# VPW - Best Withdrawal Calculator I've seen to date.....

Are you sure you got 3.7% for a 2.0% projected real return rate? I just put the numbers in VPW and got exactly what I posted.

You can tell what the real rate of return assumed by VPW is by looking at what the "internal rate" setting is on the Calculation tab.

If you put in 100% bonds in the VPW tab, for instance, it gives you a 2% real rate and you get exactly the numbers that I posted.

You are using a 'Custom VPW' and I used the Standard VPW based on Asset Allocation of 70% Bonds and 30% Stocks. Change the AA to 70% Stocks and you get an initial WR of 4.5% (which is not unreasonable, just backtest it against against any of the Bad retirement years - 1937, 1966, 1929 etc. etc.)

In fact I am not even sure how you would input 2% projected real return in VPW, unless the sheet was changed. Where do you input 2%?

Last edited:
You are using a 'Custom VPW' and I used the Standard VPW based on Asset Allocation of 70% Bonds and 30% Stocks. Change the AA to 70% Stocks and you get an initial WR of 4.5% (which is not unreasonable, just backtest it against against any of the Bad retirement years - 1937, 1966, 1929 etc. etc.)

In fact I am not even sure how you would input 2% projected real return in VPW, unless the sheet was changed. Where do you input 2%?
Yes, I get the same number that you do when I put in 30% domestic stocks and 70% bonds into the VPW spreadsheet. But all this is doing is generating a projected real return number and basing the withdrawal calculation entirely off of this single number (plus life expectancy). In the case you describe, the real return number assumed by VPW is 2.9%. The web calculator gets the same 3.7% initial withdrawal rate using 2.9%.

I got the 2% real number by inputting 100% bonds. I just picked 2% real out of the air to explain to the previous poster what he was doing wrong.

Anyway, this is all VPW is doing (besides also backtesting). The entire withdrawal calculation is based on that single real return number. I think it is fundamental to understanding what VPW is really doing. No one seemed to explain this on those bogleheads threads. It is important to understand the assumptions.

Anyway, this is all VPW is doing (besides also backtesting). The entire withdrawal calculation is based on that single real return number. I think it is fundamental to understanding what VPW is really doing. No one seemed to explain this on those bogleheads threads. It is important to understand the assumptions.

VPW certainly cannot predict the future. And Backtesting is a Big Deal. It's all we got. Firecalc would do nothing at all, if it wasn't for backtesting.

And while the percentage of withdrawal is based on the Withdrawal Calculation. The Amount is based on the remaining portfolio balance and is backtested too.

So, I think that's about all we can do.

So playing around with it a bit, it seems to just come up with a withdraw percentage based on your AA and maybe beginning and ending age?

Yes, backtesting of VPW is certainly useful.

Also, since VPW never fails but just lowers the withdrawals in bad return scenarios, it is very important to look closely at scenarios to see if one would have really wanted to cut back that much, for instance.

While balancing that with probable increased lifetime consumption using VPW as compared to other methods like the standard inflation-adjusted SWR we often discuss.

In principle anyway, this is the same approach that we are confronted with when we do RMDs. Say you turn 70.5 on Jan 1,2010. That year, you are required to withdraw the Dec31, 2009 ending balance divided by your life expectancy as specified in an IRS table. IRA Required Minimum Distributions Table | Bankrate.com

As we can see, it is a fairly conservative method, and although I have not made an in depth comparison with Cut-Throat's favorite, I doubt that they can be very different, or at least very different in effectiveness.

I think everyone understands that one must accept either variability of annual income, or variability of time until the game is over. Or some combination of these two, as in Clyatt's variation.

None of these things can suspend reality.

Ha

Boy, those rmds are real conservative (assumes one lives to over 97 years old). Dist. period goes up by a factor of .9 per year. Guess if one wanted to assume living this old, you could subtract your present age from 97 and use this figure to divide into your nest egg on annual basis. Probably would result in leaving some coin on the table after you are put in the ground. Kiddies might be happy.

I am not clear on what kind of backtesting VPW is doing. Is it just using historical numbers to get the real rate of return based on asset allocation or is it actually testing for failure using its withdrawal rates, time frame & historical asset returns? Need to go back to that thread and see if that's been addressed.

I am not clear on what kind of backtesting VPW is doing. Is it just using historical numbers to get the real rate of return based on asset allocation or is it actually testing for failure using its withdrawal rates, time frame & historical asset returns? Need to go back to that thread and see if that's been addressed.

VPW cannot fail. The rate of return is a Fixed average based on your Asset Allocation. As far as backtesting, It will adjust your withdrawal amount from your remaining portfolio balance, as well as the backtested dates actual return rate. It also compares VPW to Std SWR and other withdrawal mthods.

Cut-Throat,

2.3M
35 years
75/25 AA
(My estimated ER expense: \$92,000)

One issue that I encountered is, when backtested from 1972, I will have a string of 11 years ('75 to '85) with much lower assets to withdraw using 'Var. %' method. Those would be much lower than my estimated ER expense. (Planned to ER next year.)

'Cost. \$' method is the only one with enough assets withdrawn yearly to meet the expenses.

How do you adjust for this to use 'Var. %' method? Keep working few more years to build up a larger asset to begin with?

Cut-Throat,

2.3M
35 years
75/25 AA
(My estimated ER expense: \$92,000)

One issue that I encountered is, when backtested from 1972, I will have a string of 11 years ('75 to '85) with much lower assets to withdraw using 'Var. %' method. Those would be much lower than my estimated ER expense. (Planned to ER next year.)

'Cost. \$' method is the only one with enough assets withdrawn yearly to meet the expenses.

How do you adjust for this to use 'Var. %' method? Keep working few more years to build up a larger asset to begin with?

Be Flexible. And remember, you probably will get Social Security and this does not drop with inflation, which is what your 1972-1983 time frame had a lot of. And doesn't constant Dollar eventually fail and you get nothing? Looks like you are completely broke with 10 years left to go in your plan to me with Constant \$.

And remember no withdrawal tool will save you from reality. This is just a planning tool. It does not guarantee a \$92 Grand a year lifestyle. If we get a decade of double digit inflation, Your Social Security will be about the only thing that saves you.

Last edited:
VPW might be better if it inc more AA options. Holding short vs long term bonds can yield quite different returns over time.

As was discussed on the BH forum, (C)VPW isn't for everybody. If you do want to maximize the use of your entire portfolio, if you have quite some flexibility in your spend, and if you have some cushion (e.g. SS, annuity, etc) to help you deal with the rough times, then it works great.

On the other hand, if you want to leave a significant legacy, or want to preserve your portfolio as a guarantee against LTC costs in your twilight years, or have no stomach for a highly variable annual spend, then maybe this isn't for you. Using life expectancy as an input parameter also makes me a bit nervous, who knows what's going to happen 30 to 40 years from now.

In any case, this is certainly a very useful addition to the set of possible withdrawal methods. And as the author emphasized, this isn't a full retirement plan, and it might be wise considering to apply it to part of your savings, not all of it.

PS. I did a LOT of backtesting with it, using my own Excel sheet and more diverse AA choices. And yes, as some of you pointed out, it does overreact to bear markets while maintaining a higher annual spend would actually be fine. It's actually not hard to tweak by adding a minimal floor on your spend. You just need to be careful to keep enough room for spend variations. Then it fits well with a dual-budget strategy, and gets more realistic.

Last edited:
As was discussed on the BH forum, (C)VPW isn't for everybody. If you do want to maximize the use of your entire portfolio, if you have quite some flexibility in your spend, and if you have some cushion (e.g. SS, annuity, etc) to help you deal with the rough times, then it works great.

On the other hand, if you want to leave a significant legacy, or want to preserve your portfolio as a guarantee against LTC costs in your twilight years, or have no stomach for a highly variable annual spend, then maybe this isn't for you. Using life expectancy as an input parameter also makes me a bit nervous, who knows what's going to happen 30 to 40 years from now.

In any case, this is certainly a very useful addition to the set of possible withdrawal methods. And as the author emphasized, this isn't a full retirement plan, and it might be wise considering to apply it to part of your savings, not all of it.

PS. I did a LOT of backtesting with it, using my own Excel sheet and more diverse AA choices. And yes, as some of you pointed out, it does overreact to bear markets while maintaining a higher annual spend would actually be fine. It's actually not hard to tweak by adding a minimal floor on your spend. You just need to be careful to keep enough room for spend variations. Then it fits well with a dual-budget strategy, and gets more realistic.
Thank you for testing the spreadsheet. I agree that it is a useful tool.

In my case the Random option is of interest. I switched to that and pressed the F9 key to spin the wheel, so to speak.

The Boglehead thread starts with "for a withdrawal method I've thought up ...". I'm sure this is an original idea to the writer, but the general category of withdrawal percentages that are 1/a, where "a" = present value of an annuity at current age, has been around for a long time.

As Kramer points out, RMD's are in this category, where "a" is the pv of a fixed annuity to the current life expectancy, at a 0% discount.
IIRC, the payout phase of a variable annuity has this form, where "a" is the pv of a life annuity with some assumed rate of about 3%.

In this version, "a" is the pv of a fixed annuity to some, pre-determined, date of death, at a discount rate that's a historic average return on an input AA.
I expect if we dig back into old posts on this forum, we'll see references to this method.

All percent of portfolio methods share the problem of poor early returns. They require spending cuts roughly proportional to the portfolio shrinkage in down markets. People may say "I can be flexible" without looking at past data and really facing the issue of what they will cut if they happen to hit the bad stretch. Running this worksheet with a 60/40 AA and starting in 1973 gives an initial withdrawal of \$50,000, rapidly dropping to \$29,900, which begins a 10 year stretch where the average withdrawal is \$32,500. Somebody using this method needs to think about that \$32,500 pretty carefully.

And, they all have issues toward the end. It's true that a fixed percent can never run out of money, but again the withdrawals can be extremely (too) low in later years. Alternatively, this approach schedules rising percents, with the trade-off that you're guaranteed to run out of money on some date. (But then caps the withdrawal, at a percent, with no clear idea of what that means for payout dollars.)

I think this is a decent withdrawal method, in fact, we're using something similar. But, many people have trouble converting the rules that result in percents into actual, spendable dollars. They need to look at the downsides very carefully.

One advantage of the traditional SWR methodology (fixed withdrawal amount adjusted for inflation) is that one can compare two different investment strategies and see how they hold up. With variable withdrawals since the outputs are constantly changing it would make it difficult to compare say different AAs since the outcomes will need to evaluated on multiple different criteria.

Have the people who've FIRE'd been comparing their balances with what they projected and then possibly adjusting their withdrawal for a given year if their balance was greater or lower than their initial calculations?

Or maybe you find you aren't spending the full withdrawal for a given year so you withdraw less the following year?

Have the people who've FIRE'd been comparing their balances with what they projected and then possibly adjusting their withdrawal for a given year if their balance was greater or lower than their initial calculations?

Or maybe you find you aren't spending the full withdrawal for a given year so you withdraw less the following year?

Yes to both questions.

4th year of our retirement and very good market returns so my target of 3% WR has not been put to the test as we've not wanted to spend more than 3% up to now.

Have the people who've FIRE'd been comparing their balances with what they projected and then possibly adjusting their withdrawal for a given year if their balance was greater or lower than their initial calculations?

Or maybe you find you aren't spending the full withdrawal for a given year so you withdraw less the following year?
I am a retired single guy in my late 40's living abroad, no pension, low withdrawal rate. I find that my spending is pretty steady (or at least not closely related to my current portfolio level) and so, in my experience, the idea of spending that is varying a lot does not seem to match the way that I live. Maybe other retirees have a different experience.

However, I like the idea of knowing what I perceive to be a safe withdrawal rate for any given year. That is important to me, to know that my spending is within appropriate parameters. So I like using tools like VPW, although I think VPW is way off on the real investment returns that it estimates for a given portfolio. But I just substitute my own educated estimates for projected real returns.

All percent of portfolio methods share the problem of poor early returns. They require spending cuts roughly proportional to the portfolio shrinkage in down markets. People may say "I can be flexible" without looking at past data and really facing the issue of what they will cut if they happen to hit the bad stretch. Running this worksheet with a 60/40 AA and starting in 1973 gives an initial withdrawal of \$50,000, rapidly dropping to \$29,900, which begins a 10 year stretch where the average withdrawal is \$32,500. Somebody using this method needs to think about that \$32,500 pretty carefully.

And, they all have issues toward the end. It's true that a fixed percent can never run out of money, but again the withdrawals can be extremely (too) low in later years. Alternatively, this approach schedules rising percents, with the trade-off that you're guaranteed to run out of money on some date. (But then caps the withdrawal, at a percent, with no clear idea of what that means for payout dollars.)

BTW - All FireCalc Failures also start with poor portfolio performance in the early years. They result in Complete failures. So to compensate you get your SWR rate down to 3% or 2.5% and I've even heard of 2%! So, every portfolio suffers from poor early performance. VPW just waits until it's necessary to cut back.

Your alternative to starting with a WR of \$50k and then dropping to an average of \$32.5K is to start with a 'Conservative' SWR of 3% or \$30K and prepare for the 'Great Market Downturn' right off the bat. (It may never happen, and you may have had more money to spend all along). No thanks!, I'll be Flexible. I really don't know anyone that would continue to take an inflation adjusted 4% into the teeth of a great recession.

And as for your point that withdrawals can be too low in later years, I don't think so. Show me a reasonable example where this is true? Backtest a year and show me how the spending is too meager in later years.

Last edited:
BTW - All FireCalc Failures also start with poor portfolio performance in the early years. They result in Complete failures. So to compensate you get your SWR rate down to 3% or 2.5% and I've even heard of 2%! So, every portfolio suffers from poor early performance. VPW just waits until it's necessary to cut back.

Your alternative to starting with a WR of \$50k and then dropping to an average of \$32.5K is to start with a 'Conservative' SWR of 3% or \$30K and prepare for the 'Great Market Downturn' right off the bat. (It may never happen, and you may have had more money to spend all along). No thanks!, I'll be Flexible. I really don't know anyone that would continue to take an inflation adjusted 4% into the teeth of a great recession.

And as for your point that withdrawals can be too low in later years, I don't think so. Show me a reasonable example where this is true? Backtest a year and show me how the spending is too meager in later years.
I see some apples vs. oranges regarding both calculators and conservatism here. If I use the same calculator, and the same conservatism, the relation really is between a flexible withdrawal starting at \$50,000 and a fixed withdrawal starting at \$40,000 (or, between a flexible withdrawal starting at \$37,500 and a fixed withdrawal starting at \$30,000, etc).

The VPW calculator only allows starting years of 1972 or later. For those years, there are no failures for a flat \$40,000 withdrawal and the 60/40 AA. I think the worst year is 1973, which results in a portfolio of \$290,000 after 35 years. So, if I believe the backtesting in the VPW calculator, \$40,000 is a reasonable fixed withdrawal.

FireCalc doesn't fail either for start years of 1972 or later, with a 60/40 AA.
FireCalc will fail for start year 1973 if I use a 75/25 AA. The VPW calculator does not.
FireCalc has -\$28,000 after 30 years. The VPW calculator, with the same 75/25 AA and \$40,000 withdrawals, has \$356,000 at the end of 30 years.

This isn't surprising. All these calculations are sensitive to the source of the historic returns and to small decisions regarding calculations. For example, the standard FireCalc assumption is 18 bp of investment expense. It also applies one extra year of inflation to each withdrawal (it takes the withdrawal at the beginning of the year, but the dollar amount it withdraws includes that year's inflation).

That's what I mean by apples vs. oranges regarding calculators. It seems wrong to test the fixed dollar SWR using FireCalc, then compare it to the Variable Percentage using the VPW calculator.

The "conservatism" comes from the notion that people who talk about 3% or 2.5% SWR are not relying on backtesting. They think it's likely that the future will be worse than the past. If I ask Firecalc for a 30 year, 100% success rate for its 113 possible start years, it will give me about 3.6%. If I could load the VPW calculator with the same 113 years of data, and use FireCalc's calculation assumptions, it might tell me that I shouldn't start with a 5% or \$50,000 first year withdrawal, but maybe 4.5% or \$45,000 instead.

I'm not trying to say that a variable percentage withdrawal is a bad idea. As I've already said, I use a version of it myself. My original point was that I think some people don't realize the magnitude of the spending reductions that these strategies assume.

The point of this post is that I think your numeric example is somewhat apples to oranges in terms of assumptions.

The VPW calculator only allows starting years of 1972 or later.

Use the Shiller Data Set and you can go back as far as FireCalc. Then you can comment.

Use the Shiller Data Set and you can go back as far as FireCalc. Then you can comment.
Thanks, this helps.

I set Data Set to "Shiller" and the AA to 75/25.

I didn't test every year, but I did check those where FireCalc SWR's tend to fail -- late 1960s and early 1970s. It turns out that 3.7% is the lowest number that gets through all of those years. So, if we're going to use backtesting, I'd test against 3.7% or \$37,000.

The worst start year was 1966, where the \$37,000 constant dollar withdrawals ended 35 years with a balance of \$46,000.

That may also be the worst year for variable percentage withdrawals. The initial withdrawal is \$53,000, it eventually drops to \$22,800 before recovering.

Of course, the percentage withdrawals are scheduled to go to zero in the 36th year, and they do.

The VP withdrawals start at \$53,000 and stay above \$37,000 for 8 years.
Then they drop below, and stay below for the next 13 years.

To me, this 3.7% is a better comparison than "3% or even 2.5%".

The summary is still that percentage withdrawals have a lot going for them, but people embarking on this method should think ahead to whether they can really live on a lot less than their initial withdrawal if they happen to catch a bad retirement year. It's clear to me in this case that people who want to spend the \$53,000 in the first year need some plan to live on less than \$37,000 if the need arises. Those who really need at least \$37,000 in every year should not start with \$53,000 in the first year.

IMO, models using percentage withdrawal methods should have an additional input of "The smallest withdrawal I'm willing to accept in any year". That would help focus the user on the downside risk.

edit: I did a little more on this. I modified the VPW worksheet to accept a minimum \$ withdrawal. To offset the fact that I'd be withdrawing "too much" in some down years, I also put in a maximum \$ withdrawal.
Of course, if I set the minimum to \$37,000, then I need to set the maximum to about \$37,000, too.
Other possibilities (all approximate) are:
Min: \$35,000 Max: \$42,000
Min: \$33,000 Max: \$46,000
Min: \$31,000 Max: \$51,000

Last edited:
IMO, models using percentage withdrawal methods should have an additional input of "The smallest withdrawal I'm willing to accept in any year". That would help focus the user on the downside risk.

The problem of course, is that the downside risk is unknown and possibly catastrophic. And by setting a 'Brake' on VPW, it becomes less effective (You are taking out more in down years, which is the real devastation of portfolios).

Yes, you have to know what the risk is in any portfolio withdrawal technique. But we're all big boys and girls here, otherwise we would not even be in this position. So, that goes without saying.

But the big benefit of VPW, is that it does not start preparing for 'Bad Times', until they arrive. So, it lets you spend more in your early on in retirement. And if the 'Bad times' never come, you don't end up with the huge pile of assets at the end of your retirement like a fixed SWR does MOST of the time. Remember you had to 'hunt' for 1966 for your example.

And if you think about it, starting with a Small SWR to prepare for a period of record low returns, means if that a bad market does come, you'll have a lot more invested to lose, than if you would have spent the money early on in retirement. This was a point that I was not understanding until I ran the numbers. Saving money for a Market Downturn, the money gets lost by falling market prices instead of that African Safari Vacation (Which I did check off my bucket list this past June)

The problem of course, is that the downside risk is unknown and possibly catastrophic. And by setting a 'Brake' on VPW, it becomes less effective (You are taking out more in down years, which is the real devastation of portfolios).

Yes, you have to know what the risk is in any portfolio withdrawal technique. But we're all big boys and girls here, otherwise we would not even be in this position. So, that goes without saying.

But the big benefit of VPW, is that it does not start preparing for 'Bad Times', until they arrive. So, it lets you spend more in your early on in retirement. And if the 'Bad times' never come, you don't end up with the huge pile of assets at the end of your retirement like a fixed SWR does MOST of the time. Remember you had to 'hunt' for 1966 for your example.

And if you think about it, starting with a Small SWR to prepare for a period of record low returns, means if that a bad market does come, you'll have a lot more invested to lose, than if you would have spent the money early on in retirement. This was a point that I was not understanding until I ran the numbers. Saving money for a Market Downturn, the money gets lost by falling market prices instead of that African Safari Vacation (Which I did check off my bucket list this past June)

Note that I edited my prior post with a few "collar" numbers. If I'm going to put a minimum withdrawal in, I need to offset that with some reductions in the good year (and first year) withdrawals. There's really a spectrum here from fully percentage to fully fixed. I gave some points on the spectrum.

I'm going to guess that someone who is taking African safaris can probably live on SS + 1% of initial portfolio. So the downside "risk" is just giving up some luxuries, and it's pretty easy for me to reason that if I'm in that position should enjoy those luxuries early, knowing I've got plenty of cushion.

Other people, who could (for example) be retiring early due to an unexpected layoff, might be running much closer to their "needs" and should take the downside more seriously.