Variable Percentage Withdrawal (VPW)

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Tue Sep 06, 2016 10:27 pm

OK, one last set of answers.

Lieutenant.Columbo wrote: everything being equal the percentages of withdrawal will be the same regardless of whether the portfolio is worth $5M or $1M
correct?


Yes.

Lieutenant.Columbo wrote:please, remind me how VPW takes inflation into account, if it does; or maybe VPW is not concerned with inflation, if so disregard


If we wanted to get really technical, we could say that VPW's "internal rate" somehow takes inflation into account.

In practice, VPW calculations are not concerned with inflation. Withdrawals are adjusted to market returns, not to inflation. You want a portfolio which (hopefully) beats inflation. :wink:

Lieutenant.Columbo wrote:3. what will you adding the prior year's market results to the VPW spreadsheet do to an ongoing VPW plan?
the percentages of withdrawal won't change, will they?


Adding market results will simply increase the ranges for backtesting. Withdrawal percentages won't change except, sometimes, by a rounding error (0.1%) between versions of the spreadsheet. You might, for example, notice a very small difference (0.1% in some entries) between the Wiki's table and the spreadsheet. Pick one or the other, it doesn't matter.

In other words, once you've got one spreadsheet version, you can keep it and ignore new versions. Forum members Siamond doesn't even use the spreadsheet or Wiki table; he has simply encoded the VPW formula in his personal spreadsheet (using the Excel PMT function).
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

AlohaJoe
Posts: 2416
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Variable Percentage Withdrawal

Post by AlohaJoe » Tue Sep 06, 2016 10:30 pm

siamond wrote:
Lieutenant.Columbo wrote:3. recompute the PMT formula every year, capping the # of periods parameter to a minimum of 20
Do you mean that, No matter your age, each year you will recompute and set your spreadsheet assuming you at least have 20 yrs left, correct?

Yes. To do that, I compute the 'nper' parameter as min(20, 95+20-current-age). This is mostly to help with simulations, I just wanted to check if this would properly address an unexpected longevity in a smooth manner. By the time I'll get to 90 or so, I'm sure I'll revisit!


Lieutenant.Columbo, as you've discovered there are a lot of options for deciding "number of periods" with no real right or wrong answers. A number of different people have proposed PMT-based methods over the years -- Waring & Siegel; Pye; Steiner; and of course bogleheads' longinvest -- and different people make different starting assumptions.

You have super-conservative people like Waring & Siegel who say you should use age of 120 because, well, that's how long some people live.

Here's a chart that shows how various longevity assumptions in a PMT calculation might look like over time. (This is an idealised view because the focus is on the shape of the curves.)

I show a 40 year retirement span because that's how long it will take for 95% of Average Boglehead Couples™ to die. Around 5% will live longer than this.

Image

If you ignore the black line (which is Ken Steiner's suggestion) and the light blue line (which is "use VPW like a robot and run out of money at age 100") you'll notice that there's generally a pretty tight grouping of all the options.

For people who enjoy spreadsheets and tweaking in order to meet their personal preferences, there is room for tweaking. For people who aren't interested in that, you can pick almost any reasonable approach. Some people want a smoother step down in income as they age and are willing to do a bit more math. Some people want to front load a bit more income towards their healthy years. But in general we're not talking about massive differences; $8,000 a year for someone who starts with a $1,000,000 portfolio.

I wrote a bit more about PMT in a blog post a little while ago https://medium.com/@justusjp/flavors-of ... .p4323dntz with more charts similar to the above.

In the real world, your annual withdrawal is going to be dominated by market returns anyway.

User avatar
siamond
Posts: 3323
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal

Post by siamond » Wed Sep 07, 2016 8:36 am

AlohaJoe wrote:I wrote a bit more about PMT in a blog post a little while ago https://medium.com/@justusjp/flavors-of ... .p4323dntz with more charts similar to the above.

This is an excellent write-up, everybody interested in a VPW/PMT method should read it.

Two comments though:

1. what truly matters for a new retiree is not life expectancy as known today (in hindsight), but what it will be 20 or 30 years from now. This is where caution seems in order. It's like a trailing-PE vs a forward-PE, the latter should look better! :wink:

2. playing with the 'nper' parameter isn't all about longevity risk vs withdrawal trajectory. I listed several other goals of mine (bequest, LTC self-insurance, and also psychology-when-old) where I feel the need to protect the portfolio a little more and a little longer. Those may or may not apply to everybody, and one may draw different conclusions from those. There is more to this discussion than longevity, in other words.

User avatar
BlueEars
Posts: 3509
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: Variable Percentage Withdrawal (VPW)

Post by BlueEars » Wed Sep 07, 2016 4:29 pm

Some really good pointers in recent threads, thanks to all.

I've been using VPW for the last 2 years or so. Some of the recent comments got me to dig deeper into the underlying formulas. In particular the formulas on the "Table" sheet which has:
1) The default stock/bond historical data used to construct the backtest results. One can override this and at some point I might consider this. One should realize that changes in AA over retirement will, of course, change the allowed withdrawal rate. This can affect those backtest results which assume a constant AA.

2) The PMT calculation which I summarize here:
Image

It has been mentioned that one can run this stuff again if one gets way up there in years. But will I be running this stuff in my 80's? Hope so. :happy

User avatar
Lieutenant.Columbo
Posts: 1005
Joined: Sat Sep 05, 2015 9:20 pm
Location: 30th Parallel North

Re: Variable Percentage Withdrawal

Post by Lieutenant.Columbo » Wed Sep 07, 2016 8:54 pm

AlohaJoe,

Thank you very much for this post. it opens my views to other withdrawal thoughts and methods.

AlohaJoe wrote:...
You have super-conservative people like Waring & Siegel who say you should use age of 120 because, well, that's how long some people live.
...
If you ignore the black line (which is Ken Steiner's suggestion) and the light blue line (which is "use VPW like a robot and run out of money at age 100") you'll notice that there's generally a pretty tight grouping of all the options.
...
I wrote a bit more about PMT in a blog post a little while ago https://medium.com/@justusjp/flavors-of ... .p4323dntz with more charts similar to the above.

In the real world, your annual withdrawal is going to be dominated by market returns anyway.


I was only able to read Part 1 of your linked artice. Part 2 was way over my head.

I see you like Waring & Seigel's "average" approach the best, and I'd like to explore it further.

1. where you do You suggest I read more about the ins and outs of their "average" method, and possibly find a spreadsheet to make W&S simulations?

2. is W&S 'average" the closest method to the way you calculate with withdrawals?

thank you
Lt. Columbo: Well, what do you know. Here I am talking with some of the smartest people in the world, and I didn't even notice!

AlohaJoe
Posts: 2416
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Variable Percentage Withdrawal

Post by AlohaJoe » Wed Sep 07, 2016 9:35 pm

Lieutenant.Columbo wrote:

I'm all for people playing around and learning for themselves. But let me reiterate that most changes lead to relatively small differences. Unless you consider this a "hobby" or find it intellectually interesting, you're probably better off spending your free time on other pursuits in life :happy The difference between Ken Steiner's PMT assumptions and longinvest's PMT assumptions would be the difference between $29,704 and $27,102; or between $23,835 and $23,172.

(Ignore the yellow line and just look at the blue and green lines and how close they are over a 15 year period.)

Image

1. where you do You suggest I read more about the ins and outs of their "average" method, and possibly find a spreadsheet to make W&S simulations?


They published a paper a few years ago https://larrysiegeldotorg.files.wordpress.com/2014/09/siegel_waring_only-spending-rule-article-youll-ever-need.pdf. But their "average" approach is just to take the average of "years left until you're 120" and "your life expectancy according to a reasonable life table".

https://www.aacalc.com/calculators/le is the source of life expectancy I know of.

longinvest has put a lot of work into that VPW spreadsheet, making it a great resource for backtesting. If you're not up to making your own spreadsheet from scratch you can use VPW to test out other varieties by making modifications to it.

User avatar
FrugalInvestor
Posts: 4421
Joined: Fri Nov 07, 2008 12:20 am

Re: Variable Percentage Withdrawal (VPW)

Post by FrugalInvestor » Wed Sep 07, 2016 10:13 pm

Deleted...Read instructions and answered my own question....
IGNORE the noise! | Our life is frittered away by detail... simplify, simplify. - Henry David Thoreau

azanon
Posts: 1463
Joined: Mon Nov 07, 2011 10:34 am
Location: Little Rock, AR
Contact:

VPW and TSP

Post by azanon » Sat Sep 10, 2016 9:36 am

I was wondering if you guys thought VPW would be appropriate for a FERS retiree who has a Thrift Savings Plan. The TSP only allows for monthly withdrawals to be set up, and you can change the amount annually. But if I understand VPW correctly, it requires being able to make an annual withdrawal and then putting that cash in something safe to live on for that year (like a money market account).

When trying to think of a workaround (I haven't yet), I started to wonder if taking out the annual amount is actually a potential VPW flaw? Specifically, since that annual amount goes to cash, isn't that in effect causing portfolio "cash drag"? Especially in later years where the amount approaches and passes 10%, that's a pretty heavy cash drag. I guess its a bit like a 2-bucket retirement plan, but in a sense, you can never really consider monies separate from each other, meaning the sum total of what you have is a portfolio.

Critique aside, should I not use VPW if I intend to stay with the TSP, or should I just accept the volatility on my current year money and reset my annual percentage based on the new balance?

User avatar
siamond
Posts: 3323
Joined: Mon May 28, 2012 5:50 am

Re: VPW and TSP

Post by siamond » Sat Sep 10, 2016 10:23 am

azanon wrote:But if I understand VPW correctly, it requires being able to make an annual withdrawal and then putting that cash in something safe to live on for that year (like a money market account).

No, not really. This is just a suggestion longinvest made. Personally, I use a variant of VPW (details do not matter), I do the math at the beginning of the year, and this gives me a budget to aim at for the year. How exactly I withdraw is not consequential. Personally, I sell from my portfolio by chunks of $25k to replenish a savings account every now and then, while letting dividends trickle in my checking account. And I transfer money from my savings account to my checking account as needs be. Timing doesn't really matter, as long as my annual budget is roughly satisfied.

(I do agree with you that letting a significant amount of money sit in cash for months & months isn't terribly optimal, even though, in the grand scheme of things, this will not make the needle move very much)

User avatar
1210sda
Posts: 1208
Joined: Wed Feb 28, 2007 8:31 am

Re: VPW and TSP

Post by 1210sda » Sat Sep 10, 2016 12:30 pm

siamond wrote:(I do agree with you that letting a significant amount of money sit in cash for months & months isn't terribly optimal, even though, in the grand scheme of things, this will not make the needle move very much)

I agree, in the grand scheme of things, it doesn't move the needle much.....but it does make a difference in "simplicity" especially for DW later.

BTW, azanon said.."in the later years when the amount approaches and passes 10%, that's a pretty heavy cash drag".....I've been retired for 15 years and my "cash" is no where remotely near to 10% :happy
1210

azanon
Posts: 1463
Joined: Mon Nov 07, 2011 10:34 am
Location: Little Rock, AR
Contact:

Re: VPW and TSP

Post by azanon » Sat Sep 10, 2016 1:04 pm

1210sda wrote:
siamond wrote:(I do agree with you that letting a significant amount of money sit in cash for months & months isn't terribly optimal, even though, in the grand scheme of things, this will not make the needle move very much)

I agree, in the grand scheme of things, it doesn't move the needle much.....but it does make a difference in "simplicity" especially for DW later.

BTW, azanon said.."in the later years when the amount approaches and passes 10%, that's a pretty heavy cash drag".....I've been retired for 15 years and my "cash" is no where remotely near to 10% :happy
1210


In VPW, it's defined when you'll pass 10%. It's at age 88 with a 50/50 (see the VPW chart here https://www.bogleheads.org/wiki/Variabl ... withdrawal). Granted, it won't be 10% for the entire year because you're spending it.

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: VPW and TSP

Post by longinvest » Sat Sep 10, 2016 1:10 pm

azanon wrote:In VPW, it's defined when you'll pass 10%. It's at age 88 with a 50/50 (see the VPW chart here https://www.bogleheads.org/wiki/Variabl ... withdrawal). Granted, it won't be 10% for the entire year because you're spending it.


At age 88, my non-discretionary expenses will be covered by SS, pension, and (if necessary) inflation-indexed SPIA bought at 80. I don't know if I'll be able to spend all the money 10% of my remaining portfolio would provide for discretionary expenses. I'll be thankful just to be free of disabilities.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

azanon
Posts: 1463
Joined: Mon Nov 07, 2011 10:34 am
Location: Little Rock, AR
Contact:

Re: VPW and TSP

Post by azanon » Sat Sep 10, 2016 2:31 pm

longinvest wrote:
azanon wrote:In VPW, it's defined when you'll pass 10%. It's at age 88 with a 50/50 (see the VPW chart here https://www.bogleheads.org/wiki/Variabl ... withdrawal). Granted, it won't be 10% for the entire year because you're spending it.


At age 88, my non-discretionary expenses will be covered by SS, pension, and (if necessary) inflation-indexed SPIA bought at 80. I don't know if I'll be able to spend all the money 10% of my remaining portfolio would provide for discretionary expenses. I'll be thankful just to be free of disabilities.


As the author of VPW, what was your take on my situation? Do you think its a big deal if I just have monthly withdrawals taken from the portfolio itself, instead of a year's worth of cash set aside? It's actually part of the VPW method to have it set aside, so I just wondered what you thought of that deviation. I can at least reset the withdrawal rate in accordance with VPW annually.

I'll probably be able to spend mine. I'm aiming for a comfortable retirement, but certainly nothing lavish that I can't "take care of" even in my 80s.

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: VPW and TSP

Post by longinvest » Sat Sep 10, 2016 4:18 pm

azanon wrote:As the author of VPW, what was your take on my situation? Do you think its a big deal if I just have monthly withdrawals taken from the portfolio itself, instead of a year's worth of cash set aside? It's actually part of the VPW method to have it set aside, so I just wondered what you thought of that deviation. I can at least reset the withdrawal rate in accordance with VPW annually.


I always worry more about risk than potential (positive) returns. Usually, those who worry about losing a percent (asset return) of a percent (withdrawal) have a higher stock allocation. When I look at historical returns, I see that a high-stock portfolio could lose 30% in a few months. I wouldn't want to keep taking level withdrawals after such a drop. But, redoing the annual budget would not be nice, either. That's why I recommend a single annual withdrawal.

But, assuming one is flexible to change his budget every month (!!!), it is easy to simply derive a set of increasing monthly VPW percentages, instead of annual percentages, to take variable monthly withdrawals. But, I'll provide an approximation, for those who don't like to play with financial calculators and spreadsheets. A good enough approximation is simply to take the annual VPW percentage and divide it by twelve. In other words, to take monthly withdrawals, each month, I would withdraw:
    Monthly withdrawal = (annual VPW percentage / 12) X current portfolio balance at the time of withdrawal
    Note that I do not recommend making monthly withdrawals. I recommend taking annual withdrawals, instead.

Let me insist: monthly VPW withdrawals are a pain. If there are mandatory account withdrawals (like the TSP plan), I would simply let monthly withdrawals accumulate, along with dividends and other distributions, into a savings account, during the year, and count them as a part of next year's VPW withdrawal.

I always try to keep things as simple as they can be, specially when I consider the possibility that my wife would have to take things over if something happened to me.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

azanon
Posts: 1463
Joined: Mon Nov 07, 2011 10:34 am
Location: Little Rock, AR
Contact:

Re: VPW and TSP

Post by azanon » Sat Sep 10, 2016 5:07 pm

longinvest wrote:
azanon wrote:As the author of VPW, what was your take on my situation? Do you think its a big deal if I just have monthly withdrawals taken from the portfolio itself, instead of a year's worth of cash set aside? It's actually part of the VPW method to have it set aside, so I just wondered what you thought of that deviation. I can at least reset the withdrawal rate in accordance with VPW annually.


I always worry more about risk than potential (positive) returns. Usually, those who worry about losing a percent (asset return) of a percent (withdrawal) have a higher stock allocation. When I look at historical returns, I see that a high-stock portfolio could lose 30% in a few months. I wouldn't want to keep taking level withdrawals after such a drop. But, redoing the annual budget would not be nice, either. That's why I recommend a single annual withdrawal.


I might not have explained my situation clearly. My retirement plan just let's me set my "monthly withdrawal rate" once a year, but the monthly withdrawals have to come from the portfolio itself. So I guess if my VPW chart says "4%" (for instance) for a given year, I'll just set the monthly withdrawals for that year for 4%/12, and recalculate the next year.

Perhaps the way I can slightly correct for the extra risk of not having that year's money in cash, is to simply hold a bit fewer stocks than I planned otherwise (say 50% instead of 60%).

It's too bad their withdrawal options are so limited, but the federal G fund is so good (and their expense ratios are so low), I'd hate to rollover the money unless it was absolutely necessary.

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: VPW and TSP

Post by longinvest » Sat Sep 10, 2016 6:10 pm

azanon wrote:I might not have explained my situation clearly. My retirement plan just let's me set my "monthly withdrawal rate" once a year, but the monthly withdrawals have to come from the portfolio itself. So I guess if my VPW chart says "4%" (for instance) for a given year, I'll just set the monthly withdrawals for that year for 4%/12, and recalculate the next year.


OK, that would be good enough if you do it just for the TSP account (e.g. set the monthly withdrawal amount to: TSP balance X VPW percentage/12).

But, I wouldn't do that for the entire portfolio, though, as it introduces a small SWR-like risk into withdrawals. How much risk is added depends mostly on the AA of the TSP allocation. If it has lots of bonds, then the risk is really mild. If it has lots of stocks, it might lead to a bit more selling than one would really want in some scenarios. For example, the stock market could crash during the first month that followed the annual VPW calculation. Do you see the (small) problem?

As Siamond said, it shouldn't be too bad because withdrawal are quickly corrected within 12 months. But, still... After a 30% portfolio drop, it could lead to an effective 5.7% withdrawal instead of a 4% withdrawal.

As I wrote in my previous post, I personally would simply let the TSP withdrawals accumulate into a savings account along with any other portfolio income, then use this money as part of next year's annual VPW withdrawal. This would completely eliminate any SWR-like risk, while remaining simple.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
BlueEars
Posts: 3509
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: VPW and TSP

Post by BlueEars » Sat Sep 10, 2016 6:35 pm

azanon wrote:...
Perhaps the way I can slightly correct for the extra risk of not having that year's money in cash, is to simply hold a bit fewer stocks than I planned otherwise (say 50% instead of 60%).

It's too bad their withdrawal options are so limited, but the federal G fund is so good (and their expense ratios are so low), I'd hate to rollover the money unless it was absolutely necessary.

Let's say the worst happens and there is a 30% decline in your equities during January of a future year. Probably the bonds will go up so your portfolio takes a somewhat lesser hit. All the subsequent year's withdrawals are still going to take the hit. You are only talking about the coming year's withdrawals and protections for the spending during that year. If you don't have a pool of money in a savings account to take care of a perhaps 25% one year withdrawal hit, why not just develop that pool now? In other words, increase your withdrawals for 2017 to maybe 125% of normal and keep that extra money in emergency cash, CD, or a short term bond fund.

I do not think one has to reduce the AA by such a drastic amount to cover this kind of (unlikely but possible) event.

EDIT: I might be saying similar things as others above. My post was delayed a bit because of interruptions.

azanon
Posts: 1463
Joined: Mon Nov 07, 2011 10:34 am
Location: Little Rock, AR
Contact:

Re: VPW and TSP

Post by azanon » Sat Sep 10, 2016 10:21 pm

longinvest wrote:As I wrote in my previous post, I personally would simply let the TSP withdrawals accumulate into a savings account along with any other portfolio income, then use this money as part of next year's annual VPW withdrawal. This would completely eliminate any SWR-like risk, while remaining simple.


Ok, I understand now. Yeah, that's a potential workaround.

User avatar
Lieutenant.Columbo
Posts: 1005
Joined: Sat Sep 05, 2015 9:20 pm
Location: 30th Parallel North

Re: Variable Percentage Withdrawal

Post by Lieutenant.Columbo » Wed Sep 14, 2016 8:10 pm

longinvest wrote:Let me get philosophical, just a bit. If one intends to live forever, shouldn't one think of participating actively in society? What would be life if we had a large majority of 100 to 1000 years old people just sitting all day long, waiting for the minority of less-than 100 years old workers to feed and entertain them? Have you seen the movie "In Time"?


Just watched In Time (2011).
I thought plot was ok and premise very smart: Time is the only currency!

I get the point that if I want to live to 115, I should also be willing to work some here and there :oops:

Thank you for recommending this film!
Lt. Columbo: Well, what do you know. Here I am talking with some of the smartest people in the world, and I didn't even notice!

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sat Oct 08, 2016 4:09 pm

The VPW wiki page says:
3. Every few years, you should review your overall retirement plan. (If you are over 80 years old and your health is better than anticipated, for example, you might want to increase the Last Withdrawal Age beyond its initial 99 value).


In a private discussion, BahamaMan shared with me some anecdotal evidence. He observed a big decline in his older friends at around age 85. Here are his words: "Many friends of mine over the last 10 years that were very healthy at age 80, contracted an illness a few years later that led to a very rapid decline and they died around age 85-88 ....... I cannot stress enough how rapid the decline is from age 80 to age 85......."

He expressed the opinion that changing the Last Withdrawal Age at age 80 is not a good idea. For one thing, this could lead to an immediate and significant impact on withdrawals. For another, he thinks that this is premature, as many are still healthy at 80.

Instead, he proposes to slow the increase of VPW percentages after age 90, to avoid having to ever withdraw 100% of the portfolio. The simplest approach would have been to put a cap on percentages, but I personally don't like this because it leads to selecting an arbitrary number. Simply fixing the withdrawal percentage at the age 90 percentage leads to an abrupt change in withdrawals. So, after some back and forth, I proposed a simple formula that automatically adapts to the selected asset allocation (AA) and keeps the withdrawal decline relatively smooth.

Here are the percentages for plain VPW (pVPW) and what I'll temporary call smooth-tail VPW (sVPW) for comparison:

Code: Select all

Asset Allocation: 50% stocks / 50% bonds

Age   pVPW    sVPW
90   11.6%   11.6%
91   12.7%   12.4%
92   14.0%   13.3%
93   15.8%   14.1%
94   18.1%   14.9%
95   21.4%   15.7%
96   26.3%   16.4%
97   34.5%   17.1%
98   50.8%   17.8%
99   100.0%   18.5%
100   -   19.1%
101   -   19.8%
102   -   20.4%
103   -   20.9%
104   -   21.5%
105   -   22.0%
106   -   22.6%
107   -   23.1%
108   -   23.6%
109   -   24.0%
110   -   24.5%
111   -   24.9%
112   -   25.3%
113   -   25.8%
114   -   26.2%
115   -   26.5%
116   -   26.9%
117   -   27.3%
118   -   27.6%
119   -   28.0%


You can see how pVPW percentages keep increasing faster and faster to reach 100% at age 99, while sVPW percentages increase more slowly and remain at a level where a human is emotionally able to to make the withdrawal.

Actually, that was one of BahamaMan's points; he didn't think that he could follow through with an unmodified VPW after age 90, because percentages increased too drastically.

Here's how the withdrawal model looks like for the above 50/50 AA:
Image

In the model, withdrawals are fixed until age 90, then they start decreasing but smoothly.

Of course, this is not meant, in anyway, to replace an income floor established when approaching age 80. But, while keeping true to the principle of VPW, it takes care of not fully depleting the portfolio in case one survives beyond age 100.

For the those interested in the mathematical calculation, here's the idea. After age 90, the withdrawal percentage is calculated by assuming the the previous year's percentage was for an 11-year depletion, calculating the growth rate such that this would happen, then calculating the payment for a 10-year depletion using that rate. Here's the actual formula:

Code: Select all

= PMT(RATE(11;PreviousYearPercentage;-1;0;1;1);10;-1;0;1)


So, what are your thoughts about BahamaMan's suggestion? Should I eventually update the wiki and spreadsheet to add this smooth tail to VPW percentages? Should we eliminate the recommendation to update the Last Withdrawal Age?
Last edited by longinvest on Sat Oct 08, 2016 5:02 pm, edited 1 time in total.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
Lieutenant.Columbo
Posts: 1005
Joined: Sat Sep 05, 2015 9:20 pm
Location: 30th Parallel North

Re: Variable Percentage Withdrawal (VPW)

Post by Lieutenant.Columbo » Sat Oct 08, 2016 4:22 pm

longinvest wrote:The VPW wiki page says:
3. Every few years, you should review your overall retirement plan. (If you are over 80 years old and your health is better than anticipated, for example, you might want to increase the Last Withdrawal Age beyond its initial 99 value).


In a private discussion, Bahama man shared with me some anecdotal evidence. He observed a big decline in his older friends at around age 85. Here are his words: "Many friends of mine over the last 10 years that were very healthy at age 80, contracted an illness a few years later that led to a very rapid decline and they died around age 85-88 ....... I cannot stress enough how rapid the decline is from age 80 to age 85......."

He expressed the opinion that changing the Last Withdrawal Age at age 80 is not a good idea. For one thing, this could lead to an immediate and significant impact on withdrawals. For another, he thinks that this is premature, as many are still healthy at 80.

Instead, he proposes to slow the increase of VPW percentages after age 90, to avoid having to ever withdraw 100% of the portfolio. The simplest approach would have been to put a cap on percentages, but I personally don't like this because it leads to selecting an arbitrary number. Simply fixing the withdrawal percentage at the age 90 percentage leads to an abrupt change in withdrawals. So, after some back and forth, I proposed a simple formula that automatically adapts to the selected asset allocation (AA) and keeps the withdrawal decline relatively smooth.

Here are the percentages for plain VPW (pVPW) and what I'll temporary call smooth-tail VPW (sVPW) for comparison:

Code: Select all

Asset Allocation: 50% stocks / 50% bonds

Age   pVPW    sVPW
90   11.6%   11.6%
91   12.7%   12.4%
92   14.0%   13.3%
93   15.8%   14.1%
94   18.1%   14.9%
95   21.4%   15.7%
96   26.3%   16.4%
97   34.5%   17.1%
98   50.8%   17.8%
99   100.0%   18.5%
100   -   19.1%
101   -   19.8%
102   -   20.4%
103   -   20.9%
104   -   21.5%
105   -   22.0%
106   -   22.6%
107   -   23.1%
108   -   23.6%
109   -   24.0%
110   -   24.5%
111   -   24.9%
112   -   25.3%
113   -   25.8%
114   -   26.2%
115   -   26.5%
116   -   26.9%
117   -   27.3%
118   -   27.6%
119   -   28.0%


You can see how pVPW percentages keep increasing faster and faster to reach 100% at age 99, while sVPW percentages increase more slowly and remain at a level that a human is likely to be able to pull the trigger to make the withdrawal.

Actually, that was one of BahamaMan's points; he didn't think that he could pull the trigger on an unmodified VPW after age 90, because percentages increased too drastically.

Here's how the withdrawal model looks like for the above 50/50 AA:
Image

In the model, withdrawals are fixed until age 90, then they start decreasing but smoothly.

Of course, this is not meant, in anyway, to replace an income floor established when approaching age 80. But, while keeping true to the principle of VPW, it takes care of not fully depleting the portfolio in case one survives beyond age 100.

For the those interested in the mathematical calculation, here's the idea. After age 90, the withdrawal percentage is calculated by assuming the the previous year's percentage was for an 11-year depletion, calculating the growth rate such that this would happen, then calculating the payment for a 10-year depletion using that rate. Here's the actual formula:

Code: Select all

= PMT(RATE(11;PreviousYearPercentage;-1;0;1;1);10;-1;0;1)


So, what are your thoughts about BahamaMan's suggestion? Should I eventually update the wiki and spreadsheet to add this smooth tail to VPW percentages? Should we eliminate the recommendation to update the Last Withdrawal Age?
longinvest,
what do BahamaMan and you exactly mean by pull the trigger in this context?
thank you for this VPW modification (that I'm still trying to understand fully)
Last edited by Lieutenant.Columbo on Sat Oct 08, 2016 4:40 pm, edited 1 time in total.
Lt. Columbo: Well, what do you know. Here I am talking with some of the smartest people in the world, and I didn't even notice!

User avatar
Lieutenant.Columbo
Posts: 1005
Joined: Sat Sep 05, 2015 9:20 pm
Location: 30th Parallel North

Re: Variable Percentage Withdrawal (VPW)

Post by Lieutenant.Columbo » Sat Oct 08, 2016 5:44 pm

longinvest wrote:The VPW wiki page says:
3. Every few years, you should review your overall retirement plan. (If you are over 80 years old and your health is better than anticipated, for example, you might want to increase the Last Withdrawal Age beyond its initial 99 value).


In a private discussion, BahamaMan shared with me some anecdotal evidence. He observed a big decline in his older friends at around age 85. Here are his words: "Many friends of mine over the last 10 years that were very healthy at age 80, contracted an illness a few years later that led to a very rapid decline and they died around age 85-88 ....... I cannot stress enough how rapid the decline is from age 80 to age 85......."

He expressed the opinion that changing the Last Withdrawal Age at age 80 is not a good idea. For one thing, this could lead to an immediate and significant impact on withdrawals. For another, he thinks that this is premature, as many are still healthy at 80.

Instead, he proposes to slow the increase of VPW percentages after age 90, to avoid having to ever withdraw 100% of the portfolio. The simplest approach would have been to put a cap on percentages, but I personally don't like this because it leads to selecting an arbitrary number. Simply fixing the withdrawal percentage at the age 90 percentage leads to an abrupt change in withdrawals. So, after some back and forth, I proposed a simple formula that automatically adapts to the selected asset allocation (AA) and keeps the withdrawal decline relatively smooth.

Here are the percentages for plain VPW (pVPW) and what I'll temporary call smooth-tail VPW (sVPW) for comparison:

Code: Select all

Asset Allocation: 50% stocks / 50% bonds

Age   pVPW    sVPW
90   11.6%   11.6%
91   12.7%   12.4%
92   14.0%   13.3%
93   15.8%   14.1%
94   18.1%   14.9%
95   21.4%   15.7%
96   26.3%   16.4%
97   34.5%   17.1%
98   50.8%   17.8%
99   100.0%   18.5%
100   -   19.1%
101   -   19.8%
102   -   20.4%
103   -   20.9%
104   -   21.5%
105   -   22.0%
106   -   22.6%
107   -   23.1%
108   -   23.6%
109   -   24.0%
110   -   24.5%
111   -   24.9%
112   -   25.3%
113   -   25.8%
114   -   26.2%
115   -   26.5%
116   -   26.9%
117   -   27.3%
118   -   27.6%
119   -   28.0%


You can see how pVPW percentages keep increasing faster and faster to reach 100% at age 99, while sVPW percentages increase more slowly and remain at a level where a human is emotionally able to to make the withdrawal.

Actually, that was one of BahamaMan's points; he didn't think that he could follow through with an unmodified VPW after age 90, because percentages increased too drastically.

Here's how the withdrawal model looks like for the above 50/50 AA:
Image

In the model, withdrawals are fixed until age 90, then they start decreasing but smoothly.

Of course, this is not meant, in anyway, to replace an income floor established when approaching age 80. But, while keeping true to the principle of VPW, it takes care of not fully depleting the portfolio in case one survives beyond age 100.

For the those interested in the mathematical calculation, here's the idea. After age 90, the withdrawal percentage is calculated by assuming the the previous year's percentage was for an 11-year depletion, calculating the growth rate such that this would happen, then calculating the payment for a 10-year depletion using that rate. Here's the actual formula:

Code: Select all

= PMT(RATE(11;PreviousYearPercentage;-1;0;1;1);10;-1;0;1)


So, what are your thoughts about BahamaMan's suggestion? Should I eventually update the wiki and spreadsheet to add this smooth tail to VPW percentages? Should we eliminate the recommendation to update the Last Withdrawal Age?
longinvest,

thank you for clarifying that, in this context, "pulling the trigger" means "taking the withdrawal as per pVPW", in particular in later years

longinvest: make sVPW the new VPW :beer

it makes a lot of sense to me

Edit: one suggestion:

design the sVPW spreadsheet so that the user can adjust:

1. the age when the "fixed PW" starts,
so that newer generations can adjust that to their own highest life expectancy at that time

2. the number of years during which one will be on a "fixed PW", in case one's health declines soon, to be able to return to increasing VPWs ASAP

I love this refinement of the VPW!

thank you!
Lt. Columbo: Well, what do you know. Here I am talking with some of the smartest people in the world, and I didn't even notice!

User avatar
BlueEars
Posts: 3509
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: Variable Percentage Withdrawal (VPW)

Post by BlueEars » Sat Oct 08, 2016 6:11 pm

longinvest wrote:....
In a private discussion, BahamaMan shared with me some anecdotal evidence. He observed a big decline in his older friends at around age 85. Here are his words: "Many friends of mine over the last 10 years that were very healthy at age 80, contracted an illness a few years later that led to a very rapid decline and they died around age 85-88 ....... I cannot stress enough how rapid the decline is from age 80 to age 85......."

...
Here are the percentages for plain VPW (pVPW) and what I'll temporary call smooth-tail VPW (sVPW) for comparison:

Code: Select all

Asset Allocation: 50% stocks / 50% bonds

Age   pVPW    sVPW
90   11.6%   11.6%
91   12.7%   12.4%
92   14.0%   13.3%
93   15.8%   14.1%
94   18.1%   14.9%
95   21.4%   15.7%
96   26.3%   16.4%
97   34.5%   17.1%
98   50.8%   17.8%
99   100.0%   18.5%
100   -   19.1%

...

,,,
So, what are your thoughts about BahamaMan's suggestion? Should I eventually update the wiki and spreadsheet to add this smooth tail to VPW percentages? Should we eliminate the recommendation to update the Last Withdrawal Age?

First I think that the quote about after age 80 is very interesting. I would be very interested to know if any of his friends were in excellent physical shape (good muscle and aerobic fitness). The age 80 people were perhaps in a generation that did not emphasize exercise but I could be mistaken.

Second, so much depends on how large the dollar value of the portfolio is relative to income needs. A 15.7% withdrawal at age 95 might be way more then is needed or just barely enough. We won't know that until we are nearer that age. For us, our portfolio provides more then we need for even fairly luxurious living at age 68. If it were cut in half by age 95, a 15.7% withdrawal would probably still be well above our needs. As an example, we might have sold the big house by then and have much increased the liquid portfolio value. So the absolute liquid portfolio value is important, not just the percentages.

I guess I am saying that sVPW may be closer to the mark but still maybe is going to have to be refined a lot for an individual's situation. 20 years ago I was not able to foresee our current situation and 20 years from now we and the world may be very different. I see VPW as a good tool for looking a few years but will need forward revisions as we progress.

User avatar
LadyGeek
Site Admin
Posts: 40422
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sat Oct 08, 2016 7:18 pm

longinvest wrote:...The simplest approach would have been to put a cap on percentages, but I personally don't like this because it leads to selecting an arbitrary number. Simply fixing the withdrawal percentage at the age 90 percentage leads to an abrupt change in withdrawals. So, after some back and forth, I proposed a simple formula that automatically adapts to the selected asset allocation (AA) and keeps the withdrawal decline relatively smooth.

Here are the percentages for plain VPW (pVPW) and what I'll temporary call smooth-tail VPW (sVPW) for comparison:

Code: Select all

Asset Allocation: 50% stocks / 50% bonds

Age   pVPW    sVPW
90   11.6%   11.6%
91   12.7%   12.4%
92   14.0%   13.3%
93   15.8%   14.1%
94   18.1%   14.9%
95   21.4%   15.7%
96   26.3%   16.4%
97   34.5%   17.1%
98   50.8%   17.8%
99   100.0%   18.5%
100   -   19.1%
101   -   19.8%
102   -   20.4%
103   -   20.9%
104   -   21.5%
105   -   22.0%
106   -   22.6%
107   -   23.1%
108   -   23.6%
109   -   24.0%
110   -   24.5%
111   -   24.9%
112   -   25.3%
113   -   25.8%
114   -   26.2%
115   -   26.5%
116   -   26.9%
117   -   27.3%
118   -   27.6%
119   -   28.0%

You can see how pVPW percentages keep increasing faster and faster to reach 100% at age 99, while sVPW percentages increase more slowly and remain at a level where a human is emotionally able to to make the withdrawal.

For the those interested in the mathematical calculation, here's the idea. After age 90, the withdrawal percentage is calculated by assuming the the previous year's percentage was for an 11-year depletion, calculating the growth rate such that this would happen, then calculating the payment for a 10-year depletion using that rate. Here's the actual formula:

Code: Select all

= PMT(RATE(11;PreviousYearPercentage;-1;0;1;1);10;-1;0;1)
...

To me, this is changing the shape of the curve to fit a narrowly defined problem. I am very hesitant to implement an "under-the-hood" bias for one specific situation. Why a 10 year depletion and not 9 years or something user-defined?

I say "under-the-hood" because most users would miss this point. Remember, this is a planning tool. They would simply adjust the inputs until the numbers aligned with their desired situation. I don't think a 35 year forecast would benefit from anything more than a simple extrapolation.

I compared the changes using a different technique. First, I extended VPW's Last Withdrawal Age to 119 (aligns with longinvest) in VPW!C6 and Backtesting!C4, which will extend the VPW Table Calculation range in "Table". Then, I put the actual "actual" formula in Table!J45 and filled cells down to Table!J74.

Code: Select all

=IF(H45="","",ROUND(PMT(RATE(11,J44,-1,0,1,1),10,-1,0,1) ,3))

For additional perspective, here are the percentages starting from age 80 (the formula kicks in at 90).

Code: Select all

Age   plain   smoothed
80   4.5%   4.5%
81   4.5%   4.5%
82   4.6%   4.6%
83   4.6%   4.6%
84   4.7%   4.7%
85   4.8%   4.8%
86   4.9%   4.9%
87   4.9%   4.9%
88   5.0%   5.0%
89   5.1%   5.1%
90   5.2%   6.0%
91   5.3%   6.9%
92   5.4%   7.8%
93   5.5%   8.7%
94   5.7%   9.6%
95   5.8%   10.5%
96   6.0%   11.4%
97   6.1%   12.3%
98   6.3%   13.1%
99   6.5%   13.9%
100   6.8%   14.7%
101   7.0%   15.5%
102   7.3%   16.2%
103   7.6%   16.9%
104   7.9%   17.6%
105   8.3%   18.3%
106   8.8%   19.0%
107   9.3%   19.6%
108   10.0%   20.2%
109   10.7%   20.8%
110   11.6%   21.4%
111   12.7%   21.9%
112   14.0%   22.4%
113   15.8%   22.9%
114   18.1%   23.4%
115   21.4%   23.9%
116   26.3%   24.4%
117   34.5%   24.8%
118   50.8%   25.2%
119   100.0%   25.6%

You can see the 0.9% jump in the rightmost column at 90, which is where the formula kicks in.

Extending the current version to 119 does not make practical sense here. My point is that the current VPW calculations will always have a "cliff effect" in the last few years, regardless of the end-point.

I don't see much difference between inserting a breakpoint at 90 to change the curve slope, or, extending the current calculation a few years. Why complicate things, which increases the difficulty for the user, when you can simply move the end-point?

(Spreadsheet used: LibreOffice Calc)
To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

User avatar
siamond
Posts: 3323
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Sat Oct 08, 2016 9:15 pm

LadyGeek wrote:I don't see much difference between inserting a breakpoint at 90 to change the curve slope, or, extending the current calculation a few years. Why complicate things, which increases the difficulty for the user, when you can simply move the end-point?

I've used a 'grace period' in my own PMT models for quite a while now for this precise reason, to move the end point. Making the PMT duration equal to the expected retirement period never struck me as something realistic, I always added such a grace period (at least 10 years, maybe 20) in my models. For the reason BahamaMan listed (the shape of the portfolio trajectory in the last decade wouldn't be bearable). And also because this is probably LTC time or close to it, where variable spending budgets no longer apply (plus judgment is impaired). And by then, spending could be quite high (some of it may be covered by other means, but probably or possibly not all of it). Furthermore, for some of us, there might be a bequest motive. Oh, and some of us might actually live beyond 100...

Now in order to address those issues, this just can't be fixed by changing the shape of things starting at age 90, you need to start earlier, way earlier, and spend a little less, in order to give some breathing room to the portfolio and make the whole trajectory smoother during the retirement period. A grace period, added to the expected retirement duration, addresses those issues without changing anything to the core of the algorithm. It could be higher or lower, depending on LTC and/or bequest goals, but it should not be zero.

EDIT: if we were to make a change to the core algorithm, then I would also use the grace period as a minimum cap for the 'nper' parameter of the PMT formula. As a rough way to account for truly unexpected longevity risk.
Last edited by siamond on Sat Oct 08, 2016 10:56 pm, edited 1 time in total.

User avatar
Lieutenant.Columbo
Posts: 1005
Joined: Sat Sep 05, 2015 9:20 pm
Location: 30th Parallel North

Re: Variable Percentage Withdrawal (VPW)

Post by Lieutenant.Columbo » Sat Oct 08, 2016 9:30 pm

siamond wrote:
LadyGeek wrote:I don't see much difference between inserting a breakpoint at 90 to change the curve slope, or, extending the current calculation a few years. Why complicate things, which increases the difficulty for the user, when you can simply move the end-point?

I've used a 'grace period' in my own PMT models for quite a while now for this precise reason, to move the end point. Making the PMT duration equal to the expected retirement period never struck me as something realistic, I always added such a grace period (at least 10 years, maybe 20) in my models. For the reason BahamaMan listed (the shape of the portfolio trajectory in the last decade wouldn't be bearable). And also because this is probably LTC time or close to it, where variable spending budgets no longer apply (plus judgment is impaired). And by then, spending could be quite high (some of it may be covered by other means, but probably or possibly not all of it). Furthermore, for some of us, there might be a bequest motive. Oh, and some of us might actually live beyond 100...

Now in order to address those issues, this just can't be fixed by changing the shape of things starting at age 90, you need to start earlier, way earlier, and spend a little less, in order to give some breathing room to the portfolio and make the whole trajectory smoother during the retirement period. A grace period, added to the expected retirement duration, addresses those issues without changing anything to the core of the algorithm. It could be higher or lower, depending on LTC and/or bequest goals, but it should not be zero.
siamond,

let's see if I understand the essence of what you are saying:

you are saying one could still use longinvest's original VPW (pVPW) spreadsheet, only this time entering a retirement duration at least a decade longer (your grace period) than reasonable highest life expectancy, so that withdrawals are calculated assuming from the beginning a really long retirement, instead of starting to assume a long retirement only when one turns 90;

is that what you are saying?
Lt. Columbo: Well, what do you know. Here I am talking with some of the smartest people in the world, and I didn't even notice!

User avatar
Lieutenant.Columbo
Posts: 1005
Joined: Sat Sep 05, 2015 9:20 pm
Location: 30th Parallel North

Re: Variable Percentage Withdrawal (VPW)

Post by Lieutenant.Columbo » Sat Oct 08, 2016 9:42 pm

LadyGeek wrote:You can see the 0.9% jump in the rightmost column at 90, which is where the formula kicks in.

Extending the current version to 119 does not make practical sense here. My point is that the current VPW calculations will always have a "cliff effect" in the last few years, regardless of the end-point.

I don't see much difference between inserting a breakpoint at 90 to change the curve slope, or, extending the current calculation a few years. Why complicate things, which increases the difficulty for the user, when you can simply move the end-point?

(Spreadsheet used: LibreOffice Calc)
LadyGeek,

One of the concerns longinvest has repeatedly expressed (and addressed with his VPW spreadsheet) is avoiding under-spending during early retirement years; have you run a comparison between the same two scenarios you just looked at and this time looked at whether the differences are significant in early retirement?
Lt. Columbo: Well, what do you know. Here I am talking with some of the smartest people in the world, and I didn't even notice!

User avatar
siamond
Posts: 3323
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Sat Oct 08, 2016 9:48 pm

Lieutenant.Columbo wrote:you are saying one could still use longinvest's original VPW (pVPW) spreadsheet, only this time entering a retirement duration at least a decade longer (your grace period) than reasonable highest life expectancy, so that withdrawals are calculated assuming from the beginning a really long retirement, instead of starting to assume a long retirement only when one turns 90;

is that what you are saying?

Well, mostly yes. But I wouldn't exactly describe it that way. The point isn't to plan for an extra long retirement period. The point is to plug in the PMT formula (the core of the VPW method) a duration which is higher than your highest life expectancy. So that the portfolio doesn't start aggressively depleting in the last decade, even if you do get close to such high life expectancy. I've ran the numbers many times on this, and the impact on the early years is actually pretty small, and the benefit in the late years very clear.

EDIT: just saw your last post. To (statistically) skew withdrawals towards the early years, better play on the 'rate' parameter than the duration. Add 0.25% or 0.5% and this will do the trick.
Last edited by siamond on Sat Oct 08, 2016 9:52 pm, edited 1 time in total.

AlohaJoe
Posts: 2416
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Variable Percentage Withdrawal (VPW)

Post by AlohaJoe » Sat Oct 08, 2016 9:49 pm

longinvest wrote:He expressed the opinion that changing the Last Withdrawal Age at age 80 is not a good idea. For one thing, this could lead to an immediate and significant impact on withdrawals. For another, he thinks that this is premature, as many are still healthy at 80.


I think the goal is a good one and this is an interesting step: the use of rate like that is novel, from what I can tell, and it is cool to think about how it might be used as a building block. But I'm not yet convinced this is quite right.

Requiring people to change the Last Withdrawal Rate is the biggest failing in VPW. It should be built into the spreadsheet if that's how things should work. Elsewhere I've seen VPW proponents say that it is simple enough that you can print out a list of percentages and leave them to your heirs in case you die or are incapacitated. The need to go into a spreadsheet and make changes undermines this simplicity substantially. So fixing this one way or another is a good idea.

The proposed sVPW approach seems to ignore longevity risk until it shows up and then tries to adjust quickly. That strikes me as akin to SWR ignoring sequence of returns risk until it shows up and then (failing) to adjust. If you're worried about longevity, then it seems like a strategy should be taking that into account before age 90. It isn't like there is no longevity risk at age 89 and then it suddenly arrives the next year.

Choosing age 90 feels like a totally arbitrary magic number as is the 10 year span; if I'm gong to have magic numbers, why not just pick "cap withdrawals at 20%" and keep things simple?

(Random historical note: the very first post on this page has longinvest suggesting a cap on withdrawals of 20% :happy )

FWIW, just adjusting Last Withdrawal Age is certainly immediate, as BahamaMan says, but I'm not convinced it is any more significant that other withdrawal impacts retirees need to be able to live with under VPW. At least this one you can plan for years in advance (i.e. get rid of the vacation home). You could turn 90, have the market or inflation go wonky, and need to cut real spending by 20% as well. I guess it is about trying to reduce unnecessary income volatility but for some reason that feels like an unsatisfying or incomplete explanation. I'll have to think about it some more.

For any changes to the PMT slope I'd prefer to see it calibrated against real world research and not just mathematics. How quickly do actual retirees change their spending? Can a 90 year old actually cut their spending by amounts required? Otherwise, aren't we in the position of needing to revise it again in 5-10 years when BahamaMan reports back, "Actually, cutting my expenses by that much at age 92 turned out to be less easy & fun than I thought it would be"?

FWIW, in several papers by Larry Frank (with several co-authors) they often apply a (1/N) reduction to try to mitigate these kind of exponential increases in PMT/RMD type withdrawal schemes, acknowledging that real retirees are unlikely to ever withdraw more than 10-20% of a portfolio in a single year, regardless of their age and health. The actual calculation they use is:

Code: Select all

adjusted_withdrawal_rate = withdrawal_rate - (withdrawal_rate / life_expectancy)

(Or "end of simulation/spreadsheet" if they're not using life expectancies.)

Example
Say, you're 95 and PMT says to withdraw 21.3%:

= 21.3 - (21.3 / 4.84)
= 16.8


I've thought it was interesting but felt like a bit of a hack/workaround for reasons I can't quite put my finger on.

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 9:51 am

BlueEars wrote:A 15.7% withdrawal at age 95 might be way more then is needed or just barely enough. We won't know that until we are nearer that age.

BlueEars,

I should have been clearer. Longevity risk is a risk of ruin. I do not think that it is a good idea to expose oneself to such a risk. The way I propose to eliminate longevity risk is to:
  • Maximize cheap non-portfolio inflation-indexed base income by delaying Social Security until age 70.
  • Fill the gap between retirement and the start of Social Security payments using a liability matched TIPS ladder (or something equivalent, given the time frame involved), so that the income base is not exposed to market and inflation risk (or minimally exposed to it).
  • Accumulate a big enough portfolio (separate from the TIPS ladder) so that even after a 50% drop, VPW withdrawals will be big enough to fund, when combined with non-portfolio base income, all non-discretionary expenses and enough discretionary expenses to make life highly enjoyable for the retiree. Otherwise, one should delay retirement, because retirement should be about more than mere survival! At least, that's my opinion.
  • When approaching age 80, the price of life inflation-adjusted single premium immediate annuities (SPIA) gets more affordable, specially when bought without a guarantee (e.g. as soon as both spouses die, all payments stop without any residual value). I propose to take advantage of this to eliminate longevity risk at that point by buying enough life inflation-indexed SPIA to fund a comfortable lifelong floor of income (when combined with Social Security), just in case one of the spouses gets to live for a very long time, beyond what was projected.

It is in the above context that the withdrawal glide path takes place. Mainly, the survivor is already receiving his Social Security and SPIA payments monthly, and is taking VPW withdrawals to fund highly-discretionary expenses. In other words, one wouldn't be ruined even if the portfolio was completely depleted at age 100.

BahamaMan, taking into account his personal circumstance, does not anticipate surviving until age 100. Yet, if he was to survive until age 97, he would feel uncomfortable taking a 35% portfolio withdrawal. So, he was suggesting to limit the percentage increase for behavioral reasons, not as a way to extend portfolio survival beyond 100.

If you believe that you have a real possibility to live until age 110, you should pick something like age 114 as Last Withdrawal Age. That's why the spreadsheet allows for specifying this parameter. The proposed glide path is not meant to replace this parameter.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 10:19 am

LadyGeek,

LadyGeek wrote:To me, this is changing the shape of the curve to fit a narrowly defined problem. I am very hesitant to implement an "under-the-hood" bias for one specific situation. Why a 10 year depletion and not 9 years or something user-defined?


It's a good question. It could be user defined, but let's be realistic. 5 years would be too short as to eliminate the behavioral problems (e.g. the percentage is already very high at that point), and 20 years would indicate a desire to significantly extend portfolio survival beyond the initial limit, which is best managed by selecting a higher Last Withdrawal Age way earlier, at the time of retirement.

LadyGeek wrote:I say "under-the-hood" because most users would miss this point. Remember, this is a planning tool. They would simply adjust the inputs until the numbers aligned with their desired situation. I don't think a 35 year forecast would benefit from anything more than a simple extrapolation.


This is the real danger of the proposed glide path. Users might miss that the glide path is no meant to really extend portfolio survival, but simply to help with potential behavioral problems in one's last years.

Some members might have forgotten, but in the early versions of VPW, the percentage was capped at 20%*. I eliminated this cap to make users fully aware that the VPW plan was limited in time. I thought that the 20% cap gave the false impression that (sufficient) withdrawals could continue indefinitely.

* VPW was inspired by Canadian RRIF withdrawal rules which have such a cap.

I gather that you think that eliminating the cap was a good thing and reintroducing it (through a glide path) would not be such a good idea.

LadyGeek wrote:I don't see much difference between inserting a breakpoint at 90 to change the curve slope, or, extending the current calculation a few years. Why complicate things, which increases the difficulty for the user, when you can simply move the end-point?


Good question.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
BlueEars
Posts: 3509
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: Variable Percentage Withdrawal (VPW)

Post by BlueEars » Sun Oct 09, 2016 10:27 am

deleted
Last edited by BlueEars on Sun Oct 09, 2016 10:36 am, edited 1 time in total.

User avatar
BlueEars
Posts: 3509
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Re: Variable Percentage Withdrawal (VPW)

Post by BlueEars » Sun Oct 09, 2016 10:34 am

BlueEars wrote:
longinvest wrote:...
If you believe that you have a real possibility to live until age 110, you should pick something like age 114 as Last Withdrawal Age. That's why the spreadsheet allows for specifying this parameter. The proposed glide path is not meant to replace this parameter.

I have set the Last Withdrawal Age to 110 in the latest VPW. Then I set the start year = 1968, one of the worst in modern times. This keeps our portfolio above 50% decline, with very reasonable spending, basically out to age 102. Then the withdrawals get high, maybe too high (above 13%). But those really high withdrawals I see as very theoretical at this point. VPW is a great tool but I would not be using a frozen snapshot from 2016 in year 2050 (at age 102).

However, we could get a sequence that is even worse then the 1968 one. Not likely but who knows for sure. I want to keep my peace of mind and perhaps leave some for our heirs. That way it is more comforting if the heirs screw up financially in our lifetime. :happy Even the 1968 start year resulted in the porfolio dropping temporarily to just below the 50% point at year 15. If that happened my peace-of-mind would be out the window. It's easy to look at the year 16 and beyond upward trajectory and think "well that worked out nicely". But I'm not sure how I'd react if it actually happened at age 82 per the simulation.

Regarding SPIA, that is a decision for the future and hopefully we will not need that. If we were 80 today, the SPIA would be unnecessary given our portfolio value. Regarding SS, we took it before age 70 after looking at the options. I do not regret the decision. So Longinvest, we are different in our investment choices but more in common then differences I believe. There is such a huge variation in investment choices on Bogleheads but in general people on this forum are quite level headed ... in investing that is.

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 11:02 am

siamond wrote:I've used a 'grace period' in my own PMT models for quite a while now for this precise reason, to move the end point. Making the PMT duration equal to the expected retirement period never struck me as something realistic, I always added such a grace period (at least 10 years, maybe 20) in my models. For the reason BahamaMan listed (the shape of the portfolio trajectory in the last decade wouldn't be bearable). And also because this is probably LTC time or close to it, where variable spending budgets no longer apply (plus judgment is impaired). And by then, spending could be quite high (some of it may be covered by other means, but probably or possibly not all of it). Furthermore, for some of us, there might be a bequest motive. Oh, and some of us might actually live beyond 100...

Siamond,

See my reply to BueEars above. The glide is meant to be used in a context where longevity risk has been eliminated.

Personally, I think that using inflation-indexed SPIAs when they become cost-effective (near age 80) is less expensive than extending VPW until age 125 (the current estimated biological upper limit on human life). Even when there are bequest motives, I think that the use of SPIAs is justified for funding normal expenses, if only for simplicity when one is in old age. What's left of the portfolio will provide for extra spending and covering financial needs related to end-of-life care, and bequest. I'm not so much concerned about funding 25 years of LTC (e.g. do I really want to spend the last 25 years of my life in the equivalent of a prison cell?).

siamond wrote:Now in order to address those issues, this just can't be fixed by changing the shape of things starting at age 90, you need to start earlier, way earlier, and spend a little less, in order to give some breathing room to the portfolio and make the whole trajectory smoother during the retirement period. A grace period, added to the expected retirement duration, addresses those issues without changing anything to the core of the algorithm. It could be higher or lower, depending on LTC and/or bequest goals, but it should not be zero.


We are in agreement, here. We might disagree somewhat on the length of the buffer (10 years seems like a lot), but the principle remains. The thing is that almost all the retirement literature assumes death before age 95 (e.g. retire at age 65, and plan for a 30-year SWR). VPW has a 5-year buffer. If one has good reasons to think that he has the genes and health to live longer, he could use a higher Last Withdrawal Age, but I think that it would be more efficient to buy an inflation-indexed SPIA when approaching age 80, instead. This way, we won't make the younger self sacrifice the enjoyment of his savings when both him and his loved ones are able to build the nice memories, only to delay the availability of money to age 105, when both him and his children are riddled with disabilities and loss of short-term memory.

For those with bequest motives, I see two cases. Those who wish to give the money to their children, it might be a good idea to give it while the children could use the money effectively. This might be a case for keeping the Last Withdrawal Age at 100. For those who wish to make a big post-death donation to an organization, while keeping a maximum amount of liquidity while alive, selecting a high Last Withdrawal Age might be a good fit.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 11:36 am

BlueEars wrote:Regarding SPIA, that is a decision for the future and hopefully we will not need that. If we were 80 today, the SPIA would be unnecessary given our portfolio value. Regarding SS, we took it before age 70 after looking at the options. I do not regret the decision. So Longinvest, we are different in our investment choices but more in common then differences I believe. There is such a huge variation in investment choices on Bogleheads but in general people on this forum are quite level headed ... in investing that is.


It's all about efficient use of money. What is more efficient: extending VPW's last withdrawal age, or eliminating longevity risk once and for all when it becomes cost-effective to do so?

What I am saying is that one could make sure to get lifelong comfortable inflation-indexed income buying enough SPIA at 80 (not buying more than necessary, though).

Then the rest becomes moot. One is most likely to die before 100, so why not target spending (and donating) the money while younger, healthier, and still in full possession of short-term memory? Even though a few lucky people get to live beyond 100, just a small minority of them do so without disabilities such as reduced eyesight, reduced hearing, reduced mobility, reduced memory, etc.

Of course, if the goal is to leave a significant bequest to an organization after death, then maybe selecting age 110 is a good match.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
siamond
Posts: 3323
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Sun Oct 09, 2016 11:40 am

longinvest wrote:
siamond wrote:I've used a 'grace period' in my own PMT models for quite a while now for this precise reason, to move the end point. Making the PMT duration equal to the expected retirement period never struck me as something realistic, I always added such a grace period (at least 10 years, maybe 20) in my models.

Personally, I think that using inflation-indexed SPIAs when they become cost-effective (near age 80) are less expensive than extending VPW until age 125 (the current estimated biological upper limit on human life). Even when there are bequest motives, I think that the use of SPIAs is justified for funding normal expenses, if only for simplicity when one is in old age. What's left of the portfolio will provide for extra spending and covering financial needs related to end-of-life care, and bequest. I'm not so much concerned about funding 25 years of LTC (e.g. do I really want to spend the last 25 years of my life in the equivalent of a prison cell?).

I wasn't suggesting a life expectancy of 125 (very unlikely, I agree, even with my wariness to use today's information to judge what will happen 20 or 30 years from now), nor 25 years of LTC (this never happens). Five to 10 years of LTC can be VERY expensive though. Much more than the annual budget I would allocate to myself until then. So my point is that you'd better come to it with a portfolio which is NOT in a steep downward slope, to be able to afford a nice facility. And LTC is just one consideration among others I listed. I'm glad BahamaMan *finally* understood the behavioral issue (which I pointed out many pages ago in this thread!). Etc.

longinvest wrote:We are in agreement, here. We might disagree somewhat on the length of the buffer (10 years seems like a lot), but the principle remains. The thing is that almost all the retirement literature assumes death before age 95 (e.g. retire at age 65, and plan for a 30-year SWR). VPW has a 5-year buffer. If one has good reasons to think that he has the genes and health to live longer, he could use a higher Last Withdrawal Age, but I think that it would be more efficient to buy an inflation-indexed SPIA when approaching age 80, instead.

The fundamental issue is this assumption equating the PMT duration to the expected retirement period/lifetime. It doesn't have to be this way. YES, the use of a buffer of sorts (what I call 'grace period') can solve many problems. It just needs to be made more explicit, without pre-judging what it should be (this depends on individual goals), while the graphical representations of the portfolio trajectory should only show the expected retirement period (buffer NOT included). And then we get to much more satisfying trajectories. The portfolio going to zero just can't be a real-life scenario.

As to SPIAs, I do agree this is a possibility, but those considerations (incl. TIPS ladders and so on) are VERY dependent on the individual's circumstances. It's perfectly fine for you to mention such possibilities in the spreadsheet, because they are part of retirement planning for sure, but I don't think you should describe them as a one size-fits-all recommendation. Just make a buffer parameter explicit, and VPW will become a much more flexible and realistic tool, in addition to other tools like SPIAs and the likes. If you want to make it 5 years by default, why not, although this does seem a bit short.

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 11:44 am

AlohaJoe wrote:The proposed sVPW approach seems to ignore longevity risk until it shows up and then tries to adjust quickly.


That's a misunderstanding of the context in which it is meant to be used, which is a context where longevity risk has already been eliminated.

LadyGeek is probably right; the proposed glide path seems to open the door for such misunderstanding. Having VPW withdrawals drop to zero, in a naive use of the plan, makes it clear that longevity has to be dealt with.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 11:51 am

siamond wrote:Just make a buffer parameter explicit, and VPW will become a much more flexible and realistic tool, in addition to other tools like SPIAs and the likes. If you want to make it 5 years by default, why not, although this does seem a bit short.


The Last Withdrawal Age is already explicit. Adding second end-of-life parameters (e.g. buffer) would just complicate things, don't you think?
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
siamond
Posts: 3323
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Sun Oct 09, 2016 12:09 pm

longinvest wrote:
siamond wrote:Just make a buffer parameter explicit, and VPW will become a much more flexible and realistic tool, in addition to other tools like SPIAs and the likes. If you want to make it 5 years by default, why not, although this does seem a bit short.


The Last Withdrawal Age is already explicit. Adding second end-of-life parameters (e.g. buffer) would just complicate things, don't you think?

Well... What I think is more complicated is to overload the semantics of 'last withdrawal age', and to have to explain that you can make it higher in order to address all those issues we've been discussing. Because it no longer means 'last withdrawal age' by then. That, plus the graphical trajectories going down to zero while including an implicit buffer, would probably deter many people from using VPW. You've seen me taking quite a while to go over this perception hurdle and settle on a VPW derivative for my own plans, this was part of my problem. Showing a portfolio trajectory NOT going down to zero by default will gain VPW more users, I am pretty sure of that!

Let me put it in another way. You fully master the effect on the PMT formula of playing on one parameter or another. Regular VPW users do not. So better show them two parameters with simple and clear semantics than trying to overload too much in one single parameter. Even if the application to the PMT formula is a simple addition.

User avatar
LadyGeek
Site Admin
Posts: 40422
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sun Oct 09, 2016 12:12 pm

Lieutenant.Columbo wrote:LadyGeek, One of the concerns longinvest has repeatedly expressed (and addressed with his VPW spreadsheet) is avoiding under-spending during early retirement years; have you run a comparison between the same two scenarios you just looked at and this time looked at whether the differences are significant in early retirement?

No, as my focus was only to compare the end state of the curves. Answers subsequent to your post address this aspect.

longinvest wrote:This is the real danger of the proposed glide path. Users might miss that the glide path is no meant to really extend portfolio survival, but simply to help with potential behavioral problems in one's last years.

Some members might have forgotten, but in the early versions of VPW, the percentage was capped at 20%*. I eliminated this cap to make users fully aware that the VPW plan was limited in time. I thought that the 20% cap gave the false impression that (sufficient) withdrawals could continue indefinitely.

* VPW was inspired by Canadian RRIF withdrawal rules which have such a cap.

I gather that you think that eliminating the cap was a good thing and reintroducing it (through a glide path) would not be such a good idea.

No, I forgot the early version had a cap. :oops: :D I have a different perspective on this latest initiative...

The proposed change is no different than obtaining a consensus on the shape of the S&P 500 index over time. Some think it is lognormal, there is no consensus. When this occurs, the general approach is to assume a Gaussian distribution. Everyone knows this is not reality, but it might be "good enough" and can be accepted as a baseline to work from. For many reasons, a common baseline is needed for "apples-to-apples" comparisons.

Now, let's do some relevant comparisons. Look at much larger datasets than the situations posted in this thread.

- US: IRA distribution tables - Bogleheads (expand Table III)
- Canada: Registered Retirement Income Fund - finiki, the Canadian financial wiki

In the US version, the last few years have an increased distribution capped at 52.63%. The Canadian version is capped at 20%. The starting age is 70 in the US, 65 in Canada. So, we have two datasets with different starting ages and withdrawal rates.

At age 90, the US rate is 8.77%, Canada is 11.92%. Is this 3.15% difference significant? If you are 65 now, forecasting for 25 years is out probably well within the tolerance for accuracy ("in the noise").

If these much larger datasets (which have been constructed by a lot of experts who do this every day) arrive at a solution using a certain slope, then perhaps we should to the same?

(longinvest - For additional perspectives, I recommend submitting this proposal in the Canadian support thread: Variable Percentage Withdrawal (VPW) for Canadians - Financial Wisdom Forum)

==============================
For new investors:

This spreadsheet is one aspect of the "big picture" objective to determine available funds in retirement. For the rest of the picture, try this spreadsheet: Retiree Portfolio Model. It is US based and includes a very cool "automatic withdrawal" feature. Questions should be asked in the support thread.
To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 12:21 pm

siamond wrote:Well... What I think is more complicated is to overload the semantics of 'last withdrawal age', and to have to explain that you can make it higher in order to address all those issues we've been discussing. Because it no longer means 'last withdrawal age' by then. That, plus the graphical trajectories going down to zero while including an implicit buffer, would probably deter many people from using VPW. You've seen me taking quite a while to go over this perception hurdle and settle on a VPW derivative for my own plans, this was part of my problem. Showing a portfolio trajectory NOT going down to zero by default will gain VPW more users, I am pretty sure of that!

Let me put it in another way. You fully master the effect on the PMT formula of playing on one parameter or another. Regular VPW users do not. So better show them two parameters with simple and clear semantics than trying to overload too much in one single parameter. Even if the application to the PMT formula is a simple addition.

Siamond,

There's an implicit assumption, in your text, that one will not eliminate longevity risk when it becomes cost-effective to do so. I disagree. I think that it is probably better for users to avoid using VPW, than to use VPW without being aware of the necessity to eliminate longevity risk.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
siamond
Posts: 3323
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Sun Oct 09, 2016 12:57 pm

longinvest wrote:
siamond wrote:Well... What I think is more complicated is to overload the semantics of 'last withdrawal age', and to have to explain that you can make it higher in order to address all those issues we've been discussing. Because it no longer means 'last withdrawal age' by then. That, plus the graphical trajectories going down to zero while including an implicit buffer, would probably deter many people from using VPW. You've seen me taking quite a while to go over this perception hurdle and settle on a VPW derivative for my own plans, this was part of my problem. Showing a portfolio trajectory NOT going down to zero by default will gain VPW more users, I am pretty sure of that!

Let me put it in another way. You fully master the effect on the PMT formula of playing on one parameter or another. Regular VPW users do not. So better show them two parameters with simple and clear semantics than trying to overload too much in one single parameter. Even if the application to the PMT formula is a simple addition.

There's an implicit assumption, in your text, that one will not eliminate longevity risk when it becomes cost-effective to do so. I disagree. I think that it is probably better for users to avoid using VPW, than to use VPW without being aware of the necessity to eliminate longevity risk.

I didn't make such assumption, but yes, agreed, people have to think about such matters. Well, using a buffer parameter set by default at 5 will show without a doubt that the portfolio will not last forever. Try PortfolioCharts Retirement Spending with VPW and age=65 (since Tyler capped the retirement period at 30 years), this makes the case plenty clear. And maybe this is the right default setting, to make people think about it, and possibly adjust the buffer to 10 on their own, as a conscious decision.

As a side note, I for one do plan to consider the use an SPIA for some of our portfolio late in the game for this precise reason (having good genes in the family -my mother is 97 by now- certainly plays a role). I didn't make it a hard decision now because I don't have to, plus I can't say I like the idea of giving the principal to the insurance company, but this is certainly a possible part of my plan. Again, things are not black & white, one will use a combination of tools, somebody else will use another combination. Those tools need to have some level of flexibility though, to accommodate a broad set of scenarios (and behaviors).
Last edited by siamond on Sun Oct 09, 2016 1:00 pm, edited 1 time in total.

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 1:00 pm

LadyGeek wrote:longinvest - For additional perspectives, I recommend submitting this proposal in the Canadian support thread


LadyGeek,

Thanks for the suggestion, but you have already won the argument. You are right that adding the glide path would simply complicate the tool and could confuse users. The various posts that followed the glide path suggestion by posters who have been studying VPW for a long time should be proof enough that the glide path could be misleading.

I think that, instead, I should try to improve the wiki and spreadsheet presentation to better address the issues of longevity and disability-free life expectancy.

Somehow, insurance companies know a few things that escape many of us. At age 65, they are able to offer inflation-indexed SPIA payouts, which have no volatility whatsoever, that are more attractive* than a VPW percentage with a Last Withdrawal Age of 120 (which would expose the retiree to volatile portfolio withdrawals).

* Let me clarify: "more attractive" but not necessarily higher. When taking withdrawal volatility into account, there's a good likelyhood of higher SPIA income than VPW withdrawal in some retirement years.

A big proportion of retirees don't make it to age 85. By keeping a liquid portfolio and using a more reasonable Last Withdrawal Age of 100, most retirees are likely to get more money while they are young and healthy, and for the many who will die before 85, they'll leave a nice and big bequest behind, instead of having given a significant portion of it to the insurance company.

Near age 80, inflation-indexed SPIAs become cheaper, even though insurance companies are aware of adverse selection among SPIA buyers (e.g. those who think they'll die shortly don't buy the product). For those who survive until this point, my recommendation is to take advantage of the cheaper prices (e.g. higher payouts) to eliminate longevity risk (e.g. the risk of ruin due to long life) by buying enough SPIA so that one will be able to comfortably live on non-portfolio income. From that point on, portfolio income (using VPW withdrawals) will just provide for extra discretionary expenses.

The question is: Should one plan to keep some liquidity beyond VPW's depletion target? My answer is yes, but it should be minimal (except for bequest concerns). There are multiple approaches to this. One is to gradually increase the Last Withdrawal Age, but slowly to avoid moving much income from now to later (that's the current suggestion in the Wiki). Another is BahamaMan's glide path suggestion. And maybe the simplest (which is in the F.A.Q.) is to keep a small money reserve (similar to an emergency fund) on the side which is not subject to VPW withdrawals.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
adrian2
Posts: 27
Joined: Sun Mar 04, 2007 8:14 am
Location: Canada

Re: Variable Percentage Withdrawal

Post by adrian2 » Sun Oct 09, 2016 2:05 pm

AlohaJoe wrote:You have super-conservative people like Waring & Siegel who say you should use age of 120 because, well, that's how long some people live.

A common Jewish birthday wish, in equivalent translation, is "may you live to 120!" :sharebeer

Adrian

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 2:17 pm

siamond wrote:Try PortfolioCharts Retirement Spending with VPW and age=65 (since Tyler capped the retirement period at 30 years), this makes the case plenty clear. And maybe this is the right default setting, to make people think about it, and possibly adjust the buffer to 10 on their own, as a conscious decision.

Siamond,

It is an interesting view. But, I'm reluctant not to show the portfolio depletion. The reason I eliminated the 20% percentage cap was to explicitly show the portfolio value going down to zero to force those who don't read the instructions to be become aware of what is going on. If this pushes some people away (those who will never read instructions), maybe it's a good thing for them(?).

What should be done, though, is to improve how VPW is presented (in the instructions/wiki) to explain how VPW is just a tool within what should be a larger retirement plan. And, we should also provide some examples of simple retirement plans (at least one with bequest motives, and one without bequest motives). Of course, we won't be able to be comprehensive; that would take an entire book!
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
LadyGeek
Site Admin
Posts: 40422
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sun Oct 09, 2016 3:13 pm

In our sister Canadian forum, longinvest is requesting suggestions for improvement. I will repeat my reply here.

See: Subject: Variable Percentage Withdrawal (VPW) for Canadians

LadyGeek wrote:I would like to suggest that the spreadsheet clearly mention the legally required minimum withdrawal rates for tax-deferred accounts. While this a responsibility of the user, it is helpful to have this information inside the spreadsheet. Perhaps the "Instructions" worksheet can include a section:

===============================
Required minimum distributions
Please be aware of the legally required minimum distributions, as this takes priority over planning. Information can be found:

Canadian residents: Registered Retirement Income Fund - finiki, the Canadian financial wiki (applies to LIF, LRIF, and RRIF)
US residents: IRA distribution tables - Bogleheads (applies to IRA and certain employer provided retirement plans)

=================================

Caveat: I am US resident. Please review and modify as appropriate.

Update: See the follow-up post which starts on the next page here.
To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

User avatar
LadyGeek
Site Admin
Posts: 40422
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sun Oct 09, 2016 3:38 pm

longinvest wrote:...The question is: Should one plan to keep some liquidity beyond VPW's depletion target? My answer is yes, but it should be minimal (except for bequest concerns). There are multiple approaches to this. One is to gradually increase the Last Withdrawal Age, but slowly to avoid moving much income from now to later (that's the current suggestion in the Wiki). Another is BahamaMan's glide path suggestion. And maybe the simplest (which is in the F.A.Q.) is to keep a small money reserve (similar to an emergency fund) on the side which is not subject to VPW withdrawals.

Keeping some money "off the table" as a reserve is a very simple way to guarantee you will not deplete your funds.

From an emotional impact, you cannot touch what is not there. I think your suggestion as a similarity to an Emergency fund* is a good one. An emergency fund is for immediate use, perhaps this is better referred to as funds held in reserve (contingency planning).

Would holding an amount in reserve require modification to the spreadsheet? The F.A.Q suggests that you don't have to take out the full amount every year. To me, this is different than a clear value showing how much is held in reserve.

(From a math perspective, holding funds in reserve is an offset to the curve - it does not change the withdrawal rate.)

*The Canadian version is here: Emergency fund - finiki, the Canadian financial wiki
To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 4:06 pm

LadyGeek wrote:I would like to suggest that the spreadsheet clearly mention the legally required minimum withdrawal rates for tax-deferred accounts. While this a responsibility of the user, it is helpful to have this information inside the spreadsheet.


I agree with the follow up: RMDs are mostly a tax event. Nothing forces the retiree to spend the RMD money; it can be reinvested into the portfolio (except for money lost to taxes).
Last edited by longinvest on Sun Oct 09, 2016 4:31 pm, edited 1 time in total.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

longinvest
Posts: 2467
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Oct 09, 2016 4:31 pm

LadyGeek wrote:The F.A.Q. suggests that you don't have to take out the full amount every year. To me, this is different than a clear value showing how much is held in reserve.


LadyGeek,

I was referring to another part of the FAQ (see the bold part):
Q: Why not use a lifetime inflation-indexed single-premium immediate annuity (SPIA), instead?
A: Who told you that you couldn't also use an annuity with part of your money? Some people like to keep a certain amount of liquidity. VPW can be used for the liquid part of your portfolio, if you wish. You could also keep a money reserve on the side on which you don't apply VPW. The VPW table is simply a tool, not an overall retirement solution.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
LadyGeek
Site Admin
Posts: 40422
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sun Oct 09, 2016 4:52 pm

longinvest,

The Answer is inside the SPIA question and I totally missed the intent. Perhaps others have missed this point as well?

Would a user simply reduce the amount of the initial portfolio, or, is a portfolio final value (lower limit, floor) needed?

===============

Suggestion: Copy-n-paste the F.A.Q into the "Instructions" worksheet. This is important information that should be available within the spreadsheet directly.
To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

Post Reply