of VPW or RMD, or claim that VPW is a specific application of ABW; VPW is sufficiently different in its overall philosophy to be kept separate and fully distinct from ABW.

Just call it ABW; don't claim it's a sort of VPW. You'll be free to use ABW-specific terms, like "expected returns" to describe how it works, It will be much simpler and cleaner that way.
I wholeheartedly agree.
Variable Percentage Withdrawal (bogleheads.org/wiki/VPW) | One-Fund Portfolio (bogleheads.org/forum/viewtopic.php?t=287967)
saver7007
Posts: 83
Joined: Thu Jan 06, 2022 1:30 pm

Re: Variable Percentage Withdrawal (VPW)

Post by saver7007 »

Every model of future returns or withdrawal rates I've seen is really sensitive to the discount rate you choose for stocks and bonds, for obvious reasons I guess. You have to pick a value to do the math but going 1% up or down makes a really big difference in any conclusions you draw about how much to save or withdraw. And no value is provably correct. So maybe all portfolio and withdrawal rate forecasting is really just a WAG based on the garbage-in garbage-out principle.
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by vineviz »

longinvest wrote: Thu Aug 04, 2022 11:35 am For another, the principles on which VPW is based are quite different from those of ABW as the later is presented in our Wiki.
Nonetheless, it should be obvious to any observer who knows what "amortization" means that VPW is performing a particular flavor of amortization.

I don't see why anyone would be offended by that similarity, since the amortization feature is an important part of what makes VPW work in the way that it does.
longinvest wrote: Thu Aug 04, 2022 11:35 am VPW doesn't use return predictions. Instead, VPW provides tools such as a backtesting spreadsheet (old tool) and a -50% stocks loss test within a retirement worksheet (newer and better tool) to concretely illustrate the uncertainty of future returns and help the retiree adequately prepare for variable withdrawals.
The fact that the VPW spreadsheet uses a particularly naive implementation of expected returns doesn't mean that expected returns aren't part of the calculation.

Calling it a "guess" doesn't change the math.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

longinvest wrote: Thu Aug 04, 2022 11:35 am
VPW itself was initially inspired from Canadian Registered Retirement Income Fund (RRIF) minimum withdrawal rules (above age 70).
Is that where the tables are from?
VPW doesn't use return predictions. Instead, VPW provides tools such as a backtesting spreadsheet (old tool)
Are there any plans to support the backtesting tool?

Otherwise it’s hard to compare VPW results to anything else except through tools like FICalc that I believe you said we’re not accurate.
and a -50% stocks loss test within a retirement worksheet (newer and better tool) to concretely illustrate the uncertainty of future returns and help the retiree adequately prepare for variable withdrawals.
Given that the market can lose more than 50% I disagree that it is a better tool than backtesting.

Plus backtesting can show the impact of inflation that a 50% stock loss test does not.
SnowBog
Posts: 4699
Joined: Fri Dec 21, 2018 10:21 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by SnowBog »

For those that were interested/commenting on the "dampening" account to reduce the perception of variability of withdrawals, the latest post in the forward thread may be very interesting.
Marseille07 wrote: Thu Aug 04, 2022 12:36 am
dcabler wrote: Thu Aug 04, 2022 5:38 am
Granted, it's only been 3 years, and we haven't had a massive market correction (anything approaching 50%) that lasted for 6+ months... And by definition, VPW should vary as markets go up and down... But I remain impressed at how "stable" the retirement income has been over these past 3 years...

Here's what I think are the relevant excepts from the post, but highly recommend seeing the forward test thread for the full details:
longinvest wrote: Thu Aug 04, 2022 7:31 am ...All amounts are rounded to the nearest dollar and expressed in June 2022 dollars (using average CPI for inflation adjustments).
  • ...
  • Initial monthly total retirement income value: $7,078
    • Annualized initial income: ($7,078 X 12) = $84,885
    • ...
  • ...
  • Average monthly income over 37 months: ($265,298 / 37) = $7,170
    • Minimum monthly income: $6,910 (June 2020)
    • Maximum monthly income: $7,463 (September 2021)
  • Minimum cumulative 12-month income: $84,079 (from November 2019 to October 2020)
  • Maximum cumulative 12-month income: $88,756 (from April 2021 to March 2022)
  • This could be expressed as: $86,418 ± $2,339
In other words, annual income fluctuated by no more than ±2.7% around $86,418, so far. Meanwhile, the LifeStrategy Moderate Growth Fund investment fluctuated significantly more, losing (in inflation-adjusted terms) -13% in January-March 2020 and -21% in September-2021-June 2022. (Looking at 12-month periods: the LifeStrategy fund lost -20% from June 30, 2021 to June 30, 2022).

Two factors contributed to dampening annual income fluctuations (in addition to the 40% bond allocation of the LifeStrategy fund): the inclusion of current (work) and future (Social Security) pensions within the retirement plan, and the use of monthly withdrawals, more-or-less averaging portfolio values over 12 months with the help of a small withdrawal cushion which is a savings account containing approximately 5 months of withdrawals.
Also, longinvest points out that the 5-month dampening fund is essentially no different (in terms of AA or fund selection) than doing annual withdrawals (or at least that's how I read it), in that a retiree would spend approximately 1/12 of an annual withdrawal each month resulting in an average holding period of 5.5 months. I hadn't thought about it in this context...
longinvest wrote: Thu Aug 04, 2022 7:31 am ...
This is no more than the average* cash amount that would have been held by a retiree making annual VPW withdrawals (instead of monthly withdrawals).

* A retiree making annual withdrawals usually spends the money more-or-less equally over 12 months. So there would be 1/12 spent immediately, 1/12 held for one month, 1/12 held for 2 months, and so on, for an average holding period of 5.5 months.
Topic Author
longinvest
Posts: 5682
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest »

saver7007 wrote: Thu Aug 04, 2022 12:09 pm Every model of future returns or withdrawal rates I've seen is really sensitive to the discount rate you choose for stocks and bonds, for obvious reasons I guess. You have to pick a value to do the math but going 1% up or down makes a really big difference in any conclusions you draw about how much to save or withdraw. And no value is provably correct. So maybe all portfolio and withdrawal rate forecasting is really just a WAG based on the garbage-in garbage-out principle.
Saver7007, the VPW worksheet prominently displays an estimate of the impact of a -50% stock loss; this has a much bigger impact than a 1% up or down in discount rate.

As an example to illustrate how conservative VPW is, the latest calculations of the forward test, as of June 30, 2022, estimated the cost of Social Security bridge payments to $51,532 and discounted this amount from the after-loss portfolio:
longinvest wrote: Sat Jul 09, 2022 1:18 pm Image
...
The VPW worksheet also calculates a Required Flexibility that must be maintained by the retiree. To do so, it first applies a -50% loss to the stock allocation and then repeats its calculations. With 60/40 stocks/bonds allocation, this results into a (-50% X 60%) = -30% portfolio loss. That's a (-30% X $914,194) = -$274,258 portfolio loss, reducing the portfolio to ($914,194 - $274,258) = $639,936 after the loss. Then, it repeats its calculations:

Image
Effectively, this after-loss amount is equivalent to a pre-loss amount of ($51,532 / (100% - (60% / 2))) = $73,617.

According to my financial calculator, to generate $26,160 over two years, this represents an effective discount rate of:
  • n = 2, PV = -73617, FV = 0, PTM = 26160 ⇒ i = -44.87641444
In other words, the VPW worksheet uses a discount rate ranging from +3.76% down to -44.88% for the 2-year Social Security bridge. I'll leave the discount rate range calculations, for the other parts of the plan, as an exercise for interested readers. I think that an effective ±24% range, on the discount rate for 2 years of Social Security payments, is good enough.
Variable Percentage Withdrawal (bogleheads.org/wiki/VPW) | One-Fund Portfolio (bogleheads.org/forum/viewtopic.php?t=287967)
Marseille07
Posts: 16054
Joined: Fri Nov 06, 2020 12:41 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by Marseille07 »

SnowBog wrote: Thu Aug 04, 2022 12:34 pm For those that were interested/commenting on the "dampening" account to reduce the perception of variability of withdrawals, the latest post in the forward thread may be very interesting.
Marseille07 wrote: Thu Aug 04, 2022 12:36 am
dcabler wrote: Thu Aug 04, 2022 5:38 am
Granted, it's only been 3 years, and we haven't had a massive market correction (anything approaching 50%) that lasted for 6+ months... And by definition, VPW should vary as markets go up and down... But I remain impressed at how "stable" the retirement income has been over these past 3 years...

Here's what I think are the relevant excepts from the post, but highly recommend seeing the forward test thread for the full details:
longinvest wrote: Thu Aug 04, 2022 7:31 am ...All amounts are rounded to the nearest dollar and expressed in June 2022 dollars (using average CPI for inflation adjustments).
  • ...
  • Initial monthly total retirement income value: $7,078
    • Annualized initial income: ($7,078 X 12) = $84,885
    • ...
  • ...
  • Average monthly income over 37 months: ($265,298 / 37) = $7,170
    • Minimum monthly income: $6,910 (June 2020)
    • Maximum monthly income: $7,463 (September 2021)
  • Minimum cumulative 12-month income: $84,079 (from November 2019 to October 2020)
  • Maximum cumulative 12-month income: $88,756 (from April 2021 to March 2022)
  • This could be expressed as: $86,418 ± $2,339
In other words, annual income fluctuated by no more than ±2.7% around $86,418, so far. Meanwhile, the LifeStrategy Moderate Growth Fund investment fluctuated significantly more, losing (in inflation-adjusted terms) -13% in January-March 2020 and -21% in September-2021-June 2022. (Looking at 12-month periods: the LifeStrategy fund lost -20% from June 30, 2021 to June 30, 2022).

Two factors contributed to dampening annual income fluctuations (in addition to the 40% bond allocation of the LifeStrategy fund): the inclusion of current (work) and future (Social Security) pensions within the retirement plan, and the use of monthly withdrawals, more-or-less averaging portfolio values over 12 months with the help of a small withdrawal cushion which is a savings account containing approximately 5 months of withdrawals.
Also, longinvest points out that the 5-month dampening fund is essentially no different (in terms of AA or fund selection) than doing annual withdrawals (or at least that's how I read it), in that a retiree would spend approximately 1/12 of an annual withdrawal each month resulting in an average holding period of 5.5 months. I hadn't thought about it in this context...
longinvest wrote: Thu Aug 04, 2022 7:31 am ...
This is no more than the average* cash amount that would have been held by a retiree making annual VPW withdrawals (instead of monthly withdrawals).

* A retiree making annual withdrawals usually spends the money more-or-less equally over 12 months. So there would be 1/12 spent immediately, 1/12 held for one month, 1/12 held for 2 months, and so on, for an average holding period of 5.5 months.
Yeah, if it makes you (or other retirees) feel better operating VPW, I don't see anything wrong having a dampening account.

Practically, your withdraw-able amount (not to confuse with your *actual* expenses) fluctuating isn't a big issue unless you're constantly hitting your limit. And with proper planning & enough discretionary spending included in your estimate, you won't be hitting your limit, unless the markets return very poorly.
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

SnowBog wrote: Thu Aug 04, 2022 12:34 pm For those that were interested/commenting on the "dampening" account to reduce the perception of variability of withdrawals, the latest post in the forward thread may be very interesting.
If we haven’t seen a market correction that has exceeded the duration of the dampening period then I would say the lack of anything untoward happening is expected.

What you’d want to know is how well it would work in 1929, 2000, 2007, Nikkei 1990, etc…

Something a backtest tool would be able to tell you.

The forward test thread will be interesting in about 10-20 years or so…
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

longinvest wrote: Thu Aug 04, 2022 12:42 pm
Saver7007, the VPW worksheet prominently displays an estimate of the impact of a -50% stock loss; this has a much bigger impact than a 1% up or down in discount rate.

As an example to illustrate how conservative VPW is, the latest calculations of the forward test, as of June 30, 2022, estimated the cost of Social Security bridge payments to $51,532 and discounted this amount from the after-loss portfolio:
Given the 1929 cohort suffered a 86% loss I dunno that I would characterize 50% loss as “conservative”.

46% for 1972, 46% for 2000 and 54% for 2008 makes 50% kind of an average number for “bad bears” as opposed to “normal bears”.
HootingSloth
Posts: 1050
Joined: Mon Jan 28, 2019 2:38 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by HootingSloth »

Marseille07 wrote: Thu Aug 04, 2022 12:43 pm Yeah, if it makes you (or other retirees) feel better operating VPW, I don't see anything wrong having a dampening account.

Practically, your withdraw-able amount (not to confuse with your *actual* expenses) fluctuating isn't a big issue unless you're constantly hitting your limit. And with proper planning & enough discretionary spending included in your estimate, you won't be hitting your limit, unless the markets return very poorly.
Just to provide some contrast, and to illustrate that different people have quite different approaches, I do plan to more-or-less spend every dollar that VPW (perhaps with some modifications) says I should withdraw from my risk portfolio in each year. In fact, that is the fundamental reason that I plan on using a formulaic withdrawal method in the first place: to encourage me to make use of the money during my lifetime. If I were thinking about just using VPW as a "ceiling," I would not bother with it at all and just do everything ad hoc. Just my approach, and others of course can differ for many reasons, but you should not assume that everyone is using these methods just to come up with some kind of ceiling.
Global Market Portfolio + modest tilt towards volatility (80/20->60/40 as approach FI) + modest tilt away from exchange rate risk (80% global+20% U.S. stocks; currency-hedge bonds) + tax optimization
Marseille07
Posts: 16054
Joined: Fri Nov 06, 2020 12:41 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by Marseille07 »

HootingSloth wrote: Thu Aug 04, 2022 1:16 pm Just to provide some contrast, and to illustrate that different people have quite different approaches, I do plan to more-or-less spend every dollar that VPW (perhaps with some modifications) says I should withdraw from my risk portfolio in each year. In fact, that is the fundamental reason that I plan on using a formulaic withdrawal method in the first place: to encourage me to make use of the money during my lifetime. If I were thinking about just using VPW as a "ceiling," I would not bother with it at all and just do everything ad hoc. Just my approach, and others of course can differ for many reasons, but you should not assume that everyone is using these methods just to come up with some kind of ceiling.
How can you spend every dollar a method says you can spend? Some of my budget includes home repairs & buying a new car, but obviously these events do not happen every month. If I were hitting my ceiling month in and month out, something is wrong.

Now, I could withdraw in full and stash away the underspent $ elsewhere as a "rainy day fund." But in this case, the rainy day fund likely grows so much that I don't need a dampening account on top of that.
HootingSloth
Posts: 1050
Joined: Mon Jan 28, 2019 2:38 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by HootingSloth »

Marseille07 wrote: Thu Aug 04, 2022 1:24 pm
HootingSloth wrote: Thu Aug 04, 2022 1:16 pm Just to provide some contrast, and to illustrate that different people have quite different approaches, I do plan to more-or-less spend every dollar that VPW (perhaps with some modifications) says I should withdraw from my risk portfolio in each year. In fact, that is the fundamental reason that I plan on using a formulaic withdrawal method in the first place: to encourage me to make use of the money during my lifetime. If I were thinking about just using VPW as a "ceiling," I would not bother with it at all and just do everything ad hoc. Just my approach, and others of course can differ for many reasons, but you should not assume that everyone is using these methods just to come up with some kind of ceiling.
How can you spend every dollar a method says you can spend? Some of my budget includes home repairs & buying a new car, but obviously these events do not happen every month. If I were hitting my ceiling month in and month out, something is wrong.

Now, I could withdraw in full and stash away the underspent $ elsewhere as a "rainy day fund." But in this case, the rainy day fund likely grows so much that I don't need a dampening account on top of that.
If things go reasonably well, it is simple: I will add sufficient additional discretionary expenses, including, say, extra or more luxurious travel and additional charitable and family giving, to do what it takes to get to the number I am supposed to spend in any given year. To take the simplest possible example, if VPW tells me on January 1 that I can spend $x+y from my risk portfolio for the year, and I have only spent $x from my risk portfolio through December 30, then I can cut a check for $y on December 31 to a charity of my choice. If things go poorly, then I will have to cut back on spending in much the same way that anyone with a variable method will have to cut back.
Global Market Portfolio + modest tilt towards volatility (80/20->60/40 as approach FI) + modest tilt away from exchange rate risk (80% global+20% U.S. stocks; currency-hedge bonds) + tax optimization
Topic Author
longinvest
Posts: 5682
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest »

Monthly, in the forward test thread, I write something like this:
longinvest wrote: Sat Jul 09, 2022 1:18 pm ...
The retiree must maintain the flexibility to easily cut spending by up to -$1,195/month because stocks could easily lose -50% of their value within a short time period. In other words, at least $1,195 must be budgeted for optional discretionary spending that could be eliminated without affecting the retiree's comfort.
A -50% loss is a normal thing for stocks, even if it's infrequent. Such a loss shouldn't affect the retiree's comfort when the retiree has adequately planned (with the help of the worksheet) for variable portfolio withdrawals. This implies that required flexibility amounts, at the start of retirement, should be sufficient for the retiree to live in ample comfort.

The accumulation and retirement worksheet is a much better planning and stress testing tool than the backtesting spreadsheet. A simple yet very powerful stress test is to feed the initial "Portfolio Balance After Loss" (red cell) as "Portfolio Balance" (yellow input cell), and look at the resulting required flexibility assessment. This represents two consecutive -50% losses for stocks, or cumulative -75% losses (with rebalancing in between losses) at the worst of times, at the start of retirement immediately before first withdrawal. These are Great Depression types of losses without any quick recovery after losses. The retiree's comfort is probably affected, yet (hopefully) the retiree is likely to be doing better financially than many others in society if retirement was properly planned. The backtesting spreadsheet can't do such a simple and harsh test of the plan, as such deep losses without quick recovery never happened in the spreadsheet's data set (which includes U.S. stocks and bond returns for the last 150 years, since 1871).

Here's what I wrote in another thread in 2019. Note how I used the two consecutive -50% stock losses test to consider the retiree's various choices:
longinvest wrote: Wed Jul 24, 2019 10:04 am I like to take a broader view. I think that the magnitude of portfolio withdrawal fluctuations should be considered in context of the size of total retirement income including Social Security (possibly delayed), pension (if any), and (if necessary) inflation-indexed SPIA* relative to the retiree's financial needs for a comfortable retirement.

* Single Premium Immediate Annuity.

Note that the baseline, here, isn't a bare-bones retirement but a comfortable one!

Let's pick the example of a retiree who owns her home free and clear who could live comfortably on $80,000 (including taxes and expenses). By comfortable, I mean expenses that include travel and other fun things.

The retiree has just turned 65. She worked from age 25 to 64. Her salary started at $11,532 in 1979 ($40,687 in today's dollars) and gradually increased to $115,560 in 2018 ($117,878 in today's dollars). She has decided to delay Social Security until age 70 to receive $3,334/month (in today's dollars) and has accumulated a $1,500,000 portfolio through investing and inheritance. She has no work pension.

She is used to living below her means and could live on less than $80,000 (including taxes), if it comes to it, but it wouldn't be as comfortable.

With a 60/40 stocks/bonds portfolio, the VPW Accumulation And Retirement Worksheet tells us that she could withdraw $105,661 from the portfolio in 2019 (taking into account the future $40,008 annual income from Social Security). A -50% stocks loss would lower this to $83,176, still above her comfortable retirement needs.

With a 30/70 stocks/bonds portfolio, she could withdraw $97,864 from the portfolio in 2019 and a -50% stocks loss would lower this to $87,932.

The difference, in 2019, is ($105,661 - $97,864) = $7,797 (including taxes), which could allow for quite some additional fun while she is younger and still healthy. That's 8% more than with a 30/70 stocks/bonds portfolio.

The difference after a -50% stocks loss would be ($83,176 - $87,932) = -$4,756 (including taxes). That's -5% less than with a 30/70 stocks/bonds portfolio. As I wrote earlier, it's still within the comfortable retirement zone.

Let's also investigate the doomsday scenario of a -75% stocks loss (e.g. two 50% stocks losses). With a 60/40 stocks/bonds portfolio (reduced to $735,000) total retirement income would be $67,436. With a 30/70 stocks/bonds portfolio (reduced to $1,083,750) total retirement income would be $79,490. There's no surprise, here. The 30/70 stocks/bonds portfolio is more resilient to doomsday scenarios, yet, the 60/40 stocks/bonds portfolio still delivers sufficient retirement income, even though it's -15% below full comfort level. Remember that we're in a Great Depression (1930s) kind of situation, here. The rest of society is suffering.

All assets are risky. Stocks are significantly more volatile than bonds, but bonds have their own risks. A good argument could be made that the world bond stock portfolio is probably one of the most resilient portfolios in the long term. It currently sits at approximately 57/43 stocks/bonds according to the Bill Sharpe's preferred portfolio thread. A retiree could easily prefer this portfolio over one concentrated into the bonds asset class. I personally have a liking for Vanguard's LifeStrategy Moderate Growth Fund (VSMGX) which provides a good enough approximation of this stock-and-bond market portfolio with a moderate home bias.**

** A moderate home bias is justified for investors, across the world, by the additional frictions of international investing.

Instead of using a 30/70 stocks/bonds portfolio to reduce income fluctuations (if she wanted to reduce them), our retiree could use part of her portfolio to buy an inflation-indexed SPIA. With a 30/70 stocks/bonds allocation, her $1,500,000 portfolio would have $450,000 allocated to stocks. If she wanted to keep $450,000 in stocks with a 60/40 stocks/bonds allocation, she could retain a $750,000 portfolio and use the other $750,000 to buy an inflation-indexed SPIA which should pay her approximately $2,553/month. She would be reducing her liquidity by 50% but significantly increasing he guaranteed lifelong income. Her 2019 income would be $98,582 and a -50% stocks loss would reduce this to $87,340. The a doomsday -75% stocks loss would result into a $79,740 income; still comfortable.

One could easily argue that it's more efficient to annuitize and keep a market-like portfolio than adopting an allocation concentrated into bonds. Yet, our retiree could prefer to keep both the additional liquidity and higher income in 2019 with a 60/40 portfolio and wait until age 80 to annuitize part of her portfolio to dampen the financial risk of living beyond age 100.

"There is more than one road to Dublin." -- Taylor Larimore
Enjoy!
Variable Percentage Withdrawal (bogleheads.org/wiki/VPW) | One-Fund Portfolio (bogleheads.org/forum/viewtopic.php?t=287967)
Marseille07
Posts: 16054
Joined: Fri Nov 06, 2020 12:41 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by Marseille07 »

HootingSloth wrote: Thu Aug 04, 2022 1:30 pm If things go reasonably well, it is simple: I will add sufficient additional discretionary expenses, including, say, extra or more luxurious travel and additional charitable and family giving, to do what it takes to get to the number I am supposed to spend in any given year. To take the simplest possible example, if VPW tells me on January 1 that I can spend $x+y from my risk portfolio for the year, and I have only spent $x from my risk portfolio through December 30, then I can cut a check for $y on December 31 to a charity of my choice. If things go poorly, then I will have to cut back on spending in much the same way that anyone with a variable method will have to cut back.
Gotcha. If you are considering charity, we can always hit the limit every year. Although, in this case I'd imagine you don't really need a dampening account unless you somehow want to stabilize the amount you can donate to a charity of your choice.
saver7007
Posts: 83
Joined: Thu Jan 06, 2022 1:30 pm

Re: Variable Percentage Withdrawal (VPW)

Post by saver7007 »

longinvest wrote: Thu Aug 04, 2022 12:42 pm
In other words, the VPW worksheet uses a discount rate ranging from +3.76% down to -44.88% for the 2-year Social Security bridge. I'll leave the discount rate range calculations, for the other parts of the plan, as an exercise for interested readers. I think that an effective ±24% range, on the discount rate for 2 years of Social Security payments, is good enough.
Thank you for the response. I like how there is sort of a feedback loop built into the VPW system to handle the ups and downs of the stock market, I also like the simple usability of it for people who just want easy guidance on what to withdraw every month or year, it's a great and well thought out tool.

The discount rate I was really referring to is the 5% for stock and 1.9% for bonds that generates the VPW table values. As an experiment I copied the VPW Google sheet linked on the wiki page and lowered the discount rate for stocks to 4% see what the sensitivity would be to a 1% downward change, left all the other numbers the same, 60/40 $1,000,000 portfolio, age 65. On the retirement tab with that one tweak the VPW table value changed from 5.0% to 4.6%, and the monthly portfolio withdrawal amount went from $5356 to $5031, which is a $325/month or 6% difference.

So that was my only point about discount rates. Something has to be assumed as a discount rate for stocks and bonds in every model and small changes to those 2 variables cause relatively big changes in the model output, and nothing is provably correct to use for that value. Reasonable people could come up with different values.
HootingSloth
Posts: 1050
Joined: Mon Jan 28, 2019 2:38 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by HootingSloth »

Marseille07 wrote: Thu Aug 04, 2022 1:43 pm
HootingSloth wrote: Thu Aug 04, 2022 1:30 pm If things go reasonably well, it is simple: I will add sufficient additional discretionary expenses, including, say, extra or more luxurious travel and additional charitable and family giving, to do what it takes to get to the number I am supposed to spend in any given year. To take the simplest possible example, if VPW tells me on January 1 that I can spend $x+y from my risk portfolio for the year, and I have only spent $x from my risk portfolio through December 30, then I can cut a check for $y on December 31 to a charity of my choice. If things go poorly, then I will have to cut back on spending in much the same way that anyone with a variable method will have to cut back.
Gotcha. If you are considering charity, we can always hit the limit every year. Although, in this case I'd imagine you don't really need a dampening account unless you somehow want to stabilize the amount you can donate to a charity of your choice.
Right, I don't plan on use any dampening account for that reason, and it is also one of the reasons that VPW's withdrawal variability from year to year does not bother me.
Global Market Portfolio + modest tilt towards volatility (80/20->60/40 as approach FI) + modest tilt away from exchange rate risk (80% global+20% U.S. stocks; currency-hedge bonds) + tax optimization
Topic Author
longinvest
Posts: 5682
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest »

saver7007 wrote: Thu Aug 04, 2022 1:44 pm
longinvest wrote: Thu Aug 04, 2022 12:42 pm
In other words, the VPW worksheet uses a discount rate ranging from +3.76% down to -44.88% for the 2-year Social Security bridge. I'll leave the discount rate range calculations, for the other parts of the plan, as an exercise for interested readers. I think that an effective ±24% range, on the discount rate for 2 years of Social Security payments, is good enough.
Thank you for the response. I like how there is sort of a feedback loop built into the VPW system to handle the ups and downs of the stock market, I also like the simple usability of it for people who just want easy guidance on what to withdraw every month or year, it's a great and well thought out tool.

The discount rate I was really referring to is the 5% for stock and 1.9% for bonds that generates the VPW table values. As an experiment I copied the VPW Google sheet linked on the wiki page and lowered the discount rate for stocks to 4% see what the sensitivity would be to a 1% downward change, left all the other numbers the same, 60/40 $1,000,000 portfolio, age 65. On the retirement tab with that one tweak the VPW table value changed from 5.0% to 4.6%, and the monthly portfolio withdrawal amount went from $5356 to $5031, which is a $325/month or 6% difference.

So that was my only point about discount rates. Something has to be assumed as a discount rate for stocks and bonds in every model and small changes to those 2 variables cause relatively big changes in the model output, and nothing is provably correct to use for that value. Reasonable people could come up with different values.
Saver7007, the (upperbound) +3.76% discount rate, in my post, corresponds to the growth trend of a 60/40 stocks/bonds portfolio (used to construct the related column of the VPW table): ((60% X 5.0% stocks growth trend) + (40% X 1.9% bonds growth trend)) = 3.76% portfolio growth trend. What I explained, in my post, was that the -50% stock loss test effectively transform this into a a very wide range of possible discount rates with 3.76% on the high side.

We don't know what will happen in the next two years. Markets could soar or crash. The growth trend isn't a prediction of future returns; its role is mainly to distinguish between low and high returns, such that withdrawals generally increase when returns are higher than the trend, and decrease when lower than the trend. Along with the required flexibility calculation, it's how the worksheet accounts for the uncertainty and variability of market returns, and I think that it's good enough.

I've previously explained why using returns predictions to calibrate withdrawals could be quite harmful, even when using perfect predictions! If you missed it, I suggest reading this earlier post:
longinvest wrote: Sun Sep 20, 2020 7:33 am The use of a constant (unchanging) growth trend is fundamental to VPW. It avoids the "selling less high, selling more low" behavior.
... Continued here
Variable Percentage Withdrawal (bogleheads.org/wiki/VPW) | One-Fund Portfolio (bogleheads.org/forum/viewtopic.php?t=287967)
Lastrun
Posts: 1512
Joined: Wed May 03, 2017 6:46 pm

Re: Variable Percentage Withdrawal (VPW)

Post by Lastrun »

saver7007 wrote: Thu Aug 04, 2022 1:44 pm
…..
I like how there is sort of a feedback loop built into the VPW system to handle the ups and downs of the stock market…..

So that was my only point about discount rates. Something has to be assumed as a discount rate for stocks and bonds in every model and small changes to those 2 variables cause relatively big changes in the model output, and nothing is provably correct to use for that value. Reasonable people could come up with different values.
These are two good points. The first is that these variable methods (VPW or ABW) will self-correct if your return rate is over or under stated. I thought about this issue for a while, and then I heard Gene Fama on Rational Reminder and just let the issue go.

[Rational Reminder]: What do you think makes sense to use as an estimate for expected stock returns, just market returns?

[Fama]: “Okay. That's a very good question because I don't know what to use except for the historical average return. ……”
User avatar
siamond
Posts: 6008
Joined: Mon May 28, 2012 5:50 am

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by siamond »

longinvest wrote: Thu Aug 04, 2022 11:35 am[...] VPW doesn't use return predictions. Instead, VPW provides tools such as a backtesting spreadsheet (old tool) and a -50% stocks loss test within a retirement worksheet (newer and better tool) to concretely illustrate the uncertainty of future returns and help the retiree adequately prepare for variable withdrawals.

Forum member Siamond, one of the creators of ABW, does agree with me:
longinvest wrote: Sun Nov 15, 2020 5:08 pm [...]I suggest to stop pretending that ABW is some sort of generalization of VPW or RMD, or claim that VPW is a specific application of ABW; VPW is sufficiently different in its overall philosophy to be kept separate and fully distinct from ABW.

Just call it ABW; don't claim it's a sort of VPW. You'll be free to use ABW-specific terms, like "expected returns" to describe how it works, It will be much simpler and cleaner that way.
siamond wrote: Sun Nov 15, 2020 5:10 pm
longinvest wrote: Sun Nov 15, 2020 5:08 pmI suggest to stop pretending that ABW is some sort of generalization of VPW or RMD, or claim that VPW is a specific application of ABW; VPW is sufficiently different in its overall philosophy to be kept separate and fully distinct from ABW.

Just call it ABW; don't claim it's a sort of VPW. You'll be free to use ABW-specific terms, like "expected returns" to describe how it works, It will be much simpler and cleaner that way.
I wholeheartedly agree.
Hm. I only agreed with the sentence I quoted, agreeing that VPW and ABW were quite different in overall philosophies and agreeing both are free to use the terminology they prefer. That's all.

This being said, I always felt that calling the historical average a WAG is a dubious play on words. It is quite obviously a semi-educated guess about a number that has a decent chance of making the trajectory of future withdrawals acceptable for the retiree (certainly better than 0% or 10%!). The main difference with expected returns models suggested (but not mandated) in ABW is that it is a fixed/static guess as opposed to a more variable/adaptive guess, that's all. Either type of guess has a long list of pros and cons, none are ideal. Also, neither type of guess is a specific return prediction, which is definitely NOT required to make a PMT-based amortization scheme work in an acceptable manner. This is the beauty of PMT-based models, they self-adjust to actual (future) returns much better than say an SWR approach.

Quite frankly, I find those endless arguments about semantics quite tiring. VPW users like a fixed/static guess, fine. Most ABW users prefer a variable/adaptive guess, it's their prerogative. Why are we still arguing about semantics of "guru guess" vs. "wild a## guess" vs. "expected returns", I don't know... it's all the same 'rate' parameter injected in the underlying PMT formula. :shock: The only interesting thing is the practical impact on the trajectory of withdrawals of using one model or another.
Escapevelocity
Posts: 1145
Joined: Mon Feb 18, 2019 7:32 am

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by Escapevelocity »

HootingSloth wrote: Thu Aug 04, 2022 1:16 pm
Marseille07 wrote: Thu Aug 04, 2022 12:43 pm Yeah, if it makes you (or other retirees) feel better operating VPW, I don't see anything wrong having a dampening account.

Practically, your withdraw-able amount (not to confuse with your *actual* expenses) fluctuating isn't a big issue unless you're constantly hitting your limit. And with proper planning & enough discretionary spending included in your estimate, you won't be hitting your limit, unless the markets return very poorly.
Just to provide some contrast, and to illustrate that different people have quite different approaches, I do plan to more-or-less spend every dollar that VPW (perhaps with some modifications) says I should withdraw from my risk portfolio in each year. In fact, that is the fundamental reason that I plan on using a formulaic withdrawal method in the first place: to encourage me to make use of the money during my lifetime. If I were thinking about just using VPW as a "ceiling," I would not bother with it at all and just do everything ad hoc. Just my approach, and others of course can differ for many reasons, but you should not assume that everyone is using these methods just to come up with some kind of ceiling.
I was beginning to feel like the only one spending the full amount of the VWR suggested withdrawal amount. I have a conservative asset allocation of 50/50 (plan to increase to 60/40 eventually) and the 50% loss scenario only entails a 16% decrease to my withdrawal amount which is workable. Part of the reason for this is that approx. 1/3 of the overall portfolio is applied to the "Bridge" to social security at 67 and I also have a small pension. I'm currently building a TIPS/I-Bond ladder to make this bridge more or less riskless from an inflation standpoint.

There are some minor challenges month to month to ensure that the withdrawals are only reflecting the recurring monthly expenses and leaving out annual or semi-annual expenses like car insurance and property taxes. I choose not to use a buffering account for that stuff but rather just calibrate my annual budget to the VWR amount and then withdraw each month what I need to pay the bills as they come due knowing that my spend is aligned to the budget.

I would say that I'm not really prepared for the doomsday scenario of 75% drop in stock market, but I'm ok with rolling those dice.
Last edited by Escapevelocity on Thu Aug 04, 2022 3:19 pm, edited 1 time in total.
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by willthrill81 »

longinvest wrote: Thu Aug 04, 2022 11:35 am For one thing, VPW was created (in July 2013) many years before the creation of the Amortization based withdrawal wiki page and ABW acronym (November 2020). VPW itself was initially inspired from Canadian Registered Retirement Income Fund (RRIF) minimum withdrawal rules (above age 70).
ABW was put forth decades ago, in the 1950s, IIRC, though it wasn't called by that name. The ABW is simply an application of the time value of money formula, which obviously goes much further back than 2013, and similarly, the VPW is simply an application of the same formula with pre-determined parameters.
longinvest wrote: Thu Aug 04, 2022 11:35 am VPW doesn't use return predictions.
It has an embedded discount rate, meaning that it is using historic returns as a means of determining current withdrawals. That would be completely meaningless unless one expected the future to resemble the past. Whether you want to call that a 'prediction', 'expectation', or something else is irrelevant to me, but as Siamond has pointed out, it's only semantics.
The Sensible Steward
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

longinvest wrote: Thu Aug 04, 2022 1:36 pm The accumulation and retirement worksheet is a much better planning and stress testing tool than the backtesting spreadsheet. A simple yet very powerful stress test is to feed the initial "Portfolio Balance After Loss" (red cell) as "Portfolio Balance" (yellow input cell), and look at the resulting required flexibility assessment. This represents two consecutive -50% losses for stocks, or cumulative -75% losses (with rebalancing in between losses) at the worst of times, at the start of retirement immediately before first withdrawal.

These are Great Depression types of losses without any quick recovery after losses. The retiree's comfort is probably affected, yet (hopefully) the retiree is likely to be doing better financially than many others in society if retirement was properly planned.

The backtesting spreadsheet can't do such a simple and harsh test of the plan, as such deep losses without quick recovery never happened in the spreadsheet's data set (which includes U.S. stocks and bond returns for the last 150 years, since 1871).
This is an odd assertion to make given the 1929 crash was 86% and took 34 months peak to bottom and 186 months to full recovery.

15 years doesn’t strike me as a quick recovery…
I like to take a broader view. I think that the magnitude of portfolio withdrawal fluctuations should be considered in context of the size of total retirement income including Social Security (possibly delayed), pension (if any), and (if necessary) inflation-indexed SPIA* relative to the retiree's financial needs for a comfortable retirement.

* Single Premium Immediate Annuity.

Note that the baseline, here, isn't a bare-bones retirement but a comfortable one!

Let's pick the example of a retiree who owns her home free and clear who could live comfortably on $80,000 (including taxes and expenses). By comfortable, I mean expenses that include travel and other fun things.
$80,000 is her comfortable level. You need to stipulate her minimum spend in order to be able to assert her portfolio provided an adequate amount.

As a reminder a 4% WR (which is slightly above SWR) can only support a $60,000 annual spend for $1.5M.
The retiree has just turned 65. She worked from age 25 to 64. … She has decided to delay Social Security until age 70 to receive $3,334/month (in today's dollars) and has accumulated a $1,500,000 portfolio through investing and inheritance. She has no work pension.
If we add in $3.334/month at age 70…or 5 years after retirement we’re adding in $40,000 a year income.

Using a constant dollar approach this allows an annual draw of $84,000 a year over 35 years (age 100) with 100% historical success rate.

https://ficalc.app/?additionalIncome=%5 ... tantDollar
She is used to living below her means and could live on less than $80,000 (including taxes), if it comes to it, but it wouldn't be as comfortable.

With a 60/40 stocks/bonds portfolio, the VPW Accumulation And Retirement Worksheet tells us that she could withdraw $105,661 from the portfolio in 2019 (taking into account the future $40,008 annual income from Social Security). A -50% stocks loss would lower this to $83,176, still above her comfortable retirement needs.

Let's also investigate the doomsday scenario of a -75% stocks loss (e.g. two 50% stocks losses). With a 60/40 stocks/bonds portfolio (reduced to $735,000) total retirement income would be $67,436.

The 30/70 stocks/bonds portfolio is more resilient to doomsday scenarios, yet, the 60/40 stocks/bonds portfolio still delivers sufficient retirement income, even though it's -15% below full comfort level.
How do you know? You haven’t set a minimum spend for her…and the 75% value is below the 50% required flexibility test value.

In any case, using the same parameters for VPW shows that for the 1929 cohort the smallest spend value was $51,301.55 in 1932.

The worst case was $38,139.07…I went hunting for that but didn’t find it. I was briefly confused and then realized that must have been a year before SS kicked in.

This is far below your $67K result.

https://ficalc.app/?additionalIncome=%5 ... gyName=vpw

TL;DR:

1) the historical scenario was worse than 75% losses and full recovery wasn’t for 15 years.
2) the scenario, as presented, wasn’t very stressful given that the Constant Dollar value was above the VPW value and the retiree comfortable retirement needs of $80,000.
3) the worst case historical scenario reduced spending a lot more than the 75% did. Below $40K
Escapevelocity
Posts: 1145
Joined: Mon Feb 18, 2019 7:32 am

Re: Variable Percentage Withdrawal (VPW)

Post by Escapevelocity »

nigel_ht wrote: Thu Aug 04, 2022 3:28 pm
TL;DR:

1) the historical scenario was worse than 75% losses and full recovery wasn’t for 15 years.
2) the scenario, as presented, wasn’t very stressful given that the Constant Dollar value was above the VPW value and the retiree comfortable retirement needs of $80,000.
3) the worst case historical scenario reduced spending a lot more than the 75% did. Below $40K
Very clear, but how do these findings practically serve today's retirees looking for a workable model? Should the Great Depression serve as the stress test for all future models?
SnowBog
Posts: 4699
Joined: Fri Dec 21, 2018 10:21 pm

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by SnowBog »

nigel_ht wrote: Thu Aug 04, 2022 12:54 pm
SnowBog wrote: Thu Aug 04, 2022 12:34 pm For those that were interested/commenting on the "dampening" account to reduce the perception of variability of withdrawals, the latest post in the forward thread may be very interesting.
If we haven’t seen a market correction that has exceeded the duration of the dampening period then I would say the lack of anything untoward happening is expected.

What you’d want to know is how well it would work in 1929, 2000, 2007, Nikkei 1990, etc…

Something a backtest tool would be able to tell you.

The forward test thread will be interesting in about 10-20 years or so…
I think you miss the point... The "dampening" account is just a "shock absorber" for normal market fluctuations. Had the markets had a prolonged crash or a massive and prolonged gain, you'd see larger changes that the forward test reflects.

And the underlying point (at least as I see it) of VPW is that we make decisions on how to spend what money is available our entire working lives, and retirement is no different... Maybe this is more visible to me given that my working income is highly variable, so the idea of having more or less money in a given year in retirement (as VPW likely produces) isn't a concern to me.

But given your fixation on back testing, let's look at a few examples (excluding 2000 and 2007 as there isn't 30 years of data yet):

VPW (no min/max) https://ficalc.app?additionalIncome=%5B ... gyName=vpw

Code: Select all

Year   VPW%   VPW Avg Annual   VPW Min   VPW  Max   VPW Max Ending Portfolio   VPW Avg Ending Portfolio
----   ----   --------------   -------   --------   ------------------------   ------------------------
All    100%          $66,358   $21,386   $182,738                         $0                  
1929   100%          $51,251   $39,817    $66,338                         $0
1966   100%          $37,125   $21,386    $52,229                         $0
1990   100%          $79,864   $49,670    $99,413                         $0   


VPW with $20k min and $60k max (adjusted for inflation) - aka +/- 50% from what a 4% SWR would be: https://ficalc.app?additionalIncome=%5B ... gyName=vpw

Code: Select all

Year   VPW%   VPW Avg Annual   VPW Min   VPW  Max   VPW Max Ending Portfolio   VPW Avg Ending Portfolio
----   ----   --------------   -------   --------   ------------------------   ------------------------
All    100%          $53,676   $21,286    $60,000                 $3,411,041                   $601,164
1929   100%          $51,111   $39,817    $60,000                     $5,383
1966   100%          $37,125   $21,386    $52,229                         $0
1990   100%          $59,158   $49,670    $60,000                 $1,104,027      


SWR (4%) https://ficalc.app?additionalIncome=%5B ... tantDollar

Code: Select all

Year   SWR%   SWR Avg Annual   SWR Min   SWR  Max   SWR Max Ending Portfolio   SWR Avg Ending Portfolio
----   ----   --------------   -------   --------   ------------------------   ------------------------
All     95%          $39,909        $0    $40,000                 $4,961,134                 $1,382,889
1929    99%          $40,000   $39,999    $40,000                   $691,054
1966     0%          $38,007        $0    $40,000                         $0
1990   100%          $40,000   $40,000    $40,000                 $2,493,452     


SWR 3.33% https://ficalc.app?additionalIncome=%5B ... tantDollar

Code: Select all

Year   SWR%   SWR Avg Annual   SWR Min   SWR  Max   SWR Max Ending Portfolio   SWR Avg Ending Portfolio
----   ----   --------------   -------   --------   ------------------------   ------------------------
All    100%          $33,333   $33,333    $33,333                 $5,499,069                 $1,822,617
1929   100%          $33,333   $33,333    $33,333                 $1,148,882
1966   100%          $33,333   $33,333    $33,333                   $177,945
1990   100%          $33,333   $33,333    $33,333                 $3,000,997                        
As expected, VPW does what it intended to do, meaning it allows you to spend more of your money while alive without artificially creating more risk.
  • SWR failed in 1966, left nearly 70% of the initial portfolio behind in 1929, and > 100% of the initial portfolio in 1990 (as well as the average of all years).
  • Moving to a lower SWR (3.33%) resulted in no historic failures in this model, but increased the ending portfolio balance.
  • VPW never "failed", and never failed to provide at least $20k (50% of the 4% SWR model) in these scenarios (but it "could" have happened), on average provided 66% more spending money while alive (or 34% in our model constrained to $60k [1.5x SWR] cap).
Now with the examples, let's acknowledge what's wrong with them:
  • Neither account for things like when you die. Looking at Rich Broke or Dead with a forced $60k/year withdrawal (6% initial WR), for a 65 year old 30 year retirement, the most likely you are to be broke is at 85 (11.4%), 18% can of being rich (plan working) and 61.5% chance of being dead https://engaging-data.com/will-money-la ... 00&mort=ss In other words, instead of being worried about that 2% chance you might end up broke, realize its far more likely you'll be dead...
  • Which circles back to the whole "future being unpredictable" thing - we simply have no clue what the future holds... This is true for what the markets are going to do, what inflation will be, when we'll die, and if married when our surviving spouse dies...
  • "All models are wrong, some models are useful." There is no perfect model (or withdrawal approach), they all are based on assumptions and calculations which may or may not pan out 100% of the time (especially given that we can't predict the future).
  • Neither model includes any form of retirement income, most US investors should have at least some social security.
  • VPW "recommends" considering a SPIA around age 80 if you don't already have adequate life-long income.
  • FICalc makes assumptions in their VPW implementation.
    • Instead of using your "age" and following the table, they work backwards from 100 using your "length of retirement" input. A 45-year-old who wants to plan for a 35-year retirement is actually calculated as though they were a 65-year-old, meaning they'll be using a withdrawal rate > than the table recommends for their age. (Thus the FICalc is effectively forcing you to plan to age 100 to attempt to show VPW even remotely correctly.)
    • Instead of following the recommendation to cap withdrawal % at 10% (which applies usually in ages 88+), they do not cap withdrawal rates. This has two implications: 1) this forces portfolio completion at the end of the model - which may not be age 100 (see above) 2) it's impossible to use FICalc for estimates of you living past 100 (as in that case they'd assume you are younger than you are and use lower annual rates based on the wrong age).
  • My "approximation" of using a $60k max does not line up with the actual VPW 10% cap...
  • And ultimately, the core concept of VPW (at least in my view) goes back to having the "required flexibility." Again, I can't predict what the markets will do or when I'm going to die... But with VPW, my income will rise and fall based on the markets (or more accurately based on the size of my portfolio and my age), and I'll adapt accordingly. I'd much rather enjoy more of my hard-earned savings while alive then leave a pile untouched. Likewise, if we are able to, I'd much rather give to our heirs/charities while we are alive then to leave them a big pile of unspent cash they get when we die (which may be far less impactful on their lives at that point in time).
  • While not everyone will agree, I think the "human" factor comes in as well here.
    • Some that use VPW may well withdraw and spend (or gift) every last cent of every annual withdrawal away, and most likely they'll be just fine doing so... Others (like myself) will likely factor in the larger macro-economic situation. If we end up in another "great depression" style economy, I'm not sure I could bring myself to spend or gift 100% of what VPW says I could...
    • Conversely, if VPW says I should spend $50k, but something happens that I really need/want to spend $55k in a particular year (and overall the economic situation seems to be doing fine), I'll go ahead and do so. Obviously, overspending isn't recommended, has risks, and isn't sustainable...
    • But the mathematical beauty of the variable % of the portfolio is it becomes partially "self-healing". Conceptually, if we overspend in a particular year, that will result in less spending in future years, if we underspend, then increased spending in future years. Again, just like I'm doing during my working years...
Except edge cases (like the 1966 example for 4% SWR, where again it was more likely you'd be dead than broke), nearly all of these retirement methods are going to work out just fine the vast majority of the time. You are just making trade-offs in the assumptions you prefer and the risks you are trying to manage.

Personally, I favor "variable" models as I think they work "better" to accept that we don't know what we don't know and attempt to adapt accordingly, at least if you have the required flexibility to do so (meaning you've saved "enough" and can adapt your spending).

And back to a prior point I made "even if all else was equal", the simplicity of VPW and leaving that as the model for my surviving spouse who has no interest in finances is its primary draw. My spouse can understand a view of "spend this amount this year" model, and if they make a mistake and accidentally overspend or underspend (for any reason) - the model will adapt just fine. As hopefully by then all our "life-long" income streams will be in effect - and I know they won't starve or lose the house at that point. They'd primarily be dealing with "discretionary" expenses from age 80+.
Last edited by SnowBog on Thu Aug 04, 2022 5:18 pm, edited 1 time in total.
User avatar
Ben Mathew
Posts: 2720
Joined: Tue Mar 13, 2018 11:41 am
Location: Seattle

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by Ben Mathew »

longinvest wrote: Thu Aug 04, 2022 11:35 am
willthrill81 wrote: Wed Aug 03, 2022 7:51 pm Retirees can certainly do a lot worse than VPW, which is itself only a single application of the amortization based withdrawal (ABW) method. VPW embeds historic stock and bond returns into the formula used to derive the annual withdrawal rates, something that not everyone is aware of.
(I added the emphasis).

Forum member Willthrill81 is mistaken.

For one thing, VPW was created (in July 2013) many years before the creation of the Amortization based withdrawal wiki page and ABW acronym (November 2020). VPW itself was initially inspired from Canadian Registered Retirement Income Fund (RRIF) minimum withdrawal rules (above age 70).
While the term ABW was coined recently on Bogleheads to refer to withdrawal techniques based on amortizing the portfolio, the withdrawal techniques themselves go at least as far back as 1969. See:

The withdrawal strategy derived in Merton (1969) is ABW
longinvest wrote: Thu Aug 04, 2022 11:35 am
For another, the principles on which VPW is based are quite different from those of ABW as the later is presented in our Wiki. In particular, unlike ABW, VPW doesn't use an expected return in the hope of predicting future returns. Instead, VPW uses a growth trend, which is a wild ass guess (WAG) discount rate, to distinguish between low returns and high returns. While early in this thread, the "expected returns" term was used, that was before I learned that expected returns were just guesses by gurus, differing depending on which guru you were listening to! In other words, I've since realized that expected returns are pure garbage used to impress gullible clients of advisory firms and consumers of financial media. Obviously, the creators of ABW profoundly disagree with my assessment of expected returns (their whole withdrawal method is based on them). We don't have to agree; we're working on distinct portfolio withdrawal methods.

VPW doesn't use return predictions. Instead, VPW provides tools such as a backtesting spreadsheet (old tool) and a -50% stocks loss test within a retirement worksheet (newer and better tool) to concretely illustrate the uncertainty of future returns and help the retiree adequately prepare for variable withdrawals.
The core strategy in VPW is ABW with expected real return of 5% stocks and 1.9% bonds and spending growth of 0%. You have layered other things on top, so you are right that VPW is not just ABW. But to claim that VPW differs from ABW because it does not have expected return assumptions is wrong. Calling expected return a WAG or discount rate does not make it functionally different. If returns are lower than the WAG, withdrawals will fall, and if it's higher, withdrawals will increase. Same with expected returns in ABW. There is no meaningful difference between expected return and WAG and discount rate. It's just terminology.
Total Portfolio Allocation and Withdrawal (TPAW)
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

Escapevelocity wrote: Thu Aug 04, 2022 3:33 pm
nigel_ht wrote: Thu Aug 04, 2022 3:28 pm
TL;DR:

1) the historical scenario was worse than 75% losses and full recovery wasn’t for 15 years.
2) the scenario, as presented, wasn’t very stressful given that the Constant Dollar value was above the VPW value and the retiree comfortable retirement needs of $80,000.
3) the worst case historical scenario reduced spending a lot more than the 75% did. Below $40K
Very clear, but how do these findings practically serve today's retirees looking for a workable model? Should the Great Depression serve as the stress test for all future models?
It and 1966 should be a consideration for working models…and for now they serve well enough as a reasonable worst case scenario because…well they happened and much worse is going to be pretty bad :)

I just wouldn’t use less than that because globally the worst cases have been worse even ignoring things like war or communism.
Escapevelocity
Posts: 1145
Joined: Mon Feb 18, 2019 7:32 am

Re: Variable Percentage Withdrawal (VPW)

Post by Escapevelocity »

nigel_ht wrote: Thu Aug 04, 2022 4:36 pm
Escapevelocity wrote: Thu Aug 04, 2022 3:33 pm
nigel_ht wrote: Thu Aug 04, 2022 3:28 pm
TL;DR:

1) the historical scenario was worse than 75% losses and full recovery wasn’t for 15 years.
2) the scenario, as presented, wasn’t very stressful given that the Constant Dollar value was above the VPW value and the retiree comfortable retirement needs of $80,000.
3) the worst case historical scenario reduced spending a lot more than the 75% did. Below $40K
Very clear, but how do these findings practically serve today's retirees looking for a workable model? Should the Great Depression serve as the stress test for all future models?
It and 1966 should be a consideration for working models…and for now they serve well enough as a reasonable worst case scenario because…well they happened and much worse is going to be pretty bad :)

I just wouldn’t use less than that because globally the worst cases have been worse even ignoring things like war or communism.
I guess there is room for many opinions on such matters. I prefer to take a less cautious approach with regard to the portfolio size that I retired with. If things get to Great Depression levels, I believe that most people including myself would find ways to severely curtail consumption even beyond what is currently part of our "essential" level. For example, if we hit a global crisis, food selection would be significantly adapted (e.g., no restaurant meals).
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by willthrill81 »

Escapevelocity wrote: Thu Aug 04, 2022 4:41 pmIf things get to Great Depression levels, I believe that most people including myself would find ways to severely curtail consumption even beyond what is currently part of our "essential" level. For example, if we hit a global crisis, food selection would be significantly adapted (e.g., no restaurant meals).
I'm inclined to agree. The years that many here are willing to continue at work they don't truly love in order to increase the likelihood that they can globetrot in first-class fashion for decades don't resonate with me, especially if they don't start globetrotting until they're 70 and at high risk of experiencing significant health issues in no more than a decade.
The Sensible Steward
SnowBog
Posts: 4699
Joined: Fri Dec 21, 2018 10:21 pm

Re: Variable Percentage Withdrawal (VPW)

Post by SnowBog »

willthrill81 wrote: Thu Aug 04, 2022 4:58 pm
Escapevelocity wrote: Thu Aug 04, 2022 4:41 pmIf things get to Great Depression levels, I believe that most people including myself would find ways to severely curtail consumption even beyond what is currently part of our "essential" level. For example, if we hit a global crisis, food selection would be significantly adapted (e.g., no restaurant meals).
I'm inclined to agree. The years that many here are willing to continue at work they don't truly love in order to increase the likelihood that they can globetrot in first-class fashion for decades don't resonate with me, especially if they don't start globetrotting until they're 70 and at high risk of experiencing significant health issues in no more than a decade.
Which is far more likely for those concerned with finding something that "has never failed" before.

To try to mitigate the worst of the worst, they'll need to over save (work longer) than otherwise required. And yet when the "sh** hits the fan", they'll likely pull back on expenses (as everyone using a "variable" method is already used to doing). And then that whole not knowing how long we live for thing...

At some point, you have "enough" and your approach is "good enough"!
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

Escapevelocity wrote: Thu Aug 04, 2022 4:41 pm
nigel_ht wrote: Thu Aug 04, 2022 4:36 pm
Escapevelocity wrote: Thu Aug 04, 2022 3:33 pm
nigel_ht wrote: Thu Aug 04, 2022 3:28 pm
TL;DR:

1) the historical scenario was worse than 75% losses and full recovery wasn’t for 15 years.
2) the scenario, as presented, wasn’t very stressful given that the Constant Dollar value was above the VPW value and the retiree comfortable retirement needs of $80,000.
3) the worst case historical scenario reduced spending a lot more than the 75% did. Below $40K
Very clear, but how do these findings practically serve today's retirees looking for a workable model? Should the Great Depression serve as the stress test for all future models?
It and 1966 should be a consideration for working models…and for now they serve well enough as a reasonable worst case scenario because…well they happened and much worse is going to be pretty bad :)

I just wouldn’t use less than that because globally the worst cases have been worse even ignoring things like war or communism.
I guess there is room for many opinions on such matters. I prefer to take a less cautious approach with regard to the portfolio size that I retired with. If things get to Great Depression levels, I believe that most people including myself would find ways to severely curtail consumption even beyond what is currently part of our "essential" level. For example, if we hit a global crisis, food selection would be significantly adapted (e.g., no restaurant meals).
You’re more likely to mess up your retirement cost estimates than run into Great Depression levels.

The assumption you can “severely curtail consumption” is as certain as assuming you can “keep working in a downturn”.

But as you say, there is room for many opinions and being too conservative is as bad as too aggressive.
Topic Author
longinvest
Posts: 5682
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest »

To help readers seeing the harshness of feeding the initial "Portfolio Balance After Loss" (red cell) as "Portfolio Balance" (yellow input cell) in the retirement worksheet, here is a simulation of what this implies in the worksheet's static model (where returns are constant, except for initial losses).

Two hypothetical retirees of age 65 each have a $1,000,000 60/40 stocks/bonds portfolio from which they'll take withdrawal (in addition to receiving Social Security, which isn't shown in the simulation). Both die on their 88th birthday. The first retiree uses VPW to determine withdrawal amounts. The second retiree has full faith in the 4% SWR method and, consequently, withdraws $40,000 from the portfolio regardless of portfolio performance.

Just before the first withdrawal, stocks drop -75% (two -50% drops with rebalancing in between). Given the 60/40 allocation, this means two consecutive -30% portfolio losses, resulting into a ($1.000,000 X 70% X 70%) = $490,000 portfolio (before withdrawal). The VPW retiree adjusts the initial withdrawal accordingly. The SWR retiree, on the other hand, fully trusts the portfolio to recover and 4% SWR to work "as it did in the past" (in some markets)...

Then, after withdrawal, during the first year and all following years, the 60/40 portfolio grows by exactly 3.76%. (At this rate, it would take 20 years for an initial $10,000 lump sum (immediately dropping to $4,900) to get back again over $10,000.)

Numbers are rounded.

Image

The VPW retiree gets a little more than $24,000/year and leaves behind almost quarter of a million dollars. The SWR retiree gets $40,000/year for the first 15 years, a little less during the 16th year, and then gets $0 for the remaining 7 years before death at age 88, having depleted the portfolio at age 80.

This test is really harsh, much worse than a U.S. Great Depression back test. I think (I'm not sure; it's been many years since I've read about it) that it was a Japanese investor who faced portfolio depletion in as little as 16 years using 4% SWR.
Variable Percentage Withdrawal (bogleheads.org/wiki/VPW) | One-Fund Portfolio (bogleheads.org/forum/viewtopic.php?t=287967)
retiringwhen
Posts: 4743
Joined: Sat Jul 08, 2017 10:09 am
Location: New Jersey, USA

Re: Variable Percentage Withdrawal (VPW)

Post by retiringwhen »

nigel_ht wrote: Thu Aug 04, 2022 5:25 pm The assumption you can “severely curtail consumption” is as certain as assuming you can “keep working in a downturn”.
The US still has a basic social safety net for indigent seniors. Most of my family made it through their later years on only Social Security and Medicare. When they ended up in long-term care, their remaining (meager) assets were turned over to the state and they went on Medicaid.

Unlike Oliver Putnam in "Only Murders in the Building", most people figure out that keeping up pretenses is not worth it.

Like Bill Bernstein, I figure that true black swan (what he calls Deep Risk) can wipe me out. I'll figure it out and take prudent steps to avoid them as much as possible. Spending time looking into what is happening in the Ukraine helps focus the mind on those type of real black swans.
Topic Author
longinvest
Posts: 5682
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest »

Before anybody asks, here's what would have happened if the 4% SWR retiree of this post had decided, instead, to withdraw according the initial after-loss portfolio balance:

Image

Without any surprise, this SWR retiree who failed to trust in the 4% SWR got less than $20,000/year, that's -20% less than the VPW retiree, and died with more than $400,000, almost as much as the initial after-loss $490,000 portfolio balance.
Variable Percentage Withdrawal (bogleheads.org/wiki/VPW) | One-Fund Portfolio (bogleheads.org/forum/viewtopic.php?t=287967)
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

retiringwhen wrote: Thu Aug 04, 2022 5:46 pm
nigel_ht wrote: Thu Aug 04, 2022 5:25 pm The assumption you can “severely curtail consumption” is as certain as assuming you can “keep working in a downturn”.
The US still has a basic social safety net for indigent seniors. Most of my family made it through their later years on only Social Security and Medicare. When they ended up in long-term care, their remaining (meager) assets were turned over to the state and they went on Medicaid.

Unlike Oliver Putnam in "Only Murders in the Building", most people figure out that keeping up pretenses is not worth it.

Like Bill Bernstein, I figure that true black swan (what he calls Deep Risk) can wipe me out. I'll figure it out and take prudent steps to avoid them as much as possible. Spending time looking into what is happening in the Ukraine helps focus the mind on those type of real black swans.
Eh, I've seen that with some members of the family. Personally I don't think it matters with my health history but I prefer to have my wife in as nice a CCRC as I can afford rather than depending on the "basic social safety net" at that costs $$$.

My worst case scenario isn't 1929 or 1966 but my mom taking care of my dad who was sorta healthy for a long time but suffering from dementia.

Nope.
nigel_ht
Posts: 4742
Joined: Tue Jan 01, 2019 9:14 am

Re: [YouTube] Rob Berger's opinion of Variable Percentage Withdrawal (VPW)

Post by nigel_ht »

SnowBog wrote: Thu Aug 04, 2022 3:52 pm
But given your fixation on back testing, let's look at a few examples (excluding 2000 and 2007 as there isn't 30 years of data yet):

VPW (no min/max) https://ficalc.app?additionalIncome=%5B ... gyName=vpw

Code: Select all

Year   VPW%   VPW Avg Annual   VPW Min   VPW  Max   VPW Max Ending Portfolio   VPW Avg Ending Portfolio
----   ----   --------------   -------   --------   ------------------------   ------------------------
All    100%          $66,358   $21,386   $182,738                         $0                  
1929   100%          $51,251   $39,817    $66,338                         $0
1966   100%          $37,125   $21,386    $52,229                         $0
1990   100%          $79,864   $49,670    $99,413                         $0   
$21K is pretty rough...hopefully it wasn't for very long. With dampening it might not be so bad.

Unfortunately it's hard in FICalc to see this duration unless you download the spreadsheet.
SWR (4%)
Looks like using 50/50 and the same assumptions SWR in FICalc is 3.5%. So a SWR user who wanted 100% historical success would plan for a $35,000 ceiling.
As expected, VPW does what it intended to do, meaning it allows you to spend more of your money while alive without artificially creating more risk.
  • SWR failed in 1966, left nearly 70% of the initial portfolio behind in 1929, and > 100% of the initial portfolio in 1990 (as well as the average of all years).
A WR with a 5% failure rate which is 1966 and a couple surrounding years yes. That highlights that inflation is more deadly than bear markets.

But that's what a 5% failure rate means. If you want 100% success then the SWR amount is lower using the same assumptions as FICalc.
[*]VPW never "failed", and never failed to provide at least $20k (50% of the 4% SWR model) in these scenarios (but it "could" have happened), on average provided 66% more spending money while alive (or 34% in our model constrained to $60k [1.5x SWR] cap).
If you can live on $21K a year, awesome. On the other hand that's a 2.1% WR and you can pretty much do anything you want.

By definition 1/n strategies never fails.
[*] Neither account for things like when you die. Looking at Rich Broke or Dead with a forced $60k/year withdrawal (6% initial WR), for a 65 year old 30 year retirement, the most likely you are to be broke is at 85 (11.4%), 18% can of being rich (plan working) and 61.5% chance of being dead
Eh. Whether the risk of outliving your money is one that concerns you is a personal thing.

I am very unlikely to live to 95...but if I do, I prefer not to be broke...so I'll plan for 100 anyway because it seems there is an 8% chance either my wife or I will make it to 100.

11.5% chance of being 85 and depending on "basic social safety net for indigent seniors" is too risky for me. Might be okay for others.
[*] FICalc makes assumptions in their VPW implementation.
  • Instead of using your "age" and following the table, they work backwards from 100 using your "length of retirement" input. A 45-year-old who wants to plan for a 35-year retirement is actually calculated as though they were a 65-year-old, meaning they'll be using a withdrawal rate > than the table recommends for their age. (Thus the FICalc is effectively forcing you to plan to age 100 to attempt to show VPW even remotely correctly.)
  • Instead of following the recommendation to cap withdrawal % at 10% (which applies usually in ages 88+), they do not cap withdrawal rates. This has two implications: 1) this forces portfolio completion at the end of the model - which may not be age 100 (see above) 2) it's impossible to use FICalc for estimates of you living past 100 (as in that case they'd assume you are younger than you are and use lower annual rates based on the wrong age).
I would report this to the FICalc maintainer. And to Lauren since she maintains CFireSim which I think also has a VPW module.
[*] And ultimately, the core concept of VPW (at least in my view) goes back to having the "required flexibility." Again, I can't predict what the markets will do or when I'm going to die... But with VPW, my income will rise and fall based on the markets (or more accurately based on the size of my portfolio and my age), and I'll adapt accordingly. I'd much rather enjoy more of my hard-earned savings while alive then leave a pile untouched.
And that's great.
Likewise, if we are able to, I'd much rather give to our heirs/charities while we are alive then to leave them a big pile of unspent cash they get when we die (which may be far less impactful on their lives at that point in time).
That's up to you but the step up in basis is a huge benefit.
Except edge cases (like the 1966 example for 4% SWR, where again it was more likely you'd be dead than broke), nearly all of these retirement methods are going to work out just fine the vast majority of the time. You are just making trade-offs in the assumptions you prefer and the risks you are trying to manage.
Lol, in any year but 2022 considering 1966 as a remote edge case would likely be accepted without comment...
Wrench
Posts: 1055
Joined: Sun Apr 28, 2019 10:21 am

Re: Variable Percentage Withdrawal (VPW)

Post by Wrench »

nigel_ht wrote: Thu Aug 04, 2022 3:28 pm <SNIP>
TL;DR:

1) the historical scenario was worse than 75% losses and full recovery wasn’t for 15 years.
2) the scenario, as presented, wasn’t very stressful given that the Constant Dollar value was above the VPW value and the retiree comfortable retirement needs of $80,000.
3) the worst case historical scenario reduced spending a lot more than the 75% did. Below $40K
I am confused. When I run SIMBA's backtest from 1900 to 2021 with 60/40 asset allocation the largest drawdown was ~41% in 1932. with recovery in 4 years. Where is your 89% or 75% coming from? The other draw downs with this allocation in this ~120 year period are on the order of 20%. (As an aside, these results are consistent with my own personal experience in the last 25 - 35 years and two severe bear market in 2000-2002 and 2008-2009). A 50% loss is thus a very reasonable worst case scenario in my mind, covering the U.S. depression and more than covering multiple severe bear markets.

Wrench
User avatar
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by willthrill81 »

Wrench wrote: Thu Aug 04, 2022 6:37 pmWhen I run SIMBA's backtest from 1900 to 2021 with 60/40 asset allocation the largest drawdown was ~41% in 1932. with recovery in 4 years. Where is your 89% or 75% coming from?
100% stocks, not a 60/40.
The Sensible Steward
Post Reply