Alternative to the 4% rule

[mrWinter]

Quote:

Originally Posted by big-papa As it turns out 1/CAPE is a decent, not perfect, approximation of future real returns for the SP500 (and by extension a total stock market return).

That's very interesting and was not my understanding of the CAPE. I don't think I understand the why of that fact, I'll have to put some thought into it and read those derivations you linked. Thanks for explaining/sharing.

[thepalmersinking]

All of this is great conversation but when it’s time to actually pull out of the portfolio is there a strategy for that. If I have 15 stocks, do I take 3-4% equally out of each individual holding? What if some were losers that year and others had a high dividend rate. Maybe I am overcomplicating this...

[big-papa]

Quote:

Originally Posted by mrWinter That's very interesting and was not my understanding of the CAPE. I don't think I understand the why of that fact, I'll have to put some thought into it and read those derivations you linked. Thanks for explaining/sharing.

No problem.

Just note that there are many methods to get potential future returns to use in conjunction with a PMT based withdrawal method.
1/CAPE is probably the easiest since CAPE is readily available.
As I noted, you can probably get a little more accuracy by doing a linear regression of CAPE. In my own backtesting, though, I really didn't see much difference between that and just using 1/CAPE
And as I posted earlier in the thread, many research houses have their capital returns expectations posted - Unfortunately, I haven't been able to find a long term source of previous expectations to see if plugging it into a PMT based withdrawal would be any better. I personally doubt it given the error bars associated with any prediction like that.

And there are other mathematical methods that can be calculated or looked up including.
Tobin's Q
"greatest ever" from philosophical economics
Warren Buffet's percentage of total market cap (TMC) relative to the U.S. GNP

The possibilities are endless. I, for one want something somewhat simple and nearly autopilot. Looking up CAPE, current yield of my bond fund, and inflation expectations is about as easy as this could get.

What I don't advocate is using this information to set your AA - there's plenty enough research on that.

[RunningBum]

Quote:

Originally Posted by thepalmersinking All of this is great conversation but when it’s time to actually pull out of the portfolio is there a strategy for that. If I have 15 stocks, do I take 3-4% equally out of each individual holding? What if some were losers that year and others had a high dividend rate. Maybe I am overcomplicating this...

That would be a different subject, please don't take this thread off-topic. I suggest you start your own thread on that, or find an existing about portfolio withdraws. I know there have been a few threads on this, but I'm not sure if there's one specific to owning individual stocks, so a new thread with your situation would be appropriate.

[big-papa]

OK, I've discussed my thoughts on long term smoothing when using a PMT based method. I had mentioned earlier that sometimes people want to add short term smoothing as well. There are a few methods for that that can be applied to a PMT based withdrawal method.
1. Modify Clyatt to withdraw the % calculated by PMT or 95% of the previous year's withdrawal whichever is larger
2. Same as #1, but use 95% of the previous year's real withdrawal. When looking at inflation adjusted withdrawals instead of nominal withdrawals, this will provide additional smoothing.
(Note, there is nothing magic about 95%. You can backtest and play around with other numbers - I actually saw better results with higher percentages)
3. Finally, this is something suggested by one of the bogleheads. It bears some resemblance to Clyatt except that the calculation includes both last year's actual withdrawal and a first calculation of this year's withdrawal to calculate the actual withdrawal to be made this year.
a. Use a smoother coefficient. This is your choice. 75% is a decent choice, but you may want to play around with it.
a. Calculate the % withdrawal using the PMT formula as always
b. Multiply the % withdrawal by the amount in your porfolio to get a first calculation of what this year's withdrawal would be.
c. If the amount calculated in b above is greater than the previous year's nominal withdrawal, then this year's withdrawal is the previous year's withdrawal + the smoothing coefficient * ((the amount calculated in b) - last year's nominal withdrawal))d. Likewise, if the amount calculated in b above is less than the previous year's nominal withdrawal, then this year's withdrawal is the previous year's withdrawal - the smoothing coefficient * ((the previous year's nominal withdrawal - (the amount calculated in b))


Basically, this just let's your new withdrawal move only part of the way between whatever last year's withdrawal was and what this year's nominal calculation would be. Clyatt uses 95% of the previous year's withdrawal as the limiter. This instead let's the withdrawal move as much as 75% (or whatever % you choose) of the way between last year's withdrawal and the first calculation of what this withdrawal would be.

A modification of this would also be to do the same math but on real withdrawals and then back-calculate what the nominal withdrawal from your portfolio would be.


I've never looked at using this one on any other withdrawal method, but for those using a fixed % withdrawal, it might be interesting to compare it to Clyatt or other smoothing methods.


Cheers.

[big-papa]

Quote:

Originally Posted by big-papa OK, I've discussed my thoughts on long term smoothing when using a PMT based method. I had mentioned earlier that sometimes people want to add short term smoothing as well. There are a few methods for that that can be applied to a PMT based withdrawal method. 1. Modify Clyatt to withdraw the % calculated by PMT or 95% of the previous year's withdrawal whichever is larger 2. Same as #1, but use 95% of the previous year's real withdrawal. When looking at inflation adjusted withdrawals instead of nominal withdrawals, this will provide additional smoothing. (Note, there is nothing magic about 95%. You can backtest and play around with other numbers - I actually saw better results with higher percentages) 3. Finally, this is something suggested by one of the bogleheads. It bears some resemblance to Clyatt except that the calculation includes both last year's actual withdrawal and a first calculation of this year's withdrawal to calculate the actual withdrawal to be made this year. a. Use a smoother coefficient. This is your choice. 75% is a decent choice, but you may want to play around with it. a. Calculate the % withdrawal using the PMT formula as always b. Multiply the % withdrawal by the amount in your porfolio to get a first calculation of what this year's withdrawal would be. c. If the amount calculated in b above is greater than the previous year's nominal withdrawal, then this year's withdrawal is the previous year's withdrawal + the smoothing coefficient * ((the amount calculated in b) - last year's nominal withdrawal))d. Likewise, if the amount calculated in b above is less than the previous year's nominal withdrawal, then this year's withdrawal is the previous year's withdrawal - the smoothing coefficient * ((the previous year's nominal withdrawal - (the amount calculated in b)) Basically, this just let's your new withdrawal move only part of the way between whatever last year's withdrawal was and what this year's nominal calculation would be. Clyatt uses 95% of the previous year's withdrawal as the limiter. This instead let's the withdrawal move as much as 75% (or whatever % you choose) of the way between last year's withdrawal and the first calculation of what this withdrawal would be. A modification of this would also be to do the same math but on real withdrawals and then back-calculate what the nominal withdrawal from your portfolio would be. I've never looked at using this one on any other withdrawal method, but for those using a fixed % withdrawal, it might be interesting to compare it to Clyatt or other smoothing methods. Cheers.

One thing to note about short term smoothing methods. This is especially true for something like Clyatt. It becomes something of a pay me nor or pay me later proposition. In a bear market/high inflation environment such as one would see with a retirement starting in the late 1960's, short term smoothing definitely slowed down how quickly your withdrawals dropped year after year. But it also slowed down the time it takes to recover back to where your original withdrawals were, if ever. No smoothing had your withdrawals dropping more quickly, but the recovery was quicker. Pick your poison.

[Dtail]

Quote:

Originally Posted by big-papa One thing to note about short term smoothing methods. This is especially true for something like Clyatt. It becomes something of a pay me nor or pay me later proposition. In a bear market/high inflation environment such as one would see with a retirement starting in the late 1960's, short term smoothing definitely slowed down how quickly your withdrawals dropped year after year. But it also slowed down the time it takes to recover back to where your original withdrawals were, if ever. No smoothing had your withdrawals dropping more quickly, but the recovery was quicker. Pick your poison.

Do you find this true also in a bear market low inflation scenario?
All these Clyatt references are interesting to me for next year.
Thanks big-papa.

[audreyh1]

Quote:

Originally Posted by thepalmersinking All of this is great conversation but when it’s time to actually pull out of the portfolio is there a strategy for that. If I have 15 stocks, do I take 3-4% equally out of each individual holding? What if some were losers that year and others had a high dividend rate. Maybe I am overcomplicating this...

Is your goal to have the same amount in each stock? In that case you have to calculate how much to take out of each stock to take out to get back to equal weighting after your withdrawal (rebalance). Let your dividends accumulate in cash - that should fund part of it. And rebalance with what rest is needed.

[audreyh1]

Quote:

Originally Posted by big-papa One thing to note about short term smoothing methods. This is especially true for something like Clyatt. It becomes something of a pay me nor or pay me later proposition. In a bear market/high inflation environment such as one would see with a retirement starting in the late 1960's, short term smoothing definitely slowed down how quickly your withdrawals dropped year after year. But it also slowed down the time it takes to recover back to where your original withdrawals were, if ever. No smoothing had your withdrawals dropping more quickly, but the recovery was quicker. Pick your poison.

When comparing Clyatt's method versus the straight %remaining portfolio method, I found that during the worst scenarios, there was little difference in the drops in real income. Clyatt's method does not maintain 95% of prior spend in real terms. So if inflation is eating your lunch, it's not providing much protection.

It's probably great for those shorter drops and recoveries though.

[big-papa]

Quote:

Originally Posted by Dtail Do you find this true also in a bear market low inflation scenario? All these Clyatt references are interesting to me for next year. Thanks big-papa.

I think it's the case in a long term down market when you consider it in "real", not nominal returns. If you think of it that way, then it doesn't matter what the cause of the lower returns is: whether it's stocks being lower and/or inflation effectively given the returns a haircut. The result is the same: lower "real" returns.

[big-papa]

Quote:

Originally Posted by audreyh1 When comparing Clyatt's method versus the straight %remaining portfolio method, I found that during the worst scenarios, there was little difference in the drops in real income. Clyatt's method does not maintain 95% of prior spend in real terms. So if inflation is eating your lunch, it's not providing much protection. It's probably great for those shorter drops and recoveries though.

Yup - that's why I suggested a modified version of Clyatt to doing the math in real, instead of nominal terms. When I did that in the past, it helped quite a bit more.

[Dtail]

Quote:

Originally Posted by big-papa I think it's the case in a long term down market when you consider it in "real", not nominal returns. If you think of it that way, then it doesn't matter what the cause of the lower returns is: whether it's stocks being lower and/or inflation effectively given the returns a haircut. The result is the same: lower "real" returns.

Okay thanks.
So in nominal terms if one started with 1mm and 40k withdrawal in year 1, then year 2 the portfolio drops to 800k and the calc would be 32k, but Clyatt calc at 95% is 38k.
Now if the inflation rate is at 2%, what is the withdrawal and calculation in year 2 in real terms? Sorry if this a is simplistic question.

[big-papa]

Quote:

Originally Posted by Dtail Okay thanks. So in nominal terms if one started with 1mm and 40k withdrawal in year 1, then year 2 the portfolio drops to 800k and the calc would be 32k, but Clyatt calc at 95% is 38k. Now if the inflation rate is at 2%, what is the withdrawal and calculation in year 2 in real terms? Sorry if this a is simplistic question.

Glad to help.
So start where you started.
Year0: $1M portfolio, 4% withdrawal = $40K. Since this is year 0, the real withdrawal and the nominal withdrawal are the same.

Year1: $800K portfolio, 2% inflation happened in year 0. 4% of $800K is $32K. But relative to year 0, it is worth $32K/1.02=$31,373. So, in real terms, your withdrawal dropped (31.373-40)/40=21.6% Enter Clyatt. Let's only let it drop 5%. So, 0.95*40K=$38K. So your real withdrawal is $38K. However, you don't withdraw "real" dollars from your porfolio, you withdraw nominal dollars. So, the actual amount you withdraw from your $800K portfolio is $38K*1.02=38.76K. In nominal terms, your withdrawal dropped (38.76-40)/40=3.1% But in real terms, it dropped (38-40)/40=5.0%


If I had done all of the math in nominal terms, then it would have looked like this:
Year1: $800K portfolio, 2% inflation happened in year 0. $% of $800K is $32K. 95% of $40K from year1 would be $38K, so I would withdraw $38K nominal dollars which are worth $38K/1.02= $37.2K. By doing the math with real terms my withdrawal was actually worth $38K.


Summary:
Using real math the withdrawals went from $40K to $38K real dollars or a 5% drop.
Using nominal math, the withdrawals went from $40K to $37.2K real dollars or a 7% drop.

As the years march on, you will need to keep up with the cumulative effect of inflation on both the previous year and current year's withdrawals.

Hope this helps.

[Dtail]

Quote:

Originally Posted by big-papa Glad to help. So start where you started. Year0: $1M portfolio, 4% withdrawal = $40K. Since this is year 0, the real withdrawal and the nominal withdrawal are the same. Year1: $800K portfolio, 2% inflation happened in year 0. 4% of $800K is $32K. But relative to year 0, it is worth $32K/1.02=$31,373. So, in real terms, your withdrawal dropped (31.373-40)/40=21.6% Enter Clyatt. Let's only let it drop 5%. So, 0.95*40K=$38K. So your real withdrawal is $38K. However, you don't withdraw "real" dollars from your porfolio, you withdraw nominal dollars. So, the actual amount you withdraw from your $800K portfolio is $38K*1.02=38.76K. In nominal terms, your withdrawal dropped (38.76-40)/40=3.1% But in real terms, it dropped (38-40)/40=5.0% If I had done all of the math in nominal terms, then it would have looked like this: Year1: $800K portfolio, 2% inflation happened in year 0. $% of $800K is $32K. 95% of $40K from year1 would be $38K, so I would withdraw $38K nominal dollars which are worth $38K/1.02= $37.2K. By doing the math with real terms my withdrawal was actually worth $38K. Summary: Using real math the withdrawals went from $40K to $38K real dollars or a 5% drop. Using nominal math, the withdrawals went from $40K to $37.2K real dollars or a 7% drop. As the years march on, you will need to keep up with the cumulative effect of inflation on both the previous year and current year's withdrawals. Hope this helps.

Excellent thanks. Gets confusing sometimes with calcs on arithmetic vs. geometric terms if that makes sense.

[esterhazy]

Quote:

Originally Posted by audreyh1 Yeah - that's actually what I don't like. I prefer to take the $40K out when thing are high, even if I might only spend $30K, and let the 20% drop happen to the remaining $960K. I now have the "extra" $10K out of the portfolio that didn't get hit by 20%. I have part of that $10K to help me weather the lower income the next year.

FWIW, if you take out $40,000 but then only spend $30,000 then you haven't actually taken out $40,000. You took out $30,000 and then changed the asset allocation on the other $10,000. This results in e.g. shifting your portfolio from 80/20 to 79.8/20.2 (or whatever).

Which is perfectly fine! But I think once you open yourself up to not maintaining a static asset allocation, it is worth seeing whether this "hold the extra in cash/bonds" ends up making a difference and, if so, how much.

(There's also the argument that if your spending strategy says to spend $40,000 then you should trust your strategy instead of second guessing it. And use that extra $10,000 for a kick ass vacation, or an early gift to your heirs, or a charitable contribution.)

Anyway, I wrote up backtesting of this kind of approach -- "if the algorithm tells me to take out $120,000 but I only need $80,000 to be ridiculously happy, then put the extra $40,000 into bonds".

My first test had coding errors, so here is the fixed one -- https://medium.com/@justusjp/actuarial-harvesting-im-an-idiot-69b48c115640

Here's the original post that goes into a bit more detail on the process -- https://medium.com/@justusjp/actuarial-harvesting-be17d3a044ce

[Dtail]

Quote:

Originally Posted by esterhazy FWIW, if you take out $40,000 but then only spend $30,000 then you haven't actually taken out $40,000. You took out $30,000 and then changed the asset allocation on the other $10,000. This results in e.g. shifting your portfolio from 80/20 to 79.8/20.2 (or whatever). Which is perfectly fine! But I think once you open yourself up to not maintaining a static asset allocation, it is worth seeing whether this "hold the extra in cash/bonds" ends up making a difference and, if so, how much. (There's also the argument that if your spending strategy says to spend $40,000 then you should trust your strategy instead of second guessing it. And use that extra $10,000 for a kick ass vacation, or an early gift to your heirs, or a charitable contribution.) Anyway, I wrote up backtesting of this kind of approach -- "if the algorithm tells me to take out $120,000 but I only need $80,000 to be ridiculously happy, then put the extra $40,000 into bonds". My first test had coding errors, so here is the fixed one -- https://medium.com/@justusjp/actuari...t-69b48c115640 Here's the original post that goes into a bit more detail on the process -- https://medium.com/@justusjp/actuari...g-be17d3a044ce

First, welcome to our wonderful site esterhazy.
I am also following this concept of Audreyh1. On your section which I bolded, if one doesn't have any needs for the extra monies in that particular year, why spend it? So it is not a second guessing of strategy, just more of a desire to keep to a set % of portfolio spend and bank what's not needed for a rainy day/expensive vacation/new car, etc.
For me, I don't think that this amount would ever be more than 5% of total assets and thus if it would be included theoretically in the portfolio, it would be within the 5% band window anyway.

[big-papa]

Quote:

Originally Posted by esterhazy FWIW, if you take out $40,000 but then only spend $30,000 then you haven't actually taken out $40,000. You took out $30,000 and then changed the asset allocation on the other $10,000. This results in e.g. shifting your portfolio from 80/20 to 79.8/20.2 (or whatever). Which is perfectly fine! But I think once you open yourself up to not maintaining a static asset allocation, it is worth seeing whether this "hold the extra in cash/bonds" ends up making a difference and, if so, how much. (There's also the argument that if your spending strategy says to spend $40,000 then you should trust your strategy instead of second guessing it. And use that extra $10,000 for a kick ass vacation, or an early gift to your heirs, or a charitable contribution.) Anyway, I wrote up backtesting of this kind of approach -- "if the algorithm tells me to take out $120,000 but I only need $80,000 to be ridiculously happy, then put the extra $40,000 into bonds". My first test had coding errors, so here is the fixed one -- https://medium.com/@justusjp/actuari...t-69b48c115640 Here's the original post that goes into a bit more detail on the process -- https://medium.com/@justusjp/actuari...g-be17d3a044ce

Absolutely correct. The AA has changed and now you're in a dynamic AA situation. Nothing at all wrong with that.

For another resource on withdrawal methods that dynamically change the AA, I would recommend McClung's "Living off your Money". The first 3 chapters contain the meat of the idea and are downloadable for free. Essentially, you start with your favorite AA, then withdraw only from bonds until stocks hit a certain threshold at which point some money is drained from stocks to bonds. And what to do for corner cases such as if bonds completely drain before stocks hit the threshold. Note that this isn't so much a withdrawal method that tells you how much to withdraw - it is mainly to tell you how to withdraw from the portfolio once you have a method. Later on he does introduce actual methods to determine how much to withdraw, but you can use this idea with any withdrawal method.

Investing During Retirement

Cheers

[big-papa]

Quote:

Originally Posted by esterhazy FWIW, if you take out $40,000 but then only spend $30,000 then you haven't actually taken out $40,000. You took out $30,000 and then changed the asset allocation on the other $10,000. This results in e.g. shifting your portfolio from 80/20 to 79.8/20.2 (or whatever). Which is perfectly fine! But I think once you open yourself up to not maintaining a static asset allocation, it is worth seeing whether this "hold the extra in cash/bonds" ends up making a difference and, if so, how much. (There's also the argument that if your spending strategy says to spend $40,000 then you should trust your strategy instead of second guessing it. And use that extra $10,000 for a kick ass vacation, or an early gift to your heirs, or a charitable contribution.) Anyway, I wrote up backtesting of this kind of approach -- "if the algorithm tells me to take out $120,000 but I only need $80,000 to be ridiculously happy, then put the extra $40,000 into bonds". My first test had coding errors, so here is the fixed one -- https://medium.com/@justusjp/actuari...t-69b48c115640 Here's the original post that goes into a bit more detail on the process -- https://medium.com/@justusjp/actuari...g-be17d3a044ce

And now that I've clicked on the link, I recognize the author from BH. He and I both have a different name over here on early-retirement than we do over on BH. Anyway, he has a number of very good articles on his blog and I recommend folks go through them!

He also has a good explanation of Prime Harvesting from McClung's book, especially the "memory effect". I liken it more to "sensitivity to initial conditions", but that's not as short as "memory effect". :-)

[audreyh1]

Quote:

Originally Posted by esterhazy FWIW, if you take out $40,000 but then only spend $30,000 then you haven't actually taken out $40,000. You took out $30,000 and then changed the asset allocation on the other $10,000. This results in e.g. shifting your portfolio from 80/20 to 79.8/20.2 (or whatever). Which is perfectly fine! But I think once you open yourself up to not maintaining a static asset allocation, it is worth seeing whether this "hold the extra in cash/bonds" ends up making a difference and, if so, how much. (There's also the argument that if your spending strategy says to spend $40,000 then you should trust your strategy instead of second guessing it. And use that extra $10,000 for a kick ass vacation, or an early gift to your heirs, or a charitable contribution.) Anyway, I wrote up backtesting of this kind of approach -- "if the algorithm tells me to take out $120,000 but I only need $80,000 to be ridiculously happy, then put the extra $40,000 into bonds". My first test had coding errors, so here is the fixed one -- https://medium.com/@justusjp/actuari...t-69b48c115640 Here's the original post that goes into a bit more detail on the process -- https://medium.com/@justusjp/actuari...g-be17d3a044ce

No, it’s not second guessing. It’s just a simple way of dealing with the inconvenience of a highly variable income from my %remaining portfolio withdrawal method. There is no rule that you must spend the funds the same year that you withdraw them (OK - some people disagree with this). I can spend them at any time in the near future and am very likely to do so on various splurges. I will probably always have some minimal buffer for a rainy day, but it doesn’t mean the rest won’t be spent.

Once it’s withdrawn from the retirement portfolio it no longer matters in terms of AA. IMO, I only apply the withdrawal % and rebalancing to the retirement portfolio AA, so what happens outside of it doesn’t matter. It doesn’t screw up any models - in terms of survival characteristics and sequence of returns I am only modeling what happens to the retirement portfolio. Many people disagree with this type of segregation of a subset of their assets, but IMO, you have to set up an AA on something, withdraw and rebalance, and if it’s a subset of your investable assets, so what? All that matters to me is: the retirement portfolio large enough? If I didn’t think it was I would have already lowered my withdrawal rate.

Edited to add:
I can’t predict how much I’ll have left over any given year. I may have a budget (more like guidelines), but that doesn’t mean I spend that exact budget every year. My spending is variable.

Also, unspent funds aren’t set aside indefinitely in some bond fund that is then withdrawn from using some predetermined withdrawal rate as the retirement portfolio. They are available for spending at any time.

[Dtail]

Quote:

Originally Posted by big-papa Absolutely correct. The AA has changed and now you're in a dynamic AA situation. Nothing at all wrong with that. For another resource on withdrawal methods that dynamically change the AA, I would recommend McClung's "Living off your Money". The first 3 chapters contain the meat of the idea and are downloadable for free. Essentially, you start with your favorite AA, then withdraw only from bonds until stocks hit a certain threshold at which point some money is drained from stocks to bonds. And what to do for corner cases such as if bonds completely drain before stocks hit the threshold. Note that this isn't so much a withdrawal method that tells you how much to withdraw - it is mainly to tell you how to withdraw from the portfolio once you have a method. Later on he does introduce actual methods to determine how much to withdraw, but you can use this idea with any withdrawal method. Investing During Retirement Cheers

big-papa,
Was curious that currently our withdrawals will be from cash and TIRA accounts, does the withdrawal strategy really matter in this case as one can easily true up/ buy/sell their AA in their TIRA without tax consequences or the concept of "selling stocks when they are low".