Variable Percentage Withdrawal (VPW)

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills
longinvest
Posts: 2905
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Wed Mar 21, 2018 1:25 pm

This topic is already very long with over 700 posts. I think that it would be best to hold ADD related discussions in its own topic: Accumulation-Dynamic Decumulation Strategy
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VLB/ZRR

User avatar
LadyGeek
Site Admin
Posts: 46697
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Wed Mar 21, 2018 1:58 pm

^^^ I have moved the relevant posts to Accumulation-Dynamic Decumulation Strategy.
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: 2905
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Mon May 21, 2018 7:01 am

In another topic, I explained (yet again) that it would be a bad idea to use valuations to determine VPW's withdrawal rates:
longinvest wrote:
Sun May 20, 2018 7:00 pm
DaufuskieNate wrote:
Sun May 20, 2018 6:35 pm
There is no reason VPW can't use valuation metrics to determine initial withdrawal rates. Returns are a variable that can be defined in the VPW model.
There's a good reason not to use valuation metrics to determine initial withdrawal rates; using valuations to set withdrawal rates would lead to a "sell less when high, sell more when low" behavior. That would be a pretty bad investing approach.

The VPW table doesn't change with "valuation metrics".

The volatility of total retirement income (e.g. portfolio withdrawals + stable lifelong non-portfolio income) is best controlled by the ratio of stable lifelong non-portfolio income and the ratio of stocks and bonds in the portfolio. It's futile to try predicting future long-term returns using simplistic metrics. It's best to accept that we don't know future returns and plan accordingly.
I also explained that VPW doesn't attempt to predict future returns; that its fixed internal growth trend aims to simply be lower than high returns and higher than low returns:
longinvest wrote:
Sun May 20, 2018 11:53 pm
AlohaJoe wrote:
Sun May 20, 2018 9:23 pm
VPW assumes that future stock returns will be 5% real and bonds will return 1.8% real, so of course it will pull a lot more.
This is false. VPW doesn't assume that future stocks returns will be 5% real and future bonds returns will be 1.8% real.

VPW needs fixed internal growth trends for stocks and bonds to avoid "sell less when high, sell more when low" behavior.

A way to understand a growth trend is to view it as a growth rate that is lower than high returns and higher than low returns. In other words, when stocks have just crashed, their future growth should likely be higher than the stocks growth trend. Similarly, when stocks are in a bubble, their future growth should likely be lower than the stocks growth trend. A growth trend is just an imprecise wild-ass guess (WAG). Why 5% real and 1.8% real? Because nobody was able to provide me with better WAGs than using long-term historical world-wide stocks and bonds returns since 1900.
Finally, I explained how ignoring valuations and simply applying VPW on a balanced portfolio leads to good investing behavior:
longinvest wrote:
Mon May 21, 2018 12:25 am
The best time to sell stocks is when they are in a bubble, and the worst time is just after they've crashed. Trying to smooth withdrawals by selling less (lower percentage) when stocks are expensive during a bubble is equivalent to invest the non-withdrawn money into bubbly stocks. After the crash, this additional money will have melted along with stocks. Withdrawing more (higher percentage) after a crash hurts a portfolio at the worst time.

One robust way to protect money, in a stocks bubble, is to put it into bonds (of average short or intermediate duration, to limit volatility). Doing this perfectly is a problem, though; it would require knowing when stocks are in a bubble; unfortunately we only know after the fact.

Fortunately, the Bogleheads investment philosophy provides a solution. It suggests to never try timing the market, but to adopt a strategic asset allocation instead (like a 50/50 stocks/bonds allocation). This results in a good-enough portfolio and leads to sell more stocks when they are "expensive", and sell less stocks when they are "cheap" when rebalancing to keep the portfolio's asset allocation on target (when using VPW with a fixed growth trend, or when using the constant-percentage withdrawal method).
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VLB/ZRR

2015
Posts: 1690
Joined: Mon Feb 10, 2014 2:32 pm

Re: Variable Percentage Withdrawal (VPW)

Post by 2015 » Mon May 21, 2018 1:41 pm

longinvest, your ongoing patience in providing thorough explanations regarding VPW are most appreciated, and remain among one of the more useful and robust tools I've acquired from this forum. I've positioned myself such that I can play the investing metagame of being able to ignore valuations talk of all kinds (hat tips to Warren Buffet, Jeff Bezos, among others).

jaj2276
Posts: 350
Joined: Sat Apr 16, 2011 5:13 pm

Re: Variable Percentage Withdrawal (VPW)

Post by jaj2276 » Thu May 24, 2018 9:01 am

Is there an updated spreadsheet with 2016/2017 returns? I feel awful asking for it because the spreadsheet has been soooo useful to me already (I haven't used it for retirement but to help me with my donor advisor fund).

icetrap
Posts: 65
Joined: Mon Nov 06, 2017 12:30 pm

Re: Variable Percentage Withdrawal (VPW)

Post by icetrap » Fri May 25, 2018 7:45 am

jaj2276 wrote:
Thu May 24, 2018 9:01 am
Is there an updated spreadsheet with 2016/2017 returns? I feel awful asking for it because the spreadsheet has been soooo useful to me already (I haven't used it for retirement but to help me with my donor advisor fund).
I've partially updated the returns for 2016-2017 in the accumulation-dynamic decumulation spreadsheet (only the US rates update).

You can get the new rates directly from Vanguard using the following two links and click display annually. Use the "Total return by NAV" :
US stock proxy : VTSMX

International stock proxy : VGTSX

US bond proxy : VBMFX

The advantage here compare to using directly the index is that it reflects the management cost that you would not be able to escape. Depending on your need for the donor advisor fund, you may want to be careful to disclose that a management fee is already included.

I hope this helps

User avatar
LadyGeek
Site Admin
Posts: 46697
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Variable Percentage Withdrawal (VPW)

Post by LadyGeek » Fri May 25, 2018 7:54 am

icetrap wrote:
Fri May 25, 2018 7:45 am
I've partially updated the returns for 2016-2017 in the accumulation-dynamic decumulation spreadsheet (only the US rates update).
Here is the wiki link: Accumulation-dynamic decumulation
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: 2905
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Fri May 25, 2018 8:00 am

I'll be posting an updated spreadsheet soon, hopefully this weekend.

VPW's backtesting spreadsheet already uses real-fund (after-fee) returns for year when such information is available (e.g. for data extracted from the Simba spreadsheet including VTSMX, VBMFX, and VGTSX actual returns).

All data sources (sometimes indirect, e.g. Simba) are provided in the spreadsheet.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VLB/ZRR

longinvest
Posts: 2905
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Sun May 27, 2018 8:29 am

I have uploaded a new version of the VPW backtesting spreadsheet.

As usual, you can find the download links on the Bogleheads wiki Variable percentage withdrawal page.

Changes:
  • Added 2016 return data.
  • Added 2017 return data.
  • Started directly collecting US historical returns and inflation data (for 2015 and later) based on NAV returns of Vanguard ETFs (VTI, VXUS, and BND) and the Bureau of Labor Statistics historical CPI-U data.
Nothing was changed in the presentation.

As usual comments are welcome.

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

jaj2276
Posts: 350
Joined: Sat Apr 16, 2011 5:13 pm

Re: Variable Percentage Withdrawal (VPW)

Post by jaj2276 » Mon May 28, 2018 9:27 am

longinvest wrote:
Sun May 27, 2018 8:29 am

[*]Started directly collecting US historical returns and inflation data (for 2015 and later) based on NAV returns of Vanguard ETFs (VTI, VXUS, and BND) and the Bureau of Labor Statistics historical CPI-U data.[/list]
What is the reason for this (and why start in 2015)?

longinvest
Posts: 2905
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Mon May 28, 2018 11:01 am

Jaj2276,
jaj2276 wrote:
Mon May 28, 2018 9:27 am
longinvest wrote:
Sun May 27, 2018 8:29 am

[*]Started directly collecting US historical returns and inflation data (for 2015 and later) based on NAV returns of Vanguard ETFs (VTI, VXUS, and BND) and the Bureau of Labor Statistics historical CPI-U data.[/list]
What is the reason for this (and why start in 2015)?
This is a good question. To answer, I have to provide some context.

In July 2014, Vanguard Canada launched the first all-in-one Ex-Canada Stock ETF (VXC). The historical data source I had been using, until then, to fill the Canadian data of the VPW spreadsheet provides historical returns for US, EAFE, and emerging markets separately (no all-in-one "international" returns). Note, too, it only provided index returns, not real-life fund or ETF returns.

I've decided to start collecting myself real-life ETF returns for the three assets (Canadian stocks, Ex-Canada stocks, and Canadian bonds) using the returns of the appropriate total-market index ETFs offered by Vanguard Canada (VCN, VXC, and VAB). As 2015 is the first full-year in which all three ETFs existed, it is the year for which I started collecting the data myself.

Now, as I'm already doing the collection of Canadian return data based on 3 ETFs, I've decided to do the same with US data based on the three appropriate Vanguard ETFs (VTI, VXUS, and BND). It's very easy to do and it reduces the dependence of the VPW backtesting spreadsheet on indirect data sources for recent data. This is more transparent for users who want to double-check the data. I could try to go back a few additional years, as long as the three US ETFs existed. I might do that in a future version. For now, I just started with the same year as the Canadian data collection.
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VLB/ZRR

cap396
Posts: 17
Joined: Sun Oct 14, 2012 4:05 pm

Re: Variable Percentage Withdrawal (VPW)

Post by cap396 » Tue Jul 17, 2018 9:42 am

Question: On the "Table" tab of the VPW document, the Long-Term Real Growth Trends is set to 5.0% for stocks and 1.8% for bonds, and there is a note that this is based on data from Credit Suisse Global Investment Returns Yearbook 2016. When I look at the Credit Suisse Global Investment Returns Yearbook 2018, the numbers have been updated to 5.2% and 2.0% respectively. Should those values be changed on the "Table" tab? Doing this changes the withdrawal rates (slightly).

Sorry if I'm totally off here; I'm still learning how all of this works.

(I'm looking at page 37 of this document: https://www.credit-suisse.com/media/ass ... y-2018.pdf)

longinvest
Posts: 2905
Joined: Sat Aug 11, 2012 8:44 am

Re: Variable Percentage Withdrawal (VPW)

Post by longinvest » Fri Jul 20, 2018 8:43 pm

cap396 wrote:
Tue Jul 17, 2018 9:42 am
Question: On the "Table" tab of the VPW document, the Long-Term Real Growth Trends is set to 5.0% for stocks and 1.8% for bonds, and there is a note that this is based on data from Credit Suisse Global Investment Returns Yearbook 2016. When I look at the Credit Suisse Global Investment Returns Yearbook 2018, the numbers have been updated to 5.2% and 2.0% respectively. Should those values be changed on the "Table" tab? Doing this changes the withdrawal rates (slightly).

Sorry if I'm totally off here; I'm still learning how all of this works.

(I'm looking at page 37 of this document: https://www.credit-suisse.com/media/ass ... y-2018.pdf)
It's only is a wild-ass guess such that good returns are higher and bad returns are lower. So, I see no need to constantly change it. See this earlier post for a more detailed explanation.

There's no need for false precision. Changes in annualized historical returns over a period starting in 1900 and spanning more than a century are tiny. Updating the trends from year to year would mainly affect the rounding of percentages in the VPW table and lead to minor changes. For one thing, I don't want to give people the impression that the table changes over time. For another, I don't want them to think that the table is of high precision; it's not.

I insist: the table isn't meant to be used with ultimate precision. If the VPW table indicates 4.8%, taking 5% won't lead to premature portfolio depletion, and taking 4.5% won't lead to dying with a gigantic pile of unspent money. The percentages are the result of a calculation based on a wild-ass guess; they inherit the "wild-ass guess" property. So, they should be used as a guide to indicate approximately how much of the portfolio should be withdrawn today to pay taxes and fund the upcoming year's expenses given the current portfolio balance, the target asset allocation for the upcoming year, and the current age of the retiree.

Whether the specific percentage in the VPW table is 4.8% or 4.9% isn't important. These are rounded values resulting from a calculation based on a wild-ass guess. What's important is to understand that it wouldn't be sustainable to withdraw 7% or 8% instead of 4.8%, and keep over-withdrawing like that regularly. I've kept one decimal of precision (after being rightfully chastised by forum member Rodc, early in this thread, for the false precision of using the default two decimals of Microsoft Excel/Libreoffice Calc) so that annual percentage adjustments remain small. The difference between 4% and 5% of a portfolio is big; the difference between 4.4% and 4.5% is much smaller and leads to a more reasonable annual adjustment.

On a technical level, in the past, Credit Suisse Global Investment Returns Yearbooks provided a "growth of 1$" value since 1900. The newer free Summary editions since year 2017 don't provide this anymore. The cell's formula (based on the last of the older yearbooks) takes the 116th root of final growth value (e.g. $300 for stocks) minus one as (stocks) growth trend.

I have no strong opinion about whether to update the spreadsheet's trends based on the latest Credit Suisse Yearbook Summary or not, except that I want users to know that they can take a single copy of the VPW table and use it all retirement long.

A compromise would be to update the trends every 5 years (in 2020, 2025, etc) just to reinforce the idea of stability; that there's no need to update the VPW table yearly (or ever). Would that be more intellectually satisfying than using stale values from 2016?
Bogleheads investment philosophy | Lifelong Portfolio: 25% each of (domestic/international)stocks/(nominal/inflation-indexed)bonds | VCN/VXC/VLB/ZRR

cap396
Posts: 17
Joined: Sun Oct 14, 2012 4:05 pm

Re: Variable Percentage Withdrawal (VPW)

Post by cap396 » Sat Jul 21, 2018 7:34 am

Thank you, longinvest, that's a very clear explanation and totally makes sense to me. I know you've put a lot of work and research into all of this, and it is really benefiting me and many others.

Post Reply