Ratchet Up

[audreyh1]

For US large cap stock, an OK estimate of expected real returns is 1/PE10

For bonds, just use the current interest rate and take off expected inflation.
Then weight the two based on your AA. If you're backtesting, you can try the previous year's inflation as expected inflation or maybe an average of the last 2 or 3 years. For year's they existed, the delta between nominal bonds and TIPs is another choice.

That should get you going. If people want to discuss this further or have other ideas, then a new thread should be opened.

OK, thanks! I’ll try this next time I run VPW.

The scenarios I’ve run (50% TSM/50% 5 year treasuries) show you can go as high as 4.35% of remaining portfolio for 30-40 years and still end up on average with the portfolio you started, in real terms. Worst case ending portfolio portfolio was just over half of initial amount in real terms. You might have to deal with a 60% drop in income in the worst year, worst case, but your lowest income would still be higher than more conservative withdrawal rates. And all your other years would have higher income. The worst starting years clustered around the turn of the century: 1899, 1906, 1892, etc.

 

[Perryinva]

What is VPW?

 

[Independent]

What is VPW?

Variable Percentage Withdrawal.

You withdraw some percent of the current portfolio. The percent grades up slowly as you age because your remaining lifespan shortens.

 

[mpeirce]

You might have to deal with a 60% drop in income in the worst year, worst case, but your lowest income would still be higher than more conservative withdrawal rates.

Which is why I use a trailing average of the last three years. It takes the edge off the wild swings both up and down. As does a cash buffer for those bad years when you don't want to sell too much.

 

[big-papa]

Variable Percentage Withdrawal.

You withdraw some percent of the current portfolio. The percent grades up slowly as you age because your remaining lifespan shortens.

It basically uses the PMT function in Excel. Think of it as the equivalent of the principal you owe on a loan. In this case, the percentage is the percentage withdrawn from your portfolio. If you use just a fixed estimate of stock and bond returns (as the VPW spreadsheet on bogleheads does), then the percentage you withdraw starts off low, increasing each year until the last year is 100% of the remaining portfolio. It makes no attempt to automatically keep up with inflation, though the increasing percentage helps a little along with gains from your portfolio.

 

[big-papa]

OK, thanks! I’ll try this next time I run VPW.

The scenarios I’ve run (50% TSM/50% 5 year treasuries) show you can go as high as 4.35% of remaining portfolio for 30-40 years and still end up on average with the portfolio you started, in real terms. Worst case ending portfolio portfolio was just over half of initial amount in real terms. You might have to deal with a 60% drop in income in the worst year, worst case, but your lowest income would still be higher than more conservative withdrawal rates. And all your other years would have higher income. The worst starting years clustered around the turn of the century: 1899, 1906, 1892, etc.

If you use the method I mention, you will need to modify the VPW spreadsheet or create your own because you need to update the expected returns each year. It's a personal choice, but I don't use starting years from the 1800's or even the early 1900's because I think the world was just too different then. I usually start with either the late 1920's or the 1950's, depending on what I'm trying to look at. When you do that, the worst year for VPW is still 1966, just like it is for a 4% rule.

VPW as it is written only gives you a couple of choices. The default is to, by design, end up with $0 at the end. But you can also change the PMT function to leave exactly a certain amount on the last year. What most people who use it do, however, is add buffer years. So if they think 30 years is a reasonable length retirement, they'll set up VPW for 35 or 40 years. Alternatively, you can make the number of years variable by using the IRS tables. Many possibilities.

 

[audreyh1]

Which is why I use a trailing average of the last three years. It takes the edge off the wild swings both up and down. As does a cash buffer for those bad years when you don't want to sell too much.

That 60% drop in income did not occur in a single year. In several cases it took 22 years to drop that low before finally turning around. Usually it took at least 15.

I’m not sure averaging over three years would make much difference in a long slow decline that usually causes failure or very low income scenarios.

Personally I prefer to take the full chunk out after a big up year figuring that it could go poof the next, so I might as well pull out the entire amount, accepting the raise the market gave me. I stockpile my excess income instead.

 

[audreyh1]

If you use the method I mention, you will need to modify the VPW spreadsheet or create your own because you need to update the expected returns each year. It's a personal choice, but I don't use starting years from the 1800's or even the early 1900's because I think the world was just too different then. I usually start with either the late 1920's or the 1950's, depending on what I'm trying to look at. When you do that, the worst year for VPW is still 1966, just like it is for a 4% rule.

VPW as it is written only gives you a couple of choices. The default is to, by design, end up with $0 at the end. But you can also change the PMT function to leave exactly a certain amount on the last year. What most people who use it do, however, is add buffer years. So if they think 30 years is a reasonable length retirement, they'll set up VPW for 35 or 40 years. Alternatively, you can make the number of years variable by using the IRS tables. Many possibilities.

Right, I noticed people added more years. Guessed life expectancy plus 5 or whatever.

It had also occurred to me to just stop at 7% withdrawal rate or something like that.

Yeah - I’ll look at where to modify the spreadsheet to update the expected returns each year. Thanks.

 

[big-papa]

Right, I noticed people added more years. Guessed life expectancy plus 5 or whatever.

It also occurred to me to just stop at 7% withdrawal rate or something like that.

Yeah - I’ll look at where to modify the spreadsheet to update the expected returns each year. Thanks.

Yep. I think the author of vpw suggested something like maxing out the percentage as well.

Sent from my Moto G (4) using Early Retirement Forum mobile app

 

[Danmar]

So are you saying that a fixed 4% WR resulted in a 94.9% success rate and a 4% ratchet (renewing in any year where 4% of beginning of year portfolio exceeds last years withdrawals plus inflation) resulted in a 92.3% success rate?

And with the ratchet strategy that, on average, your available spending is ~40% higher in years 11-20?

If so, for me 40% more spending in exchange for a 2.6% increase in failure seems like a reasonable tradeoff.

This is my conclusion as well. Things have been very good (on balance) since I retired in 2006. But up until last year I just spent divs from portfolio. Now I am liquidating relatively small amounts (1-2%) which allows me to boost spending by 10-20%. Will carry on this way until the correction hits and then will revert to divs.

Obviously you can spend more in good times as long as you are able to cut back in bad times. For the vast majority of cases we leave a lot of money on the table otherwise.

 

[pb4uski]

Agree... flexibility in withdrawals effectively offsets the negligible increase in failures... as long as you don't use the increases to make irrevocable lifestyle changes the ratchet should be sound.

 

[Independent]

It basically uses the PMT function in Excel. Think of it as the equivalent of the principal you owe on a loan. In this case, the percentage is the percentage withdrawn from your portfolio. If you use just a fixed estimate of stock and bond returns (as the VPW spreadsheet on bogleheads does), then the percentage you withdraw starts off low, increasing each year until the last year is 100% of the remaining portfolio. It makes no attempt to automatically keep up with inflation, though the increasing percentage helps a little along with gains from your portfolio.

Yes, I understand the math. I didn't think I needed to include it in that post.

Regarding the math, I think that 90% of the decision to use VPW is in the "P" and only 10% is in the "V".

People embarking on a "percent of current balance" withdrawal plan need to think about how they'll feel if their balance goes down. They should look at historic drops, not just single year but longer term bear markets, and be sure they will be willing to adjust spending accordingly.

Considering cases where that happens very soon after retirement makes that "real". In those years, you don't get much help from the "V" because the percent isn't varying very fast.

Many will consider some moderating strategy, but then it makes sense to back test those strategies and make sure they don't set themselves up for failure by being too slow to react on the downside. And, they might consider moderating up the upside so they don't overspend based on a short term bump. Again, some testing seems necessary.

The "V" comes into play more in the back end because your willingness to spend down principle in the later years can allow you to maintain spending in those years in the face of bad news that comes late in retirement.

 

[big-papa]

Yes, I understand the math. I didn't think I needed to include it in that post.

Regarding the math, I think that 90% of the decision to use VPW is in the "P" and only 10% is in the "V".

People embarking on a "percent of current balance" withdrawal plan need to think about how they'll feel if their balance goes down. They should look at historic drops, not just single year but longer term bear markets, and be sure they will be willing to adjust spending accordingly.

Considering cases where that happens very soon after retirement makes that "real". In those years, you don't get much help from the "V" because the percent isn't varying very fast.

Many will consider some moderating strategy, but then it makes sense to back test those strategies and make sure they don't set themselves up for failure by being too slow to react on the downside. And, they might consider moderating up the upside so they don't overspend based on a short term bump. Again, some testing seems necessary.

The "V" comes into play more in the back end because your willingness to spend down principle in the later years can allow you to maintain spending in those years in the face of bad news that comes late in retirement.

It's one of the reasons I don't like VPW as it was written. Year on year changes on "P' are fairly small in the early years, certainly making it the dominant factor. It's why, as I posted earlier, several of us have modified the algorithm to update expected returns each year. Once you do that V and P are both fluctuating year on year which tends to smooth the dollar withdrawals (after adjusting for inflation), especially over longer periods. Still, some apply some short term smoothing on top of that, limiting both upside and downsides year on year. Long term smoothing is relatively safe in all of the backtesting I've done since it's still a PMT function that uses remaining years. More care needs to be taken with the short term smoothing, however, as it could drain the portfolio quicker by sustaining higher withdrawals longer.

 

[Independent]

It's one of the reasons I don't like VPW as it was written. Year on year changes on "P' are fairly small in the early years, certainly making it the dominant factor. It's why, as I posted earlier, several of us have modified the algorithm to update expected returns each year. Once you do that V and P are both fluctuating year on year which tends to smooth the dollar withdrawals (after adjusting for inflation), especially over longer periods. Still, some apply some short term smoothing on top of that, limiting both upside and downsides year on year. Long term smoothing is relatively safe in all of the backtesting I've done since it's still a PMT function that uses remaining years. More care needs to be taken with the short term smoothing, however, as it could drain the portfolio quicker by sustaining higher withdrawals longer.

I understand the general direction you're going.

It appears that when market values go down, you increase your expected returns. I can see from other posts that you've walked through 1966 as a test case. Presumably, you've explained this in a Bogleheads thread, do you have a link?

 

[kcowan]

Originally Posted by pb4uski View Post
So are you saying that a fixed 4% WR resulted in a 94.9% success rate and a 4% ratchet (renewing in any year where 4% of beginning of year portfolio exceeds last years withdrawals plus inflation) resulted in a 92.3% success rate?

And with the ratchet strategy that, on average, your available spending is ~40% higher in years 11-20?

If so, for me 40% more spending in exchange for a 2.6% increase in failure seems like a reasonable tradeoff.

This is my conclusion as well. Things have been very good (on balance) since I retired in 2006. But up until last year I just spent divs from portfolio. Now I am liquidating relatively small amounts (1-2%) which allows me to boost spending by 10-20%. Will carry on this way until the correction hits and then will revert to divs.

Obviously you can spend more in good times as long as you are able to cut back in bad times. For the vast majority of cases we leave a lot of money on the table otherwise.

This is a big relief. It is what I follow anyway but I am pleased to see statistical support!

 

[big-papa]

I understand the general direction you're going.

It appears that when market values go down, you increase your expected returns. I can see from other posts that you've walked through 1966 as a test case. Presumably, you've explained this in a Bogleheads thread, do you have a link?

Hah, I can only wish that this was my idea. Siamond, among others have discussed this some. Here are some links with some lively conversations between Siamond and Longinvest (the author of VPW).

viewtopic.php?f=2&t=160073&start=50#p2418402
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=144089
AlohaJoe also has some posts over there but he also has a blog that discussed PMT based withdrawal methods here:
https://medium.com/@justusjp/flavors-of-pmt-based-withdrawals-part-1-of-2-mortality-dbe09aed5be1

And here is the original VPW thread. the discussions of smoothing don't show up until later, but it's a start.
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=120430


Just FYI before you jump in on bogleheads. Longinvest very much likes the pure form of VPW that he authored with a fixed expected return lasting the entire retirement. He has stated a view that if one wants a smoother trajectory, then one should have a different AA to support it and is not really in favor of long term smoothing like this. There is a lot of back and forth between Longinvest and Siamond because of this.

 

[big-papa]

Hah, I can only wish that this was my idea. Siamond, among others have discussed this some. Here are some links with some lively conversations between Siamond and Longinvest (the author of VPW).

viewtopic.php?f=2&t=160073&start=50#p2418402
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=144089
AlohaJoe also has some posts over there but he also has a blog that discussed PMT based withdrawal methods here:
https://medium.com/@justusjp/flavors-of-pmt-based-withdrawals-part-1-of-2-mortality-dbe09aed5be1

And here is the original VPW thread. the discussions of smoothing don't show up until later, but it's a start.
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=120430


Just FYI before you jump in on bogleheads. Longinvest very much likes the pure form of VPW that he authored with a fixed expected return lasting the entire retirement. He has stated a view that if one wants a smoother trajectory, then one should have a different AA to support it and is not really in favor of long term smoothing like this. There is a lot of back and forth between Longinvest and Siamond because of this.

Additional. https://www.bogleheads.org/forum/viewtopic.php?f=10&t=120430&start=500#p2436122

 

[Independent]

Additional. https://www.bogleheads.org/forum/viewtopic.php?f=10&t=120430&start=500#p2436122

Thanks, lots of reading there. Just skimming, I didn't see the particular method I thought you had suggested.

several of us have modified the algorithm to update expected returns each year. Once you do that V and P are both fluctuating year on year which tends to smooth the dollar withdrawals

That sounded to me like your smoothing method used varying estimates of future returns. Presumably, in years that your portfolio goes up you adjust to a lower future return, so the percent number goes down and the dollars you withdraw aren't as high as they would have been.

Based on an earlier post, I thought you might be tying to a trailing P/E ratio. In particular, when P/E is high, you lower the expected future return.

 

[big-papa]

Thanks, lots of reading there. Just skimming, I didn't see the particular method I thought you had suggested.



That sounded to me like your smoothing method used varying estimates of future returns. Presumably, in years that your portfolio goes up you adjust to a lower future return, so the percent number goes down and the dollars you withdraw aren't as high as they would have been.

Based on an earlier post, I thought you might be tying to a trailing P/E ratio. In particular, when P/E is high, you lower the expected future return.

It's based on Shiller's PE10 (aka CAPE) for stocks (1/CAPE is an OK guess for future expected returns and has the advantage of being easy to look up and compute each year) and for bonds it's the current interest rate of your favorite bond fund minus your best guess a future inflation.

Yup, it's a lot of reading.

Unfortunately, I don't know how to reference an exact post. But this should guide you.
https://www.bogleheads.org/forum/viewtopic.php?f=2&t=160073&start=50#p2418402
Refer to Siamond's post on March 14, 2015 at 7:12pm
The second paragraph is the explanation. This is the first place where Siamond made the suggestion of updating expected returns each year in the PMT equation instead of just using a fixed expected return for the duration of retirement. Now he used the interest rate for 10 year bonds which is probably OK. It's probably better to use the interest rate for whatever bond fund you're using.

So you can follow the trail from there.

siamond later recognized that this method takes care of longer term smoothing, but still sees the need for short term smoothing. He and longinvest went back and forth about it, but ultimately what siamond uses is described in the following post:
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=222480&p=3433727&hilit=smoothing+PMT#p3433869
Refer to Siamond's post on July 3rd, 2017 at 5:40pm

That's basically the gist of it all. Seems to work well when I backtest. If you deviate from a basic US large cap + treasury portfolio, then you'll have to add some more terms, get data from other places (like for international, etc),

 

[Cut-Throat]

That 60% drop in income did not occur in a single year. In several cases it took 22 years to drop that low before finally turning around. Usually it took at least 15.

This completely depends on your Asset Allocation. And by 'income', I'm assuming you mean just the Withdrawal Amount of your Portfolio? Income usually includes Social Security and Pensions as well. VPW should be looked at within the entire Framework for Post Retirement Spending....

60% Drop in Income is no where near my Personal Historical Worst Case. I've been using VPW for the last 5 years now.

 

[Perryinva]

Of course not. The last five years have been a solid part of the 9 year rising bull. You wouldn’t expect much variation. And it should always be understood that the income only refers to the withdrawals from portfolio, whether needed or Not. Needing to live on it or not is an entirely different discussion and is what determines your risk level and requirements for smoothing vs desire or comfort with smoothing. That is why back testing CAN provide valuable insights, but still is not definitive.

 

[audreyh1]

This completely depends on your Asset Allocation. And by 'income', I'm assuming you mean just the Withdrawal Amount of your Portfolio? Income usually includes Social Security and Pensions as well. VPW should be looked at within the entire Framework for Post Retirement Spending....

60% Drop in Income is no where near my Personal Historical Worst Case. I've been using VPW for the last 5 years now.

Not VPW. This is from running FIRECALC through the %remaining portfolio case with a 50/50 allocation and various withdrawal rates and looking at the real income (i.e. withdrawal) drops during the worst case years. Over many years, in the worst historical cases, you do see that much of a real decline because your portfolio drops that far down before finally recovering, and how far it drops depends on the withdrawal rate. I used TSM and 5 year treasuries for the 50/50. I’ll try to post the results table later.

I haven’t modeled other asset allocations. One can speculate that a higher equity allocation will make the portfolio drop more under the worst case historical scenarios, whether or not it improves average outcomes.

Yes, I was just looking at portfolio withdrawals. My income is without pension nor SS which I’m still 12 years from taking and it will then likely be quite small compared to my portfolio withdrawals. Every individual has to take into account their own total retirement income picture and spending requirements.

 

[Independent]

It's based on Shiller's PE10 (aka CAPE) for stocks (1/CAPE is an OK guess for future expected returns and has the advantage of being easy to look up and compute each year) and for bonds it's the current interest rate of your favorite bond fund minus your best guess a future inflation.

Yup, it's a lot of reading.

Unfortunately, I don't know how to reference an exact post. But this should guide you.
https://www.bogleheads.org/forum/viewtopic.php?f=2&t=160073&start=50#p2418402
Refer to Siamond's post on March 14, 2015 at 7:12pm
The second paragraph is the explanation. This is the first place where Siamond made the suggestion of updating expected returns each year in the PMT equation instead of just using a fixed expected return for the duration of retirement. Now he used the interest rate for 10 year bonds which is probably OK. It's probably better to use the interest rate for whatever bond fund you're using.

So you can follow the trail from there.

siamond later recognized that this method takes care of longer term smoothing, but still sees the need for short term smoothing. He and longinvest went back and forth about it, but ultimately what siamond uses is described in the following post:
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=222480&p=3433727&hilit=smoothing+PMT#p3433869
Refer to Siamond's post on July 3rd, 2017 at 5:40pm

That's basically the gist of it all. Seems to work well when I backtest. If you deviate from a basic US large cap + treasury portfolio, then you'll have to add some more terms, get data from other places (like for international, etc),

Thanks

It's going to take a while to digest this.

 

[big-papa]

Thanks

It's going to take a while to digest this.

Understandable. Also both siamond and I are still playing with this offline. What I suggested (1/CAPE) for an estimate of stock returns and bond interest rate minus expected inflation, updated each year seems pretty good. But those aren't the only forward looking models around.

 

[audreyh1]

Understandable. Also both siamond and I are still playing with this offline. What I suggested (1/CAPE) for an estimate of stock returns and bond interest rate minus expected inflation, updated each year seems pretty good. But those aren't the only forward looking models around.

That looks straightforward enough. Simple enough for me anyway.

I just need ballpark.