Allocation according to CAPE?

[RenoJay]

Is anyone aware of (and can share a link) on any studies that use index funds but rebalance according to metrics of stock market valuation as opposed to metrics of the investor’s personal asset allocation? (For instance, is there a model which would say something like: “If the CAPE > 30, reduce equity to just 40% of portfolio.”? I’d be curious to see the results.

 

[MichaelB]

Meb Faber has written a lot on CAPE, and even has some ETFs. His web site is here https://mebfaber.com/

 

[T-Minus]

GREAT question which I’ve been thinking about as well. Subscribed.

 

[sengsational]

I don't remember who, but someone on this board uses a formula that incorporates the CAPE ratio into their asset allocation percentages.


Mine isn't a formula, but I implemented a "step function" into my asset allocation scheme. It goes like this... if the "rock breaks scissors" sell rule is triggered, set my asset allocation to that of someone 30 years older (based on the Marotta Gone Fishing allocation). The sell was triggered September 1, 2015. So I haven't participated as much in the last few years' advances, but I had the pedal to the metal before. And of course I have the hard and fast rule to get me back in, if the time comes. I say "if" because there have only been nine signals in 138 years, lol!



I just searched and found a couple of old posts if you're interested:


http://www.early-retirement.org/for...t-performance-thread-84822-4.html#post1837897


http://www.early-retirement.org/forums/f28/using-shiller-pe-to-time-the-market-73166.html

 

[DrBrisket]

ERN has some research about CAPE based investing. https://earlyretirementnow.com/?s=Cape

Be sure to read the comments section as there are links that are also enlightening. I found the critical assessments of relying on accounting “principles” interesting along with the fact that the R^2 of CAPE regressions are about 1/3.

A deeply technical reference is Michael McClung’s book Living Off Your Money. McClung looks at CAPE in conjunction with other indicators to adjust asset allocation.

 

[samclem]

We've had some good discussions here on this topic. You can find context and links to academic studies here (oops, already provided by sengsational!) and here, plus some other spots.


IMO, using PE10 or CAPE to shift allocations is a worthwhile strategy with good academic backing. You'll find some indications that using govt bonds/notes rather than commercial bonds is the best approach when stock valuations get exceptionally high.
Anyone using this approach has to be prepared to be "wrong" for a long time, which is probably one reason it is not more popular among professional money managers (i.e. a good chance you'll get fired before your decision looks right). One can lose a lot of upside and a lot of dividends while waiting for the crash. Also, I would not be comfortable using this signal to move entirely in or out of stocks, but would instead use it to move to heavier/lighter weightings within my overall comfort zone based on stock valuations. So, maybe 70s/30b when stocks are cheap, 40s/60b when they are expensive.

Good luck!


Edited to add:

For a taste, see Table2 and Table 3 at the end of this paper.

Pfau paper on PE10

 

[USGrant1962]

We've had some good discussions here on this topic. You can find context and links to academic studies here and here, plus some other spots.


...

Anyone using this approach has to be prepared to be "wrong" for a long time, which is probably one reason it is not more popular among professional money managers (i.e. a good chance you'll get fired before your decision looks right). One can lose a lot of upside and a lot of dividends while waiting for the crash.

...

This. Except for a very short time after the 2008 crash, CAPE10 has been above 20 since 1992 - basically an investing lifetime. Shiller PE Ratio

 

[sengsational]

Above 20 isn't the problem, IMO. It's when it's above 30, like it has been. It's been above 30 a total of 4% of the time in the last 138 years, most of that was about 4 years starting in 1997, before the lackluster results of the 2000's. And now, since a year ago. Oh, and of course 1929.


But it stayed in the 30's for over 4 years before, so, as mentioned, you might have to wait a LONG time before you'll be proven right if you take this approach.

 

[audreyh1]

Kitces has a paper on Valuation-based Tactical Allocation which uses CAPE10: https://www.kitces.com/blog/valuati...on-on-declining-and-rising-equity-glidepaths/

Overall, though, the conclusions remain that whether based on safe withdrawal rates as a measure of minimizing risk, or median wealth as a measure of maximizing wealth accumulation, the valuation-based tactical portfolios (that start at 45% in equities but fluctuate between 30% and 60%) appear to do slightly better than either a fixed 60% equity exposure or a rising equity glidepath. When market valuation starts out unfavorable, the valuation-based portfolio starts out similar to a rising equity glidepath, but is able to more quickly adjust to higher equity allocations if valuation becomes favorable. By contrast, when valuations start out favorable, the valuation-based portfolio starts out similar to a fixed 60% equity portfolio, but becomes more conservative when necessary to avoid damaging drawdowns.

 

[audreyh1]

I don't remember who, but someone on this board uses a formula that incorporates the CAPE ratio into their asset allocation percentages.


Mine isn't a formula, but I implemented a "step function" into my asset allocation scheme. It goes like this... if the "rock breaks scissors" sell rule is triggered, set my asset allocation to that of someone 30 years older (based on the Marotta Gone Fishing allocation). The sell was triggered September 1, 2015. So I haven't participated as much in the last few years' advances, but I had the pedal to the metal before. And of course I have the hard and fast rule to get me back in, if the time comes. I say "if" because there have only been nine signals in 138 years, lol!



I just searched and found a couple of old posts if you're interested:


http://www.early-retirement.org/for...t-performance-thread-84822-4.html#post1837897


http://www.early-retirement.org/forums/f28/using-shiller-pe-to-time-the-market-73166.html

Me. It's a minor adjustment to my AA, just swinging between 50/50 and 60/40 based on CAPE10. And I use the CAPE10 values of 18 and 25 as my "limits". Between them it's a slope - so a gradual change in AA. These values are based on the past 25 years of CAPE 10, not the past 6 decades or more.

 

[RenoJay]

Above 20 isn't the problem, IMO. It's when it's above 30, like it has been. It's been above 30 a total of 4% of the time in the last 138 years, most of that was about 4 years starting in 1997, before the lackluster results of the 2000's. And now, since a year ago. Oh, and of course 1929.


But it stayed in the 30's for over 4 years before, so, as mentioned, you might have to wait a LONG time before you'll be proven right if you take this approach.

I'm the OP, and yep, this is pretty similar to how I've been doing things. Back in the late 90's, I had never heard of CAPE or PE10, but I was running a dotcom and realized what a ridiculous bubble the market was, so I really started jumping into equity investing after the inevitable crash. In recent years, I was around 80% equity during the 2008/2009 time frame when stocks were objectively cheap by historical measures. Now I'm 30% equity, and would go even lower except the cap gains taxes are killing me. Time will tell, but it's nice to see there are a few others here who factor overall market valuation into their decisions instead of just a static asset allocation.

 

[sengsational]

If you're having trouble sleeping some night, search "Tim Ferris" "Howard Marks", find the netcast, and listen to it. Marks generally says it's better to get average return in good markets and get less bad results in bad markets than beating the market in good times. So, yeah, he plays it defensively. I think it was August when the recording was made, and he was saying you can't predict the future, but you can take the temperature of the market, and he's saying it's pretty hot. Stops short of predicting the inevitable because it's not knowable.

 

[samclem]

Me. It's a minor adjustment to my AA, just swinging between 50/50 and 60/40 based on CAPE10. And I use the CAPE10 values of 18 and 25 as my "limits". Between them it's a slope - so a gradual change in AA. These values are based on the past 25 years of CAPE 10, not the past 6 decades or more.

Audrey, did you choose CAPE10 limits of 18 and 25 based on backtesting, or was it based on a look at the graph and "that looks about right"? Since the data set is rather limited and you are gradually sliding your allocation, I'd guess a look at the graph is probably just as valid as a more rigorous analysis. I'd think trying to "optimize" is just kidding ourselves, we're just trying to do a little better.
25 year look back based on relatively recent changes in accounting rules?

 

[haha]

Now I'm 30% equity, and would go even lower except the cap gains taxes are killing me. Time will tell, but it's nice to see there are a few others here who factor overall market valuation into their decisions instead of just a static asset allocation.

There are more than a few of us here. We just get tired of knee jerk condemnations of our deviation from scripture, so we keep our mouths shut. People with atypical positions know that these may not work out. People following the scripture are often sure that they are right. Until the market seems to perhaps not know the scripture.

Anyway, this time around I expect unprecedented government interference. Markets as important as US stocks and bonds will not be left free when there is stress.

Ha

 

[audreyh1]

Audrey, did you choose CAPE10 limits of 18 and 25 based on backtesting, or was it based on a look at the graph and "that looks about right"? Since the data set is rather limited and you are gradually sliding your allocation, I'd guess a look at the graph is probably just as valid as a more rigorous analysis. I'd think trying to "optimize" is just kidding ourselves, we're just trying to do a little better.
25 year look back based on relatively recent changes in accounting rules?

Looked at graph but also aware of accounting rule changes. I figured it was “close enough” and didn’t need to be that precise. CAPE10 has been higher, and often much higher, than mean/median most of my investing life.

 

[NW-Bound]

I don't remember who, but someone on this board uses a formula that incorporates the CAPE ratio into their asset allocation percentages...

I definitely remember FIREd posting his strategy, which I may not be able to find again.

 

[Fedup]

If you're having trouble sleeping some night, search "Tim Ferris" "Howard Marks", find the netcast, and listen to it. Marks generally says it's better to get average return in good markets and get less bad results in bad markets than beating the market in good times. So, yeah, he plays it defensively. I think it was August when the recording was made, and he was saying you can't predict the future, but you can take the temperature of the market, and he's saying it's pretty hot. Stops short of predicting the inevitable because it's not knowable.

I’m glad I read this. Somebody out there agrees with me.

 

[samclem]

Some observations on the current PE10:
1) Some analysts have noted that the present CAPE (PE10, which is 35.09) is high partly because of the low corporate earnings in the aftermath of the 2008 financial crisis, and that as these bad years fall off the back of the 10 year data set, the average earnings would go up and the CAPE would go down, which would indicate maybe stocks aren't quite so overvalued . I looked at what the "PE8" would be--stock price /inflation adjusted earnings for only the last 8 years, and that number is 31.23. So, if over the next 2 years, stock prices and earnings stay relatively stable, it is true that the CAPE will decrease into "slightly less overvalued" territory.
but:
2) The "PE1" (includes only last year's earnings) is over 55. Maybe that's some sort of statistical/calculation artifact. Here are the other "PEs, rounded to the nearest whole number:"
PE10: 35
PE9: 32
PE8: 31
PE7: 31
PE6: 31
PE5: 32
PE4: 33
PE3: 35
PE2: 38
PE1: 55

For geeks: The CAPE calculator I found at this site makes it easy to do this sort of lookback. Unfortunately, the site/calculator does >not< include a function to find the mean/median CAPE over various time periods, it only figures it for the whole data set (1873(?) to present). The site does offer CSV data of the Schiller series for those who want to torture the data even more in their own spreadsheet.

 

[big-papa]

This topic has been discussed over on Bogleheads a few times, but a very interesting thread is this one:
https://www.bogleheads.org/forum/viewtopic.php?t=168852

Pay attention to some of the work from the poster named bh7. He's doing backtesting based only on available data. In other words, to do a fair backtest, if this year is 1985, you shouldn't include any data in your backtest from years beyond that year since in 1985, you didn't have a crystal ball. When done this way, he shows that the CAPE mean is also a roller coaster...

I don't see any earth shattering excess returns here and if your stocks are in taxable accounts, there's a good chance that whatever excess returns you do achieve could be wiped out by taxes, depending on your situation.

Also pay attention to some comments from siamond regarding max correlation of CAPE to future returns (peaks somewhere 10-15 years out).

 

[USGrant1962]

2) The "PE1" (includes only last year's earnings) is over 55. Maybe that's some sort of statistical/calculation artifact. Here are the other "PEs, rounded to the nearest whole number:"
PE10: 35
PE9: 32
PE8: 31
PE7: 31
PE6: 31
PE5: 32
PE4: 33
PE3: 35
PE2: 38
PE1: 55

For geeks: The CAPE calculator I found at this site makes it easy to do this sort of lookback. Unfortunately, the site/calculator does >not< include a function to find the mean/median CAPE over various time periods, it only figures it for the whole data set (1873(?) to present). The site does offer CSV data of the Schiller series for those who want to torture the data even more in their own spreadsheet.

I've used that calculator too, and like it. But there is something wrong with how they calculate PE1. TTM PE for the S&P 500 is currently 23.5, not 55.

 

[samclem]

I've used that calculator too, and like it. But there is something wrong with how they calculate PE1. TTM PE for the S&P 500 is currently 23.5, not 55.

Thanks, that's what I figured. Maybe only partial earnings data for the last year (denominator), but the data set for stock prices is complete (numerator), so the number is in error to the high side.

 

[sengsational]

Interesting about the PE8. I've "tortured" the Schiller data a bit myself (lol), but always stuck with 10 years.



I'll admit to only applying brief first level thinking on the shorter analysis span, but it seems not to give any momentum to the "it's different this time" position, does it?

 

[samclem]

I'll admit to only applying brief first level thinking on the shorter analysis span, but it seems not to give any momentum to the "it's different this time" position, does it?

No, I don't think so. If anything, it excuses the need to try to rationalize the present high PE10--"it's not anything unusual, stock prices aren't as igh as they seem. Look what happens when we drop those low earnings from 2008 and 2009. See! It is heading back to the mean. Nothing to see here--keep buying!"

Interestingly, when we lengthen the PE "smoothing period," the average for the whole period goes up slightly. For example, with the same data set (1873 to present), the:
PE2 average is 15.7
PE5 average is 16.0
PE10 average is 16.9
PE15 average is 17.7
PE20 average is 18.4

I don't know enough about how the number is calculated to explain why this is so. It might be due to the overall upward trend in prices over the data set, or the difference in the "slope" of uptrends vs downtrends. But there it is: if you cut the series into smaller chunks, stocks get cheaper relative to earnings.

 

[audreyh1]

Some observations on the current PE10:
1) Some analysts have noted that the present CAPE (PE10, which is 35.09) is high partly because of the low corporate earnings in the aftermath of the 2008 financial crisis, and that as these bad years fall off the back of the 10 year data set, the average earnings would go up and the CAPE would go down, which would indicate maybe stocks aren't quite so overvalued . I looked at what the "PE8" would be--stock price /inflation adjusted earnings for only the last 8 years, and that number is 31.23. So, if over the next 2 years, stock prices and earnings stay relatively stable, it is true that the CAPE will decrease into "slightly less overvalued" territory.
but:
2) The "PE1" (includes only last year's earnings) is over 55. Maybe that's some sort of statistical/calculation artifact. Here are the other "PEs, rounded to the nearest whole number:"
PE10: 35
PE9: 32
PE8: 31
PE7: 31
PE6: 31
PE5: 32
PE4: 33
PE3: 35
PE2: 38
PE1: 55

For geeks: The CAPE calculator I found at this site makes it easy to do this sort of lookback. Unfortunately, the site/calculator does >not< include a function to find the mean/median CAPE over various time periods, it only figures it for the whole data set (1873(?) to present). The site does offer CSV data of the Schiller series for those who want to torture the data even more in their own spreadsheet.

Right - I did those calcs a year ago, looking at PE7, and saw the difference was ~4 points, not nearly enough to get us out of well overvalued territory.

 

[TwoByFour]

Deleted. Misread OP