Variable Percentage Withdrawal (VPW)

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sat May 17, 2014 3:53 pm

1210sda wrote:Longinvest,

Here's my quick feedback.

On the "withdrawals" chart, there is no legend for the red line. The same is true for the "withdrawal statistics" chart.

On the "portfolio balance" chart, the blue line legend only shows B..

Small stuff, OK

1210

That's in Microsoft Excel, right? Could you take a screen capture? I use tinypic.com to display images in my posts.
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: 39148
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sat May 17, 2014 4:08 pm

Longinvest - I see the same thing in MS Excel 2010. It's nothing more than increasing the size of the legend boxes, but I'll let you confirm.

The wiki shows how to post images, which was just updated to show an example with tinypic.com. See: Posting images in the Bogleheads forum (If the image needs no editing, but it is not yet on a Web page)

For those wondering why I spend so much time on minor details, I go by this rule:

You never get a second chance to make a first impression.

The first thing a reader sees are typos and unclear information. It's the least challenging from a technical point of view, but it's the most important thing to get right.

The quality of your user interface is a reflection on the quality of your work. No matter how good everything is "under the hood" - it all goes for naught if the reader doesn't use your product because they can't understand what you are trying to convey (or it's full of typos).
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: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sat May 17, 2014 5:19 pm

LadyGeek, I uploaded a new version with legends above the charts, just in case it helps. I do not have Excel.

I tried changing the size of the legend box, but LibreOffice moves the chart over the legend as soon as I do so. I need a little of your magic, here! :)
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun May 18, 2014 8:42 am

I've uploaded a new version (May 18th) implementing LadyGeek's chart legend solution.

Unfortunaltely, when I try to visualize it using Microsoft's free ExcelViewer, it shows zeros in the legend:
Image
The correct display should show "Withdrawal" and "Inflation", something like:
Image

In Excel, does the legend show correctly?
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: 39148
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sun May 18, 2014 9:38 am

The ExcelViewer agrees with the zeroes displayed in the Google Drive preview (which I just noticed).

In Excel 2010, files downloaded from the internet open in Protected View - no editing is possible until you click on Enable Editing. I see zeroes in the chart legend. :shock:

When I click on Enable Editing, the legends are updated to the correct labels.

I think that the file viewers, including the one in Google Drive, are doing exactly that - viewing the file. The legends are linked to cells in the spreadsheet, and are not displayed until Excel is allowed to "calculate" what the value should be. So in hindsight... everything is working correctly.
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: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sat May 24, 2014 11:06 am

I have uploaded a new version of the VPW spreadsheet on the Bogleheads Wiki VPW Page.

Changes:

Enjoy!
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: 39148
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sat May 24, 2014 1:49 pm

I'm using MS Excel 2010. In the VPW tab:

- Consider saving the file with the Start Year to 1980, which is the earliest Canadian year. This will set the value used when the file is opened - it's currently set to the Simba data set and 1972.

When I flip the Data Set to Canada, quite a number of invalid cell values appear and the charts are broken. Setting the Start Year to 1980 (latest start date of the data set) will fix the problem.

- Something doesn't look right in the VPW table, Year 35. Set the Start Year to 1980. Either I'm not interpreting the data correctly, or the last data point needs to be taken care of.

Shiller - 2013 shows up as the last year, but the data ends in 2012.
Simba - 2014 shows up as the last year, but the data ends in 2013.
Canada - 2014 shows up as the last year, but the data ends in 2013.

The Withdrawal Statistics chart is pulling data from the entire data set (Shiller: 1871 - 2012). Is that the intention, or should it align with the Simulation?
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: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sat May 24, 2014 2:15 pm

LadyGeek,

LadyGeek wrote:Consider saving the file with the Start Year to 1980, which is the earliest Canadian year. This will set the value used when the file is opened - it's currently set to the Simba data set and 1972.

When I flip the Data Set to Canada, quite a number of invalid cell values appear and the charts are broken. Setting the Start Year to 1980 (latest start date of the data set) will fix the problem.

That makes a lot of sense. I've just uploaded a new version with 1980 as default start year.

- Something doesn't look right in the VPW table, Year 35. Set the Start Year to 1980. Either I'm not interpreting the data correctly, or the last data point needs to be taken care of.

Shiller - 2013 shows up as the last year, but the data ends in 2012.
Simba - 2014 shows up as the last year, but the data ends in 2013.
Canada - 2014 shows up as the last year, but the data ends in 2013.

It looks correct to me... Let me explain. Withdrawals are taken at the beginning of the year. So, we don't need the last year's return to compute its withdrawal. :D

(I have yet to display an additional year when its balance is 0$ and the previous year has no return.)

The Withdrawal Statistics chart is pulling data from the entire data set (Shiller: 1871 - 2012). Is that the intention, or should it align with the Simulation?

Argh! The title is wrong. I made a mistake when I recreated the chart to solve the Excel problem. The title should be "Withdrawal Statistics for Every Start Year". So, yes, the intention is to cover the entire data set.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sat May 24, 2014 3:38 pm

longinvest wrote:
LadyGeek wrote:- Something doesn't look right in the VPW table, Year 35. Set the Start Year to 1980. Either I'm not interpreting the data correctly, or the last data point needs to be taken care of.

Shiller - 2013 shows up as the last year, but the data ends in 2012.
Simba - 2014 shows up as the last year, but the data ends in 2013.
Canada - 2014 shows up as the last year, but the data ends in 2013.

It looks correct to me... Let me explain. Withdrawals are taken at the beginning of the year. So, we don't need the last year's return to compute its withdrawal. :D

LadyGeek,

I'm starting to have doubts. I was looking at the display in LibreOffice. Maybe it doesn't show correctly in Excel?

longinvest
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: 39148
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sat May 24, 2014 4:34 pm

The numbers match in both LibreOffice Calc and Excel.

The only difference is in the display of the chart title "Withdrawal Statistics for Every Start Year", where Excel inserts extra space between the bottom of the title and the top of the chart area (minor discrepancy, not worth fixing).

longinvest wrote:It looks correct to me... Let me explain. Withdrawals are taken at the beginning of the year. So, we don't need the last year's return to compute its withdrawal. :D

I'm always off by one... somewhere. :)
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: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sat May 24, 2014 5:17 pm

LadyGeek wrote:
longinvest wrote:It looks correct to me... Let me explain. Withdrawals are taken at the beginning of the year. So, we don't need the last year's return to compute its withdrawal. :D

I'm always off by one... somewhere. :)

There are 35 withdrawals. So, we need 34 years of returns (as the 35th withdrawal of 100% is made at the start of year 35).

1980: 1
1981: 2
1982: 3
...
2013: 34 (that's all we need)

Added: Here's the tail of the Shiller simulation for 1980.
Image
See how the last year with a return is 2012 (that's where the data set ends). The last simulated withdrawal in 2013 is the 34th withdrawal at age 98. There is one unknown withdrawal left in 2014 (percentage 100%) at the age of 99.

You'll probably enjoy the missing dollar riddle: viewtopic.php?f=10&t=138661#p2052009 :D
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: 39148
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sun May 25, 2014 3:58 pm

To our Canadian readers:

longinvest has started a discussion in our sister Canadian forum, Financial Wisdom Forum. See: Variable Percentage Withdrawal (VPW) for Canadians

Refer to this thread for Canadian-specific topics. (Or ask your questions in that thread.)

Both myself and longinvest are members of the Financial Wisdom Forum.
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: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Wed May 28, 2014 8:13 pm

I have uploaded a new version of the VPW spreadsheet on the Bogleheads Wiki VPW Page.

Changes:
  • LadyGeek updated the Instructions sheet
  • LadyGeek modified the VPW sheet:
    • Inserted a blank column to separate the entry / VPW table from the backtesting area
    • Retitled the backtesting columns and charts (removed a "subtitle") entry in each chart
  • longinvest further improved the Statistics for All Start Years table subtitles.

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

longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Thu Jun 12, 2014 8:33 am

From viewtopic.php?f=2&t=140976&p=2087603#p2087599
ResearchMed wrote:...
Next problem: I'm getting a message about
"Significant loss of functionality"
"One or more cells in this workbook contain data validation rules which refer to values on other worksheets. These data validation rules will not be saved. Location: 'VPW' "

Will this affect how the software "works"?

RM


There shouldn't be any dependency on other worksheets. Something has gone wrong if you get this message. I will look at it and, if necessary, I will completely rebuild the spreadsheet.

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

freebeer
Posts: 1853
Joined: Wed May 02, 2007 8:30 am
Location: Seattle area USA

Re: Variable Percentage Withdrawal (VPW)

Post by freebeer » Thu Jun 12, 2014 8:50 am

Cut-Throat wrote:
ot1138 wrote:* You can't create a plan based on the number of years until you die. You don't know this with any certainty.



Sure you Can !

You don't have to be completely Certain about this. If you pick age 100 you'll do Fine...
And if you use VPW..... You'll never run out of money!


These points are certainly true. To take a step further, if you pick age 120 you'll be even more certain, and if you always withdraw 3% of your current portfolio value you'll never run out of money. But in that case you will definitely be over-saving and you may or may not be spending enough to fund your obligations.

If you have an extraordinarily large bequest motive and a job you love (so don't mind working many years longer) that pays you a lot more than you need (so you can super-save or else you are OK to super-scrooge during working years) and you don't mind highly variable spending (so you may be eating a lot of rice and beans some years) then this is all OK.

But the average person wants to look at retirement from the perspective of the minimum savings necessary to provide "consumption smoothing" in both working years and retirement years, including the ability to retire as soon as feasible and reasonable predictability in spending. It is not clear to me that the particular algorithm of VPW is an optimal way to meet these constraints.

I'm also confess to being skeptical any time someone tries to label one particular approach to something as "the" thing, especially without comparison of how that particular approach compares to other similar approaches. There are many variable percentage withdrawal strategies ranging from Taylor Larimore's pragmatic approach to more complex algorithms. Wade Pfau has written a bit about it here: http://wpfau.blogspot.com/2012/04/varia ... ement.html but I would love to see more of this type of analysis.

longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Thu Jun 12, 2014 8:51 am

RM,

longinvest wrote:From viewtopic.php?f=2&t=140976&p=2087603#p2087599
ResearchMed wrote:...
Next problem: I'm getting a message about
"Significant loss of functionality"
"One or more cells in this workbook contain data validation rules which refer to values on other worksheets. These data validation rules will not be saved. Location: 'VPW' "

Will this affect how the software "works"?

RM

...

I am unable to reproduce the problem using LibreOffice or Microsoft Excel's viewer.

What this message usually indicates is that an external spreadsheet would be needed to validate user provided input data.

The only place where I use validation rules is for drop-down selections, like the "Data Set" and "Start Year" entries in the Backtesting Parameters box of the main VPW sheet. Do these two drop-down selections work for you?

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

ResearchMed
Posts: 4720
Joined: Fri Dec 26, 2008 11:25 pm

Re: Variable Percentage Withdrawal (VPW)

Post by ResearchMed » Thu Jun 12, 2014 8:58 am

longinvest wrote:From viewtopic.php?f=2&t=140976&p=2087603#p2087599
ResearchMed wrote:...
Next problem: I'm getting a message about
"Significant loss of functionality"
"One or more cells in this workbook contain data validation rules which refer to values on other worksheets. These data validation rules will not be saved. Location: 'VPW' "

Will this affect how the software "works"?

RM


There shouldn't be any dependency on other worksheets. Something has gone wrong if you get this message. I will look at it and, if necessary, I will completely rebuild the spreadsheet.

Thanks for the report.


Thanks very much.

I've just played around with it, and it *seems* to be working just fine :happy

In case it helps, back in about Dec, 2013, I was able to download the file using the link in the "red bar".
(I found the old file I had downloaded in a backup drive, in hopes it would help with the date.)

RM

longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Thu Jun 12, 2014 9:54 am

freebeer wrote:...
But the average person wants to look at retirement from the perspective of the minimum savings necessary to provide "consumption smoothing" in both working years and retirement years, including the ability to retire as soon as feasible and reasonable predictability in spending. It is not clear to me that the particular algorithm of VPW is an optimal way to meet these constraints.

I'm also confess to being skeptical any time someone tries to label one particular approach to something as "the" thing, especially without comparison of how that particular approach compares to other similar approaches. There are many variable percentage withdrawal strategies ranging from Taylor Larimore's pragmatic approach to more complex algorithms. Wade Pfau has written a bit about it here: http://wpfau.blogspot.com/2012/04/varia ... ement.html but I would love to see more of this type of analysis.


For one thing, I do not claim that VPW is "the" thing. I've actually repeated many times, on this thread, that VPW is just a tool that one can use within an overall retirement plan.

Dr. Pfau's blog post is very interesting. I've just taken three VPW backtesting samples that match his Good, OK, and Bad investment outcome scenarios. Please note that my samples distribute the withdrawals over a 35-year retirement instead of Dr. Pfau's 30 years. Using 30 years would make my numbers nicer, but I like to play it safe by assuming a 65 to 100 years-old retirement.

Good investment outcome (retiring in 1982):

Image
Image

OK investment outcome (retiring in 1950):

Image
Image

Bad investment outcome (retiring in 1966):

Image
Image

A few notes:

The 1982 retirement is not yet finished. Its target depletion year is 2017.

I so wish to be as lucky as to retire in a 1982-like year. :moneybag But, hey, any normal year, like 1950, is not too bad, either.

The most interesting is the bad outcome (1966), of course. Looking at the actual withdrawal sequence is very interesting:

Code: Select all

$48 405
$46 458
$48 867
$50 680
$47 278
$50 391
$53 722
$57 418
$52 551
$47 371
$56 099
$59 556
$56 191
$59 865
$64 224
$71 957
$70 694
$87 123
$95 484
$101 393
$118 666
$141 076
$136 190
$147 831
$167 733
$168 515
$197 410
$204 506
$216 832
$209 260
$252 184
$279 812
$321 047
$371 971
$384 670

Notice how withdrawals (in nominal dollars) almost never get down. Yes, there was some high inflation in the 1970s, but having a balanced portfolio delivered its promise of stable withdrawals. As inflation is something that we experience differently (owner vs renter, cost of oil if you don't use your car much, etc.), it is quite possible that the 1966 retiree that would have had used VPW would not have suffered too much from the loss of purchasing power during a few years. Note that the inflation line, in the Withdrawals graph, is anchored on a constant $48,405 (the start withdrawal amount), not on the usual $40,000.

Overall, the median withdrawal of the 1966 retiree would have been a little over a constant $40,000 over a 35 year period. That does compare quite favorably to the failure of a 4% SWR to deliver a 30 year retirement to a 1966 retiree.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
Cut-Throat
Posts: 2011
Joined: Sun Oct 17, 2010 9:46 am

Re: Variable Percentage Withdrawal (VPW)

Post by Cut-Throat » Thu Jun 12, 2014 10:37 am

freebeer wrote:
Cut-Throat wrote:
ot1138 wrote:* You can't create a plan based on the number of years until you die. You don't know this with any certainty.



Sure you Can !

You don't have to be completely Certain about this. If you pick age 100 you'll do Fine...
And if you use VPW..... You'll never run out of money!


These points are certainly true. To take a step further, if you pick age 120 you'll be even more certain, and if you always withdraw 3% of your current portfolio value you'll never run out of money. But in that case you will definitely be over-saving and you may or may not be spending enough to fund your obligations.



Except that VPW does not do this. It increases the Withdrawal percentage as you age.

So, I don't think you understand how VPW works. Why don't you download the spreadsheet and try it out.

User avatar
midareff
Posts: 4772
Joined: Mon Nov 29, 2010 10:43 am
Location: Biscayne Bay, South Florida

Re: Variable Percentage Withdrawal (VPW)

Post by midareff » Thu Jun 12, 2014 10:51 am

Longinvest............. I looked at the xls and plugged in my age (66) and a desire for a 40 year plan (younger wife). The first thin I noticed is the forecasting started using 1974, which produced 15 straight positive years, including the sequential double digit returns of the 80's.

From that, unless I am missing something, I don't see a realistic set of dependable back cast data to go from in 2014.

What am I missing?

User avatar
Cut-Throat
Posts: 2011
Joined: Sun Oct 17, 2010 9:46 am

Re: Variable Percentage Withdrawal (VPW)

Post by Cut-Throat » Thu Jun 12, 2014 10:58 am

midareff wrote:Longinvest............. I looked at the xls and plugged in my age (66) and a desire for a 40 year plan (younger wife). The first thin I noticed is the forecasting started using 1974, which produced 15 straight positive years, including the sequential double digit returns of the 80's.

From that, unless I am missing something, I don't see a realistic set of dependable back cast data to go from in 2014.

What am I missing?


You are missing that there are two data sets. You are using Simba. Use Shiller and go back to 1871.

longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Thu Jun 12, 2014 11:07 am

midareff wrote:Longinvest............. I looked at the xls and plugged in my age (66) and a desire for a 40 year plan (younger wife). The first thin I noticed is the forecasting started using 1974, which produced 15 straight positive years, including the sequential double digit returns of the 80's.

From that, unless I am missing something, I don't see a realistic set of dependable back cast data to go from in 2014.

What am I missing?

Dear midareff,

Maybe you forgot to read the Instructions sheet.

You are most probably interested in the Retirement sheet, where you can enter your start retirement year and your actual portfolio balance on each January 1st to deduce the recommended withdrawal amount. This is an ongoing process, each year of retirement.

The VPW sheet allows you to backtest what would have happened in the past. It provides 3 historical data sets.

If you didn't realize it, the past is not the future. I do not possess a working crystal ball. But, backtesting allows you to gauge the riskiness of the proposed withdrawal method.

Please read the Frequently Asked Questions section of this post:
viewtopic.php?f=10&t=120430&p=1761580#p1761563

Regards,

longinvest
Last edited by longinvest on Fri Jun 13, 2014 9:08 am, 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: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Thu Jun 12, 2014 11:12 am

I forgot to mention:

You must set your Asset Allocation on the VPW sheet before using the Retirement sheet.

You can select the "Data Set" as well as the simulation "Start Year" to study what would have happened in different past retirement scenarios. The Shiller data set goes back to 1871, if that's what you are looking for.

Enjoy!
Last edited by longinvest on Thu Jun 12, 2014 11:16 am, 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
midareff
Posts: 4772
Joined: Mon Nov 29, 2010 10:43 am
Location: Biscayne Bay, South Florida

Re: Variable Percentage Withdrawal (VPW)

Post by midareff » Thu Jun 12, 2014 11:14 am

longinvest wrote:
midareff wrote:Longinvest............. I looked at the xls and plugged in my age (66) and a desire for a 40 year plan (younger wife). The first thin I noticed is the forecasting started using 1974, which produced 15 straight positive years, including the sequential double digit returns of the 80's.

From that, unless I am missing something, I don't see a realistic set of dependable back cast data to go from in 2014.

What am I missing?

Dear midareff,

Maybe you forgot to read the Instructions sheet.

You are most probably interested in the Retirement sheet, where you can enter your start retirement year and your actual portfolio balance on each January 1st to deduce the recommended withdrawal amount. This is an ongoing process, each year of retirement.

The VPW sheet allows you to backtest what would have happened in the past. It provides 3 historical data sets.

If you didn't realize it, the past is not the future. I do not possess a working cristal ball. But, backtesting allows you to gauge the riskiness of the proposed withdrawal method.

Please read the Frequently Asked Questions section of this post:
viewtopic.php?f=10&t=120430&p=1761580#p1761563

Regards,

longinvest



Got it and thanks....

While nothing is a cure all it should be a good guide, especially if one compares it to a CPI adjusted fixed starting withdrawal rate. LOL.. in all cases keep a new refrigerator box and a blanket handy.

longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Fri Jun 27, 2014 2:43 pm

Hi,

Prof. Shiller has not yet updated http://www.econ.yale.edu/~shiller/data/chapt26.xls for 2013 returns. This has lead me to question why I provide two distinct US data sets.

Why not provide, instead, a single US data set composed of Prof. Shiller's data for the period 1871-1971, and Simba's data for 1972 and later? I think that this would be much simpler. This would be consistent with the way the Simba data set is built, as it also uses different data sources for years when Vanguard was not yet offering total market bond/us/international funds.

What do you think?
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
Cut-Throat
Posts: 2011
Joined: Sun Oct 17, 2010 9:46 am

Re: Variable Percentage Withdrawal (VPW)

Post by Cut-Throat » Fri Jun 27, 2014 4:06 pm

longinvest wrote:Hi,

Prof. Shiller has not yet updated http://www.econ.yale.edu/~shiller/data/chapt26.xls for 2013 returns. This has lead me to question why I provide two distinct US data sets.

Why not provide, instead, a single US data set composed of Prof. Shiller's data for the period 1871-1971, and Simba's data for 1972 and later? I think that this would be much simpler. This would be consistent with the way the Simba data set is built, as it also uses different data sources for years when Vanguard was not yet offering total market bond/us/international funds.

What do you think?


I think it sounds Perfect ! Simpler is always better.

User avatar
LadyGeek
Site Admin
Posts: 39148
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Fri Jun 27, 2014 9:34 pm

I'm confused on the difference between VPW and Simba's backtesting spreadsheet models. Perhaps this can be viewed from a different perspective - Why doesn't Simba's spreadsheet utilize the Shiller data?

Perhaps I should ask this question in the Simba 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: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Fri Jun 27, 2014 9:44 pm

LadyGeek wrote:Perhaps I'm confused on the difference between VPW and Simba's backtesting spreadsheet models. Could the converse proposal work?
IOW, why doesn't Simba's spreadsheet utilize the Shiller data?

Perhaps I should ask this question in the Simba thread?

LadyGeek,

You're asking the right question, as usual. I would be more than happy delegating the whole data aggregation job, including fetching Canadian data (and why not other countries?) to the Simba team.

All I need is the return data for building 3-fund portfolios:
  • Domestic Stocks
  • International Stocks
  • Domestic Bonds
  • Inflation

My preference is for real-world Index Fund returns, when available.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sat Jun 28, 2014 3:28 pm

I have uploaded a new version of the VPW spreadsheet on the Bogleheads Wiki VPW Page.

Changes:
  • Simplified the spreadsheet: Merged the Simba and Shiller U.S. data sets into a single U.S. data set.

Enjoy!
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: 39148
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sat Jun 28, 2014 4:02 pm

All OK with MS Excel. I tried a few extreme test cases, nothing broke.

In the Returns tab row 149, the formula is configured to use Shiller's data (blank - to be a future entry for 2014). I didn't know which data set will be used for 2014 - perhaps that is not yet decided.
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: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sat Jun 28, 2014 4:16 pm

LadyGeek wrote:All OK with MS Excel. I tried a few extreme test cases, nothing broke.

Thanks!
In the Returns tab row 149, the formula is configured to use Shiller's data (blank - to be a future entry for 2014). I didn't know which data set will be used for 2014 - perhaps that is not yet decided.

Good catch. That's because I merged the data sets by dragging the Simba returns over the Shiller columns. I forgot to handle the remaining rows.

I've just uploaded a fixed version.
Last edited by longinvest on Sat Jun 28, 2014 4:59 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
LadyGeek
Site Admin
Posts: 39148
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Sat Jun 28, 2014 4:29 pm

Looks OK. I'm also testing the spreadsheet by putting in data that doesn't make any sense (checking for user entry errors).

For example: Putting in a depletion years of 100 with 100% equities (75% domestic, 25% international), start at 1996. You run out of data, but the VPW table shows the full range of withdrawal.

Nothing breaks, but you can see that something doesn't line up. Should you put in error checking? Probably not, as you'll come up with quite a number of "what if" cases and make the coding all that much harder to implement.
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: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun Jun 29, 2014 1:04 pm

How about VPW and Japan?

I'm curious. I've decided to go ahead and play with the Japan return data found in the following post: viewtopic.php?f=10&t=23036&p=1979403#p1971254 . Using basic mathematical formulas, I was able to recover the required nominal domestic and international stock returns, nominal bond returns, and domestic inflation.

I inserted this data in the VPW spreadsheet and was able to get the following simulation of the default portfolio (50/50, with half international stocks). Using a 35-year depletion starting in 1990 (25 withdrawals, out of 35, so far) gives a median inflation-adjusted withdrawal of $39,490, and a minimum withdrawal of $29,667.

Here's a screenshot:
Image

I hesitate to publish the spreadsheet before getting the OK from Bogleheads authorities, as I have copied this data from the forum (and before asking bpp, the author of the post).

Added: It seems that there might be restrictions on part of the data. So, I can't publish it. :( Hopefully, one day, the Simba spreadsheet team will start collecting international data that I'll be able to use in the VPW spreadsheet.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
siamond
Posts: 3154
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Mon Jun 30, 2014 11:09 am

longinvest wrote:Hi,

Prof. Shiller has not yet updated http://www.econ.yale.edu/~shiller/data/chapt26.xls for 2013 returns. This has lead me to question why I provide two distinct US data sets.

Why not provide, instead, a single US data set composed of Prof. Shiller's data for the period 1871-1971, and Simba's data for 1972 and later? I think that this would be much simpler. This would be consistent with the way the Simba data set is built, as it also uses different data sources for years when Vanguard was not yet offering total market bond/us/international funds.

What do you think?


LadyGeek wrote:I'm confused on the difference between VPW and Simba's backtesting spreadsheet models. Perhaps this can be viewed from a different perspective - Why doesn't Simba's spreadsheet utilize the Shiller data?

Perhaps I should ask this question in the Simba thread?


Totally agreed. I've used exactly such combo of input data in my own Excel sheets for quite a while now.

The only difficulty is when trying to model portfolios with more asset classes. The Fama-French data is very helpful for small-value and the likes back to 1926, but then other asset classes (e.g. REITs, emerging countries) are more tricky for pre-1974 returns, and I ended up approximating them to the nearest broad asset class (anyway, investors in the past may not have had access to such index funds!).

And yes, it would be a good project for the Simba sheet, we actually discussed something similar in the past, and didn't proceed. Maybe one day, I'll get to it...

Near retirement
Posts: 114
Joined: Mon Oct 08, 2012 4:57 pm

Re: Variable Percentage Withdrawal (VPW)

Post by Near retirement » Mon Jun 30, 2014 3:34 pm

Will the software download to an iPad?

User avatar
LadyGeek
Site Admin
Posts: 39148
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Mon Jun 30, 2014 6:45 pm

The spreadsheet will work in both LibreOffice Calc (available for Mac OS X) and Microsoft Excel. I understand that Microsoft Excel for iPad is available, but I don't know anything about it.
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.

Beliavsky
Posts: 688
Joined: Sun Jun 29, 2014 10:21 am

Re: Variable Percentage Withdrawal (VPW)

Post by Beliavsky » Thu Jul 10, 2014 11:50 am

longinvest wrote:So, does VPW make sense?


I do think VPW makes sense. It has you withdrawing a percentage of assets each year, with the percentage depending on life expectancy. The commonly mentioned alternative of withdrawing a fixed inflation-adjusted dollar amount each year ignores how your portfolio has done and can cause you to run out of money.

The fixed spending strategy, although flawed, does maintain the retiree's standard of living, until the money runs out! With VPW or fixed percentage withdrawal, a 20% rise/fall in your portfolio tells you to spend 20% more/less. But people may not be able to cut their rent/mortgage, utility bills, and prescription spending by 20%, and they may not want to cut discretionary spending on vacations and club memberships that much. It is human nature to become accustomed to a certain level of spending.

To reduce the consumption volatility of the VPW rule, you could set spending at the lesser of
(1) the VPW amount
(2) c * last year's real spending -- maybe c = 1.1

So when your portfolio does really well, you increase your real spending by only 10%, not creating expensive habits that will be painful to break. If the markets have done very well and amount (2) is 20% less than amount (1), you have created a buffer such that a 20% fall in your portfolio does not force you to cut spending. "C" is a free parameter, and if you make it very large, you get back to the VPW rule. If you make it 1.0, you never let real spending rise, regardless of portfolio performance, which I think it too austere.

User avatar
Cut-Throat
Posts: 2011
Joined: Sun Oct 17, 2010 9:46 am

Re: Variable Percentage Withdrawal (VPW)

Post by Cut-Throat » Thu Jul 10, 2014 11:58 am

Beliavsky wrote:I do think VPW makes sense. It has you withdrawing a percentage of assets each year, with the percentage depending on life expectancy. The commonly mentioned alternative of withdrawing a fixed inflation-adjusted dollar amount each year ignores how your portfolio has done and can cause you to run out of money.

The fixed spending strategy, although flawed, does maintain the retiree's standard of living, until the money runs out! With VPW or fixed percentage withdrawal, a 20% rise/fall in your portfolio tells you to spend 20% more/less. But people may not be able to cut their rent/mortgage, utility bills, and prescription spending by 20%, and they may not want to cut discretionary spending on vacations and club memberships that much. It is human nature to become accustomed to a certain level of spending.

To reduce the consumption volatility of the VPW rule, you could set spending at the lesser of
(1) the VPW amount
(2) c * last year's real spending -- maybe c = 1.1

So when your portfolio does really well, you increase your real spending by only 10%, not creating expensive habits that will be painful to break. If the markets have done very well and amount (2) is 20% less than amount (1), you have created a buffer such that a 20% fall in your portfolio does not force you to cut spending. "C" is a free parameter, and if you make it very large, you get back to the VPW rule. If you make it 1.0, you never let real spending rise, regardless of portfolio performance, which I think it too austere.


This has been suggested before, but then VPW falls very short of what it was trying to accomplish. The design goals were to be able to spend all of your money by the end of your plan, and to not fail along the way.

If you need to reduce volatility alter your asset allocation, buy annuities etc.

And as a side comment, if you cannot reduce spending by 20%, you probably should not be retired or you need to reduce your standard of living so that you are able to reduce spending. This is no different than when we working. We did not spend all of our take home pay, but tried to save 30% or more for our retirement. When you finally retire you need as big a cushion on your withdrawal amount.

User avatar
siamond
Posts: 3154
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Fri Jul 11, 2014 9:59 am

I've been swapping a few PMs with Longinvest, and he suggested to share a few highlights.

First, I stumbled upon an old BH thread where the very exact same idea of using PMT down to depletion was discussed. This is an interesting read.
http://www.bogleheads.org/forum/viewtopic.php?t=41514

Next, like many others, the jittery trajectory of VPW's withdrawals (as a direct function of the portfolio balance jitters) doesn't quite work for me. And I really do not want to change my AA to address this issue. I played with a few ideas, and the one that seems to work the best is to compute the PMT formula as usual for each year, but then use as withdrawal goal the rolling average of the past few years (say 5). Such math has to be done in real dollars. This goes a long way in smoothing the trajectory of a sideways market, while not introducing issues with big bulls or bears. Somebody suggested the same thing on this old BH thread.

Next (and this fits well with the smoothing idea), I much prefer to have a lower cap on the # of remaining periods (i.e. 5 years before depletion, if you're still alive, you may want to foresee more than 5 years of life to come!). A lower cap of 10 years, or something like that, seems reasonable. This works better than simply extending the depletion period by 10 years.

Finally, if your personal goal is NOT to spend it all (e.g. leave a significant bequest, self-insure for LTC, etc), then an interesting idea could be to use the 'Final Value' of the PMT formula. It could be half of your initial portfolio, or even all of it, that is up to you. This also fits well with the smoothing and lower cap ideas. This one is a longer discussion, I modeled it, it works, but the outcome is more sensitive to the expected market returns vs reality than I originally expected. Anyhoo, this is an interesting avenue for investigation.

PS. to better understand my post, let me remind you that the VPW math for a given year is equal to PMT (expected-real-returns, years-periods-left, minus portfolio-balance, final-value, type), with FV=$0 and Type=1. Example: PMT (4%, 20, -$2M, $0, 1) for 20 years left to spend it all.

User avatar
Cut-Throat
Posts: 2011
Joined: Sun Oct 17, 2010 9:46 am

Re: Variable Percentage Withdrawal (VPW)

Post by Cut-Throat » Fri Jul 11, 2014 10:52 am

siamond wrote:I've been swapping a few PMs with Longinvest, and he suggested to share a few highlights.

First, I stumbled upon an old BH thread where the very exact same idea of using PMT down to depletion was discussed. This is an interesting read.
http://www.bogleheads.org/forum/viewtopic.php?t=41514

Next, like many others, the jittery trajectory of VPW's withdrawals (as a direct function of the portfolio balance jitters) doesn't quite work for me. And I really do not want to change my AA to address this issue. I played with a few ideas, and the one that seems to work the best is to compute the PMT formula as usual for each year, but then use as withdrawal goal the rolling average of the past few years (say 5). Such math has to be done in real dollars. This goes a long way in smoothing the trajectory of a sideways market, while not introducing issues with big bulls or bears. Somebody suggested the same thing on this old BH thread.

Next (and this fits well with the smoothing idea), I much prefer to have a lower cap on the # of remaining periods (i.e. 5 years before depletion, if you're still alive, you may want to foresee more than 5 years of life to come!). A lower cap of 10 years, or something like that, seems reasonable. This works better than simply extending the depletion period by 10 years.

Finally, if your personal goal is NOT to spend it all (e.g. leave a significant bequest, self-insure for LTC, etc), then an interesting idea could be to use the 'Final Value' of the PMT formula. It could be half of your initial portfolio, or even all of it, that is up to you. This also fits well with the smoothing and lower cap ideas. This one is a longer discussion, I modeled it, it works, but the outcome is more sensitive to the expected market returns vs reality than I originally expected. Anyhoo, this is an interesting avenue for investigation.

PS. to better understand my post, let me remind you that the VPW math for a given year is equal to PMT (expected-real-returns, years-periods-left, minus portfolio-balance, final-value, type), with FV=$0 and Type=1. Example: PMT (4%, 20, -$2M, $0, 1) for 20 years left to spend it all.


I hope longinvest leaves the VPW tool 'As is', it is a great tool and I hope the 'perfect' does not ruin the great. I think you should develop another tool using your approach with backtesting as well. The "jittery trajectory of VPW's withdrawals" is exactly why it is the perfect withdrawal tool for most Bogelheads. You don't need to anticipate a bad market (i.e. 2% SWRs), VPW puts the brakes on at exactly the right time. Most Bogelheads here save more than 30% of their take home pay. Why should their withdrawal amount be so inflexible at retirement? We don't live from paycheck to paycheck here. And in retirement you shouldn't live from withdrawal amount to withdrawal amount.

I worked in the software industry for 30 years and I have seen endless modifications that have ruined many an application. I'm sure Google struggles daily to maintain 1 input Box on a Single Screen.

User avatar
siamond
Posts: 3154
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Fri Jul 11, 2014 3:47 pm

Cut-throat, I am very glad that you found the perfect match for your own retirement goals. But fact is other people have other goals. Numerous people expressed strong reservations about the 'jitters' on this thread. And I would find hard to believe that numerous people wouldn't share my own goal to leave a significant bequest to my children and keep my portfolio value roughly unchanged on average over time (in real $) while benefitting from a proper income flow.

I worked in the Software and Telecom industry for 30 years... Google was a customer... And I know that a product satisfying only one customer or two ain't going to make it... :wink:

Let's just be open-minded and explore other options beyond the existing VPW. We may call it something else, or it may not materialize at all, doesn't really matter to me, I do have indeed my own Excel sheet with my own backtesting... But I thought it was appropriate to share some key ideas that helped me mold the key idea behind VPW to my own goals. Which I don't think are that special.

User avatar
Cut-Throat
Posts: 2011
Joined: Sun Oct 17, 2010 9:46 am

Re: Variable Percentage Withdrawal (VPW)

Post by Cut-Throat » Fri Jul 11, 2014 4:58 pm

siamond wrote: And I know that a product satisfying only one customer or two ain't going to make it...


If this is an implication that VPW would only satisfy 1 or 2 people, I completely disagree.

There is nothing stopping you from coming up with your own tool, starting another thread and letting people download it, as longinvest has. The changes that you propose don't have to be implemented in VPW.

In the end, it will be up to Longinvest to decide, but I think we covered this ground before.

longinvest
Posts: 2391
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Fri Jul 11, 2014 8:31 pm

siamond wrote:I've been swapping a few PMs with Longinvest, and he suggested to share a few highlights.

First, I stumbled upon an old BH thread where the very exact same idea of using PMT down to depletion was discussed. This is an interesting read.
http://www.bogleheads.org/forum/viewtopic.php?t=41514


Thank you, Siamond, for uncovering this prior art. I really enjoyed reading that thread.

Next, like many others, the jittery trajectory of VPW's withdrawals (as a direct function of the portfolio balance jitters) doesn't quite work for me. And I really do not want to change my AA to address this issue.


I think that not changing AA is one of the main points that leads you to want to modify VPW. (I'll come back to this later). I propose that this is not a VPW issue; it is a Siamond issue with using VPW as-is. :wink:

So, let me restate what you wrote differently: VPW will only deliver a jittery trajectory if you keep an inappropriate withdrawal-time AA.

Just for the benefit of readers of this thread: I'll illustrate how VPW does actually deliver very smooth withdrawals, when an appropriate AA is selected. I'll show graphs of VPW withdrawals for two retirement backtest simulations using Vanguard Target Retirement's 30% stocks / 70% bonds AA (with stocks: 70% US / 30% International) for a retired person:
  1. A 35-year retirement starting in 1966; that's the retirement year for which a traditional 4% SWR fails.
  2. An incomplete (35-year) retirement starting in 2000; it contains two big stock market drawdowns in 2000 and 2008.

I used a logarithmic scale to make sure not to hide any volatility in smaller withdrawals, in early retirement.

Here's the 1966 withdrawal path:
Image

The biggest drop from 1973 to 1975 was approximately 10%.

And here's the 2000 partial withdrawal path:
Image

In 2009, after experiencing the horrible 2008 year for stocks, the withdrawal was down something like 10%.

As you can see, VPW is not jittery when you have an appropriate AA.

I played with a few ideas, and the one that seems to work the best is to compute the PMT formula as usual for each year, but then use as withdrawal goal the rolling average of the past few years (say 5). Such math has to be done in real dollars. This goes a long way in smoothing the trajectory of a sideways market, while not introducing issues with big bulls or bears. Somebody suggested the same thing on this old BH thread.

Next (and this fits well with the smoothing idea), I much prefer to have a lower cap on the # of remaining periods (i.e. 5 years before depletion, if you're still alive, you may want to foresee more than 5 years of life to come!). A lower cap of 10 years, or something like that, seems reasonable. This works better than simply extending the depletion period by 10 years.

Finally, if your personal goal is NOT to spend it all (e.g. leave a significant bequest, self-insure for LTC, etc), then an interesting idea could be to use the 'Final Value' of the PMT formula. It could be half of your initial portfolio, or even all of it, that is up to you. This also fits well with the smoothing and lower cap ideas. This one is a longer discussion, I modeled it, it works, but the outcome is more sensitive to the expected market returns vs reality than I originally expected. Anyhoo, this is an interesting avenue for investigation.


Siamond, let me try to summarize what you are seeking, just so that you can correct me if I am wrong. You're looking for a withdrawal method:
  • that can deliver smooth withdrawals regardless of AA,
  • that can continue to deliver sufficient withdrawals, in case you live longer than anticipated,
  • that lets you specify a target bequest.

Is this what you seek? If yes, I will start a new thread with a different proposal that addresses such constraints using some (hopefully nice) new ideas that I got! (It's actually different from VPW, due to your hard constraints).


Just for the benefit of readers of this thread:

I have already shown that VPW can deliver smooth withdrawals.

VPW can also deliver withdrawals for a longer time; just select a longer initial depletion period. (By default, VPW uses a 35-year depletion period for a 65-year old person). You should also revise your plan every few years and set up a new (longer) depletion period if you think that your life expectancy is higher than you previously thought. Just make it longer than your best estimate.

As for leaving a bequest, that must be done separately from VPW. VPW will, of course, leave bequest as long as you die before the end of the depletion period. That is actually expected, otherwise, you failed to revise your plan every few years! If you end-up broke at 120 years-old, it is because you misused VPW. You really think you can make it to 120? Just select a depletion period that ends at 125 or 130. Just let me know when you get there! :wink:

But, let's be serious. Let say you have $3,000,000 (for VPW and bequest) and you'd like to leave a $1,000,000 (constant dollar) bequest. One sure way to do so is to buy $1,000,000 of TIPS. Then, you can apply VPW on the remaining $2,000,000 and add the TIPS interest payments to your VPW withdrawals. Or pick any variation on this.

In other words, I think that VPW:
  • Is good at doing what it was meant to do: To deplete a portfolio over a time-certain period.
  • Is not a complete retirement plan in itself.
  • Is not a comprehensive retirement calculator that integrates SS, pensions, etc.
  • Is very simple to use.
  • Provides a nice backtesting framework in its spreadsheet.
  • Is fairly easy to recreate from scratch using a piece of paper, a pen, and a simple financial calculator.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VAB/ZRR

User avatar
siamond
Posts: 3154
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Sun Jul 13, 2014 8:07 am

longinvest wrote:As you can see, VPW is not jittery when you have an appropriate AA.

Some retirees may choose a 30/70 allocation, others would be very nervous about the consequences of doing so, and would hardly see it as an appropriate AA... There has to be a better way to address the jitters - which are truly a side-effect of the withdrawal formula of choice (here PMT). I suggested one way, I have another one in mind, let's see what other idea you have in your creative mind. I would strongly suggest that you look at graphs in real $ when exploring such options though, this shows the issues much more clearly.

longinvest wrote:Siamond, let me try to summarize what you are seeking, just so that you can correct me if I am wrong. You're looking for a withdrawal method:
  • that can deliver smooth withdrawals regardless of AA,
  • that can continue to deliver sufficient withdrawals, in case you live longer than anticipated,
  • that lets you specify a target bequest.

Is this what you seek? If yes, I will start a new thread with a different proposal that addresses such constraints using some (hopefully nice) new ideas that I got! (It's actually different from VPW, due to your hard constraints).

Well, I would rephrase.

1. The primary point is really to aim at keeping an average trajectory of the portfolio centered on a predefined & sustainable value (in real $) while extracting (variable & optimal) income through withdrawals.
=> This addresses longevity risk, this addresses early retirement (hence -hopefully- long retirement), this addresses the desire for a bequest, this addresses self-insured LTC, this addresses the desire to plan for the worst as well as plan for the best (and everything in-between), etc. Simply focusing on the bequest isn't quite the point. And viewing it as a separate bucket of sorts is actually CONTRARY to my goals. No buckets! :wink:

2. Then yes, get variable yet reasonably smooth withdrawals - again while planning for the worst as well as planning for the best
=> Note that short-term smoothing is easy (the average technique I described does it). What is trickier is mid-term smoothing (avoid unnecessary humps).

3. And indeed not imply undue constraints on the AA in the process.

Sounds to me those should be fairly common goals. Taking maximum advantage of a sustainable portfolio, plus limited jitters on withdrawals, is really the bottomline. Those could be viewed as two separate problems solved by their own separate devices, actually. Or maybe kill two birds with one stone. I'm all ears...
Last edited by siamond on Sun Jul 13, 2014 8:59 am, edited 3 times in total.

User avatar
Cut-Throat
Posts: 2011
Joined: Sun Oct 17, 2010 9:46 am

Re: Variable Percentage Withdrawal (VPW)

Post by Cut-Throat » Sun Jul 13, 2014 8:44 am

siamond wrote:Sounds to me those should be fairly common goals. A sustainable portfolio and limited jitters is really the bottomline. Those could be viewed as two separate problems solved by their own separate devices, actually. Or maybe kill two birds with one stone. I'm all ears...


What made VPW different from the previous withdrawal methods was the following design criteria.
1.) It cannot fail
2.) It consumes all of your money at the end of your plan.
3.) It backtests against past history, so that you can see if you can live with your asset allocation and associated Volatility.

While 'Limited Jitters' is up to the individual to decide, you can't push this too far. VPW - Does mean VARIABLE. (That's part of the Objective!) While a Perpetual Motion Machine is easy to understand, it flies in the face of Physics.

What you seek has completely different goals than VPW. Not that they're bad, just completely different. I think this has been tackled quite a few times, with withdrawal methods such as 'Floor and ceiling' etc. Have you looked at other withdrawal methods that are probably closer to your goals? I think this one is a better place to start your desired withdrawal method.

http://www.cbsnews.com/news/vanguard-proposes-flexible-way-to-determine-retirement-paycheck/

User avatar
siamond
Posts: 3154
Joined: Mon May 28, 2012 5:50 am

Re: Variable Percentage Withdrawal (VPW)

Post by siamond » Sun Jul 13, 2014 8:57 am

I slightly rephrased my bottomline as you made me realize this wasn't properly worded.

Oh yes, I've looked at quite a few withdrawal methods, including the Vanguard article... I do believe in the virtues of variable methods. But I also believe there is a middle ground between extreme variability and extreme stability. Guyton-Klinger found a pretty nice middle-ground, but has some drawbacks. A PMT-based method has a lot of advantages though, and what I described in my earlier post comes pretty close to my goals... Still I am curious about longinvest new ideas.

Beliavsky
Posts: 688
Joined: Sun Jun 29, 2014 10:21 am

Re: Variable Percentage Withdrawal (VPW)

Post by Beliavsky » Sun Jul 13, 2014 10:42 am

Strategies that withdraw x% of the portfolio in the first year and increase withdrawals by inflation in later years are commonly evaluated based on the fraction of time money runs out over the retirement time horizon, say 30 years. If you never withdraw more than a fraction of your portfolio, you cannot run out of money unless your portfolio has a -100% return. A different metric is needed to evaluate such strategies. With VPW the withdrawals may be insufficient to maintain the desired lifestyle.

You could define a utility function for annual consumption and look for a withdrawal strategy that maximizes the sum of the annual utilities. Saying that utility is proportional to the log of consumption is plausible to me. Given the expected real return and volatility of your portfolio, and given a life expectancy table, you could calculate whether consuming 4, 5, or 6% annually of the portfolio maximizes expected cumulative utility. If year-to-year volatility in consumption reduces utility, a penalty term could be added penalizing decreases in annual consumption.

User avatar
Cut-Throat
Posts: 2011
Joined: Sun Oct 17, 2010 9:46 am

Re: Variable Percentage Withdrawal (VPW)

Post by Cut-Throat » Sun Jul 13, 2014 11:04 am

Beliavsky wrote:Strategies that withdraw x% of the portfolio in the first year and increase withdrawals by inflation in later years are commonly evaluated based on the fraction of time money runs out over the retirement time horizon, say 30 years. If you never withdraw more than a fraction of your portfolio, you cannot run out of money unless your portfolio has a -100% return. A different metric is needed to evaluate such strategies. With VPW the withdrawals may be insufficient to maintain the desired lifestyle.


The metric that is used to evaluate VPW is backtesting against prior Market History. That is all we have to work with as no one has developed a forward looking tool.

Have you downloaded the VPW tool and backtested your portfolio and Asset Allocation against history to see if your "Withdrawals may be insufficient to maintain your desired lifestyle"?

If your VPW "withdrawal is insufficient to maintain your desired lifestyle", it is NOT VPW's problem.
The Problem is one of the following:
1.) Your portfolio is too small to begin with.
2.) Your Asset Allocation is incorrect for what you are trying to do.
3.) Your desired Lifestyle is unrealistic for the size of your portfolio.

apog
Posts: 29
Joined: Fri Sep 05, 2008 9:13 pm

Re: Variable Percentage Withdrawal (VPW)

Post by apog » Tue Jul 15, 2014 10:47 am

To longinvest: Thanks! for all of the work regarding VPW. Steady state solutions to a dynamic problem always bug me, but I'm neither imaginative nor knowledgeable enough to develop an approach like yours. The simplicity (ie multiply balance by up to max of VPW percentage) & 'never fail' are very attractive to me and the higher VPW vs SWR amount sure would be FUN.

Haven't committed to VPW yet, but we (age 59 for me, 57 for wife) are seriously considering implementing under this scenario beginning next year:
Have set aside 6 years worth of expected health care insurance / expenses taking me to 65 when Medicare kicks in. This amount is excluded from VPW balance. Wife's social security should take care of health care
expenses thereafter.

My SS at 70 is $40K. Expect to be taxed at 25% so $30K will be ours to spend. Would set aside $330K of taxable to supplement withdrawals by $30k / year. This bridging amount is problematic in that only 1/3 the amount is fixed income while the other 2/3 consists of index funds so we're subject to market declines. Originally had planned this because it reduced needed SWR to around 3%, but it also functions as a nice smoother for VPW during portfolio down years. This amount is excluded from VPW balance.

We include significant cost items like fence, HVAC, roof, electronics, auto, flooring etc replacement in our budget and set aside dollars into a checking account. This amount is excluded from VPW balance.

Remaining $2.2m would be withdrawn using up to VPW percentages assuming wife departure age of 105. Portfolio would continue to be 60% stock & 40% bond. Withdrawal variability backtesting with years like 1966 produce acceptable results for us (ie might take just one trip in bad years).

Have really enjoyed this thread. Thanks once again for the ideas and spreadsheet.

User avatar
Cut-Throat
Posts: 2011
Joined: Sun Oct 17, 2010 9:46 am

Re: Variable Percentage Withdrawal (VPW)

Post by Cut-Throat » Tue Jul 15, 2014 3:40 pm

apog wrote: Portfolio would continue to be 60% stock & 40% bond. Withdrawal variability backtesting with years like 1966 produce acceptable results for us (ie might take just one trip in bad years).


I am in a similar position to yours, although I have implemented VPW this year.

Can I ask why you would be still 60% Stock in your retired years? It sounds like you have enough ???? --- So, why risk it? I am 30% Stock/70 Bond..... At this point just trying to not lose $$$. I feel I have won the game, so why keep playing.

You mentioned if the 1966 scenario played out, you would only take 1 trip in a bad year. You could adjust your asset allocation and pretty much keep doing what you're doing. IOW - What are you trying to accomplish with 60% stocks? One of the goals of VPW is to spend your $$$, so you don't leave a pile on the table.

Post Reply