1/CAPE10 at the extremes

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
GAAP
Posts: 591
Joined: Fri Apr 08, 2016 12:41 pm

1/CAPE10 at the extremes

Post by GAAP » Thu Jun 28, 2018 2:02 pm

I would like to hear from those who use 1/CAPE10 to predict earnings (or anything else). Since that formula is not linear, how do you handle the extremes in values?

For example, if CAPE10 is 10, 1/CAPE10 yields a 10% real expected return. In 1921, CAPE10 got down to 5.3, predicting 19% real returns -- the extreme of 1 would predict 100% returns. The other end of the scale approaches, yet never reaches zero -- but always expecting positive real returns seems unrealistic, and basing a withdrawal on near-zero returns would clearly not work for long.

I've lately been playing with limiting the range of values I use -- outliers get calculated at my limits. It seems to work for "normal" cases

Alternate solutions anyone?

User avatar
willthrill81
Posts: 5823
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: 1/CAPE10 at the extremes

Post by willthrill81 » Thu Jun 28, 2018 3:36 pm

GAAP wrote:
Thu Jun 28, 2018 2:02 pm
but always expecting positive real returns seems unrealistic,
Why? In over 85% of the historic 10 year periods, returns for the S&P 500 were positive. Valuations alone are not good at predicting negative returns.
GAAP wrote:
Thu Jun 28, 2018 2:02 pm
and basing a withdrawal on near-zero returns would clearly not work for long.
Yes, it could. From 2000-2009, a 60/40 portfolio of TSM and TBM had a real return of just .38%. Add 4% fixed withdrawals to that and the portfolio went down in real terms by 39% over that period. Yet those who retired in the year 2000 with such a portfolio and made '4% rule' withdrawals are doing fine now with about 69% of their starting portfolio in real dollars still intact after 18 years of withdrawals.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

User avatar
siamond
Posts: 4187
Joined: Mon May 28, 2012 5:50 am

Re: 1/CAPE10 at the extremes

Post by siamond » Thu Jun 28, 2018 4:42 pm

1/CAPE is a reasonable model for expected (total, inflation-adjusted) returns over the long-term (10 to 20 years). Expected returns has to be understood in the statistical sense, the epicenter of a wide cloud of possibilities. If you try to use it as a specific forecast, you will be very disappointed for sure. If you use it as part of a sound self-correcting process (e.g. dynamic withdrawals), then this can present significant value.

OP, please think about your questions while being careful about the timeframe and the statistical nature of the metric. (US) stocks real returns never went negative over a period of 20 years. They did hover close to zero in some periods of 15 years, but a specific outcome isn't the epicenter of possibilities. They definitely went over 10% quite a few times for 15 years and 20 years time periods. Check the portfolio cycles charts (with CAGR Real as metric) to convince yourself. Here is an example with 15 years cycles based on numerous starting years.

Image

Then this logic needs to be applied to an entire portfolio. You probably don't have 100% US stocks in your portfolio if you're retired. What matters is the expected return of the portfolio, including its International and Bonds components. This will smooth out the outcomes, depending on your AA. This being said, yes, if you do the math near the peak of the Internet crisis (US) or the 1929 big bull market, you'll find very low expected returns (rightfully). The portfolio value is extremely elevated by then though. Combining the low and high values in a withdrawal formula will end up with a reasonable outcome. And the reverse way around at the bottom of a deep crisis. I see no reason to try to normalize the 1/CAPE formula. You may want to normalize the outcome of a withdrawal formula though (e.g. with a floor for your barebone expenses), but that's a separate discussion.

GAAP
Posts: 591
Joined: Fri Apr 08, 2016 12:41 pm

Re: 1/CAPE10 at the extremes

Post by GAAP » Thu Jun 28, 2018 7:09 pm

I will admit to some sloppiness in the question...

My primary concern at low CAPE values is the rapid change in the slope. I do expect accelerated differences -- but at some point that acceleration is tough to buy. Even when charted with a log function, the curve is notably non-linear -- Ln(1/1) to Ln(1/5) looks a lot different than L(1/5) to Ln(1/10).

The limited information I've seen on international CAPE is that some countries went well beyond 40 -- and the corresponding returns were actually negative. There is no value of 1/CAPE that would mimic that result, since even 1/infinity is still positive.
willthrill81 wrote:
Thu Jun 28, 2018 3:36 pm
In over 85% of the historic 10 year periods, returns for the S&P 500 were positive.
Which means that 1/CAPE is unable to predict 15% of the cases...

To be clear: I do use 1/CAPE, albeit with some constraints. I do use it only for the equity portion of my AA, not fixed income, and I am concerned with a globally-allocated portfolio, not just domestically. And yes, the rate is used as part of an internal trend for a PMT function which will naturally adjust for errors in estimation -- safely, if the errors are reasonable. I'm hoping that my constraints are more conservative in a down market, allowing more time to recover. The differences at the high end are far less extreme, and reasonably handled by the PMT function.

User avatar
siamond
Posts: 4187
Joined: Mon May 28, 2012 5:50 am

Re: 1/CAPE10 at the extremes

Post by siamond » Thu Jun 28, 2018 9:05 pm

GAAP wrote:
Thu Jun 28, 2018 7:09 pm
The limited information I've seen on international CAPE is that some countries went well beyond 40 -- and the corresponding returns were actually negative. There is no value of 1/CAPE that would mimic that result, since even 1/infinity is still positive.
I think you're alluding at Japan here. I agree with you that Japan's CAPE trajectory is hard to swallow. From the little I understand, this is due to a practice which is very unique to Japan, where companies do a LOT of cross-investments (buying shares) with each other, artificially elevating valuations. This seems peculiar enough to not disprove any regular use of CAPE in the rest of the world, but I don't quite understand the next level of details, to be honest.

Overall, valuations are very blunt tools, no question about it. This is why it is so crucial to only use them in a solid self-correcting long-term context, and nowhere else... Personally, I also avoid to make *any* assumption about some sort of return-to-the-mean of said valuations, because the historical mean varied quite significantly over time, and I don't see why the mean as of today has special semantics. Bottomline: either use 1/CAPE in a very simple manner, with a minimum set of assumptions, or don't use valuations at all. Any further complication will just create garbage math.

User avatar
siamond
Posts: 4187
Joined: Mon May 28, 2012 5:50 am

Re: 1/CAPE10 at the extremes

Post by siamond » Thu Jun 28, 2018 9:08 pm

GAAP wrote:
Thu Jun 28, 2018 7:09 pm
willthrill81 wrote:
Thu Jun 28, 2018 3:36 pm
In over 85% of the historic 10 year periods, returns for the S&P 500 were positive.
Which means that 1/CAPE is unable to predict 15% of the cases...
Not true. Again, 1/CAPE is NOT exact predictive science. This is what trips a lot of CAPE critics on this board. An expected returns model is statistical, an epicenter in a range of possibilities. Not a specific forecast by any mean.

letsgobobby
Posts: 11590
Joined: Fri Sep 18, 2009 1:10 am

Re: 1/CAPE10 at the extremes

Post by letsgobobby » Thu Jun 28, 2018 10:18 pm

There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.

Valuethinker
Posts: 36024
Joined: Fri May 11, 2007 11:07 am

Re: 1/CAPE10 at the extremes

Post by Valuethinker » Fri Jun 29, 2018 7:36 am

siamond wrote:
Thu Jun 28, 2018 9:05 pm
GAAP wrote:
Thu Jun 28, 2018 7:09 pm
The limited information I've seen on international CAPE is that some countries went well beyond 40 -- and the corresponding returns were actually negative. There is no value of 1/CAPE that would mimic that result, since even 1/infinity is still positive.
I think you're alluding at Japan here. I agree with you that Japan's CAPE trajectory is hard to swallow. From the little I understand, this is due to a practice which is very unique to Japan, where companies do a LOT of cross-investments (buying shares) with each other, artificially elevating valuations. This seems peculiar enough to not disprove any regular use of CAPE in the rest of the world, but I don't quite understand the next level of details, to be honest.
I think we might find there are quite a few markets with these sorts of issues re cross holdings. I am thinking China (but I really don't know). Chinese banks (stockmarket listed) lend money which in turn is used to buy stocks (or to buy property and the cash released is used to buy stocks), etc.

The American version of this is pension funds and other institutional investors own large amounts of stock in companies, but are themselves associated with those companies. Or situations like Softbank which has raised this $100bn fund to invest in unquoted tech companies. A bit like the interbank lending system which collapsed during the Crash (apparently Eurodollar deposits were key in this - the dollar borrowings of European banks).

GAAP
Posts: 591
Joined: Fri Apr 08, 2016 12:41 pm

Re: 1/CAPE10 at the extremes

Post by GAAP » Fri Jun 29, 2018 9:11 am

siamond wrote:
Thu Jun 28, 2018 9:08 pm
GAAP wrote:
Thu Jun 28, 2018 7:09 pm
willthrill81 wrote:
Thu Jun 28, 2018 3:36 pm
In over 85% of the historic 10 year periods, returns for the S&P 500 were positive.
Which means that 1/CAPE is unable to predict 15% of the cases...
Not true. Again, 1/CAPE is NOT exact predictive science. This is what trips a lot of CAPE critics on this board. An expected returns model is statistical, an epicenter in a range of possibilities. Not a specific forecast by any mean.
That actually is my point. It's a tool that provides a reasonable estimate under most conditions. But most conditions does not mean always -- which implies that at some point, you need to consider the tool less appropriate. I'm simply proposing that the extreme values may generate less appropriate results.

User avatar
willthrill81
Posts: 5823
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: 1/CAPE10 at the extremes

Post by willthrill81 » Fri Jun 29, 2018 9:37 am

letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

GAAP
Posts: 591
Joined: Fri Apr 08, 2016 12:41 pm

Re: 1/CAPE10 at the extremes

Post by GAAP » Fri Jun 29, 2018 9:51 am

willthrill81 wrote:
Fri Jun 29, 2018 9:37 am
letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
Agreed. I understand using multiple measures to refine an estimate -- but my distinct impression was that they refine it the other way (toward greater error). I guess that makes good marketing sense, and it probably helps sell the need for professional help. I don't see that being very helpful for me.

User avatar
siamond
Posts: 4187
Joined: Mon May 28, 2012 5:50 am

Re: 1/CAPE10 at the extremes

Post by siamond » Fri Jun 29, 2018 9:58 am

GAAP wrote:
Fri Jun 29, 2018 9:11 am
That actually is my point. It's a tool that provides a reasonable estimate under most conditions. But most conditions does not mean always -- which implies that at some point, you need to consider the tool less appropriate. I'm simply proposing that the extreme values may generate less appropriate results.
Well, I gave you multiple data points showing that this was not really the case in the US history, see my 15yrs chart a few posts ago. The actual outcomes from extremes were well inside a reasonable range of possible outcomes. Note that I used 15 years horizons because 1/CAPE is a tad better for such horizons (or at least has been in the past). Other countries are much harder to judge, by lack of history, which is unfortunate, as solid out-of-sample testing would be very welcome for this stuff.

This being said, I don't disagree with your assertion that sometimes the tool is really quite off. The main counter-example (besides Japan) isn't about extreme values of CAPE. It's the years preceding the oil crisis, in the US. Stocks did NOT appear terribly over-valued by then. And yet the late 60s and early 70s started some of the worst periods for a US investor. 1929, 2002, 2009 (in the US) and 1990 (in Japan) crises were quite clearly (in hindsight, mind you) speculative bubbles. 1973 was not, it was much more of a price/inflation crisis. Valuations do NOT have full explanatory power, far from it.

So yeah, the tool is very blunt and can be quite off. But I wouldn't try to 'fix it' (probably adding more issues than improvements in the process), I would just accept it for what it is, a probabilistic model, a definite improvement over historical averages, and yet very far from perfect. Where I would apply common sense and normalization would be the outcomes of derivative math, like withdrawal formulas (e.g. hard floor or hard ceiling or smoothing from past year), for very practical real-life reasons.

User avatar
siamond
Posts: 4187
Joined: Mon May 28, 2012 5:50 am

Re: 1/CAPE10 at the extremes

Post by siamond » Fri Jun 29, 2018 11:10 am

letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
Here is the thread. The part about Buckingham’s expected returns vs. reality is really a joke, but the core point is of course correct (er, by sheer definition of the term 'expected returns'!).

letsgobobby
Posts: 11590
Joined: Fri Sep 18, 2009 1:10 am

Re: 1/CAPE10 at the extremes

Post by letsgobobby » Sat Jun 30, 2018 12:47 am

willthrill81 wrote:
Fri Jun 29, 2018 9:37 am
letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
For example, at the extreme of an elevated CAPE, to paraphrase the article and my understanding, the median expected return shifts from, say, +10% with a +/-5% 90% CI (completely made up numbers), to +3% with a strong left skew. Given this, at extremely high valuations in CAPE, I could save more, defer retirement, sell equities, etc. So I find it very helpful.

My IPS requires I react to CAPEs>30. The only reason I haven’t is that a valid argument can be made that today’s CAPE is not directly comparable to the CAPE from 15 years ago, and thus today’s 32.2 is more like yesteryear’s 27. And when the 2008 earnings drop off in six more months, that should bump down CAPE another 10% or so. So CAPE looks like 24-25. That is the only reason I haven’t substantially reduced US equity exposure despite the seemingly high CAPE.

marcopolo
Posts: 1126
Joined: Sat Dec 03, 2016 10:22 am

Re: 1/CAPE10 at the extremes

Post by marcopolo » Sat Jun 30, 2018 9:02 am

letsgobobby wrote:
Sat Jun 30, 2018 12:47 am
willthrill81 wrote:
Fri Jun 29, 2018 9:37 am
letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
For example, at the extreme of an elevated CAPE, to paraphrase the article and my understanding, the median expected return shifts from, say, +10% with a +/-5% 90% CI (completely made up numbers), to +3% with a strong left skew. Given this, at extremely high valuations in CAPE, I could save more, defer retirement, sell equities, etc. So I find it very helpful.

My IPS requires I react to CAPEs>30. The only reason I haven’t is that a valid argument can be made that today’s CAPE is not directly comparable to the CAPE from 15 years ago, and thus today’s 32.2 is more like yesteryear’s 27. And when the 2008 earnings drop off in six more months, that should bump down CAPE another 10% or so. So CAPE looks like 24-25. That is the only reason I haven’t substantially reduced US equity exposure despite the seemingly high CAPE.
When did you create your your IPS? Were the accounting changes that affected CAPE and the poor 2008 earning already known at that time?
Once in a while you get shown the light, in the strangest of places if you look at it right.

User avatar
willthrill81
Posts: 5823
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: 1/CAPE10 at the extremes

Post by willthrill81 » Sat Jun 30, 2018 9:03 am

letsgobobby wrote:
Sat Jun 30, 2018 12:47 am
willthrill81 wrote:
Fri Jun 29, 2018 9:37 am
letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
For example, at the extreme of an elevated CAPE, to paraphrase the article and my understanding, the median expected return shifts from, say, +10% with a +/-5% 90% CI (completely made up numbers), to +3% with a strong left skew. Given this, at extremely high valuations in CAPE, I could save more, defer retirement, sell equities, etc. So I find it very helpful.
I suppose that I'm just not convinced that CAPE alone can accurately predict that the median expected return would shift downward from something like 10% to 3%. Kitces found that historically low and high starting CAPE values impacted the subsequent 30 year returns by only +/- 1%, and I'm more concerned about what happens over that time frame than the next decade.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

GAAP
Posts: 591
Joined: Fri Apr 08, 2016 12:41 pm

Re: 1/CAPE10 at the extremes

Post by GAAP » Sat Jun 30, 2018 9:31 am

willthrill81 wrote:
Sat Jun 30, 2018 9:03 am
I suppose that I'm just not convinced that CAPE alone can accurately predict that the median expected return would shift downward from something like 10% to 3%. Kitces found that historically low and high starting CAPE values impacted the subsequent 30 year returns by only +/- 1%, and I'm more concerned about what happens over that time frame than the next decade.
Accurately predict, no. Reasonably estimate is something else. To me, it makes general sense that the more expensive stocks get, the less return you're likely to make. I don't need precision, but I'm a lot more comfortable with a general direction than a historical average.

30-year returns could easily be two complete cycles. That matters for me, since I'm not using a classic 30-year 4% plan. Even if I was, I would still care about whether I was heading into a downturn, since that really means that my "number" should be higher to account for that potential.

letsgobobby
Posts: 11590
Joined: Fri Sep 18, 2009 1:10 am

Re: 1/CAPE10 at the extremes

Post by letsgobobby » Sat Jun 30, 2018 10:39 am

marcopolo wrote:
Sat Jun 30, 2018 9:02 am
letsgobobby wrote:
Sat Jun 30, 2018 12:47 am
willthrill81 wrote:
Fri Jun 29, 2018 9:37 am
letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
For example, at the extreme of an elevated CAPE, to paraphrase the article and my understanding, the median expected return shifts from, say, +10% with a +/-5% 90% CI (completely made up numbers), to +3% with a strong left skew. Given this, at extremely high valuations in CAPE, I could save more, defer retirement, sell equities, etc. So I find it very helpful.

My IPS requires I react to CAPEs>30. The only reason I haven’t is that a valid argument can be made that today’s CAPE is not directly comparable to the CAPE from 15 years ago, and thus today’s 32.2 is more like yesteryear’s 27. And when the 2008 earnings drop off in six more months, that should bump down CAPE another 10% or so. So CAPE looks like 24-25. That is the only reason I haven’t substantially reduced US equity exposure despite the seemingly high CAPE.
When did you create your your IPS? Were the accounting changes that affected CAPE and the poor 2008 earning already known at that time?
Apparently they were known, but not by me.

letsgobobby
Posts: 11590
Joined: Fri Sep 18, 2009 1:10 am

Re: 1/CAPE10 at the extremes

Post by letsgobobby » Sat Jun 30, 2018 11:01 am

willthrill81 wrote:
Sat Jun 30, 2018 9:03 am
letsgobobby wrote:
Sat Jun 30, 2018 12:47 am
willthrill81 wrote:
Fri Jun 29, 2018 9:37 am
letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
For example, at the extreme of an elevated CAPE, to paraphrase the article and my understanding, the median expected return shifts from, say, +10% with a +/-5% 90% CI (completely made up numbers), to +3% with a strong left skew. Given this, at extremely high valuations in CAPE, I could save more, defer retirement, sell equities, etc. So I find it very helpful.
I suppose that I'm just not convinced that CAPE alone can accurately predict that the median expected return would shift downward from something like 10% to 3%. Kitces found that historically low and high starting CAPE values impacted the subsequent 30 year returns by only +/- 1%, and I'm more concerned about what happens over that time frame than the next decade.
In the thread I referenced and siamond linked, Larry’s research showed CAPE10> 25 had a mean 0.3% real return over 10 years with about a +/- 6% range. In contrast CAPE10 <9.6 had mean real returns of 10.3%. This is a huge difference. It suggests that when CAPE10 is very high, not only will subsequent real returns be nearly zero, on average, but on average there will be periods of losses during which you can buy stocks back at a lower real price; and furthermore that while you are waiting you can earn almost the same expected real return from bonds, with minimal risk of volatility. That is the origin of my IPS.

User avatar
grayfox
Posts: 4933
Joined: Sat Sep 15, 2007 4:30 am

Re: 1/CAPE10 at the extremes

Post by grayfox » Sat Jun 30, 2018 11:46 am

If there is average earnings growth over the next 10 years, and no change valuation, 1/CAPE10 is a quite accurate predictor of future returns. The big wild card is future valuations.

So the obvious next step in the art of forecasting returns is to forecast future valuations. Today CAPE10 is 32.30. If you knew what CAPE10 will be in 10 years, your 10-year forecast will be more accurate. What are the predictions for CAPE10 in 30-Jun-2028?

Alrighty then, somebody get on it one time, eh.

User avatar
willthrill81
Posts: 5823
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: 1/CAPE10 at the extremes

Post by willthrill81 » Sat Jun 30, 2018 1:10 pm

letsgobobby wrote:
Sat Jun 30, 2018 11:01 am
willthrill81 wrote:
Sat Jun 30, 2018 9:03 am
letsgobobby wrote:
Sat Jun 30, 2018 12:47 am
willthrill81 wrote:
Fri Jun 29, 2018 9:37 am
letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
For example, at the extreme of an elevated CAPE, to paraphrase the article and my understanding, the median expected return shifts from, say, +10% with a +/-5% 90% CI (completely made up numbers), to +3% with a strong left skew. Given this, at extremely high valuations in CAPE, I could save more, defer retirement, sell equities, etc. So I find it very helpful.
I suppose that I'm just not convinced that CAPE alone can accurately predict that the median expected return would shift downward from something like 10% to 3%. Kitces found that historically low and high starting CAPE values impacted the subsequent 30 year returns by only +/- 1%, and I'm more concerned about what happens over that time frame than the next decade.
In the thread I referenced and siamond linked, Larry’s research showed CAPE10> 25 had a mean 0.3% real return over 10 years with about a +/- 6% range. In contrast CAPE10 <9.6 had mean real returns of 10.3%. This is a huge difference. It suggests that when CAPE10 is very high, not only will subsequent real returns be nearly zero, on average, but on average there will be periods of losses during which you can buy stocks back at a lower real price; and furthermore that while you are waiting you can earn almost the same expected real return from bonds, with minimal risk of volatility. That is the origin of my IPS.
That sounds reasonable. Although adjusting one's investment strategy based on market conditions is market timing, I personally have no problem with that being that I'm a trend follower.
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

User avatar
siamond
Posts: 4187
Joined: Mon May 28, 2012 5:50 am

Re: 1/CAPE10 at the extremes

Post by siamond » Sat Jun 30, 2018 2:10 pm

grayfox wrote:
Sat Jun 30, 2018 11:46 am
If there is average earnings growth over the next 10 years, and no change valuation, 1/CAPE10 is a quite accurate predictor of future returns. The big wild card is future valuations.
Hm, where did you get this idea? The typical research (which I reproduced myself multiple times) is that 1/CAPE10 is a fairly decent expected returns model of future actual returns (over 10 years or more) -- assuming we use a proper definition of 'expected returns' (i.e. in the statistical sense). Note that I did NOT say anything about future earnings or future valuations.

I know that Larry keeps making weird assumptions about future valuations not changing, but this runs counter to any basic common sense (try telling that to a 1929 or 2002 retiree! or to Prof. Shiller!). More to the point, I don't believe he provided any research about 1/CAPE in the time periods where earnings growth was average and valuations didn't change (heck, was there ANY such period?). Were you thinking to something else?

User avatar
grayfox
Posts: 4933
Joined: Sat Sep 15, 2007 4:30 am

Re: 1/CAPE10 at the extremes

Post by grayfox » Sat Jun 30, 2018 4:26 pm

siamond wrote:
Sat Jun 30, 2018 2:10 pm
grayfox wrote:
Sat Jun 30, 2018 11:46 am
If there is average earnings growth over the next 10 years, and no change valuation, 1/CAPE10 is a quite accurate predictor of future returns. The big wild card is future valuations.
Hm, where did you get this idea? The typical research (which I reproduced myself multiple times) is that 1/CAPE10 is a fairly decent expected returns model of future actual returns (over 10 years or more) -- assuming we use a proper definition of 'expected returns' (i.e. in the statistical sense). Note that I did NOT say anything about future earnings or future valuations.

I know that Larry keeps making weird assumptions about future valuations not changing, but this runs counter to any basic common sense (try telling that to a 1929 or 2002 retiree! or to Prof. Shiller!). More to the point, I don't believe he provided any research about 1/CAPE in the time periods where earnings growth was average and valuations didn't change (heck, was there ANY such period?). Were you thinking to something else?
Where does Total Return come from? Bogle's Formula:
TR = Dividend Yield + Earnings Growth + Change in Valuation

Previous Thread: Whence S&P500 Return past 10 years?

User avatar
siamond
Posts: 4187
Joined: Mon May 28, 2012 5:50 am

Re: 1/CAPE10 at the extremes

Post by siamond » Sat Jun 30, 2018 5:22 pm

grayfox wrote:
Sat Jun 30, 2018 4:26 pm
Where does Total Return come from? Bogle's Formula:
TR = Dividend Yield + Earnings Growth + Change in Valuation
Oh, silly me. And yes, it doesn't take much algebra to get to E/P from there. I actually did exactly that in a blog article a while ago. Sorry for being slow, I was thinking empirical, you were thinking common sense... There are three variable parameters in this equation though, it's more than valuation.

MathWizard
Posts: 3005
Joined: Tue Jul 26, 2011 1:35 pm

Re: 1/CAPE10 at the extremes

Post by MathWizard » Tue Oct 09, 2018 5:27 pm

GAAP wrote:
Thu Jun 28, 2018 2:02 pm
I would like to hear from those who use 1/CAPE10 to predict earnings (or anything else). Since that formula is not linear, how do you handle the extremes in values?

For example, if CAPE10 is 10, 1/CAPE10 yields a 10% real expected return. In 1921, CAPE10 got down to 5.3, predicting 19% real returns -- the extreme of 1 would predict 100% returns. The other end of the scale approaches, yet never reaches zero -- but always expecting positive real returns seems unrealistic, and basing a withdrawal on near-zero returns would clearly not work for long.

I've lately been playing with limiting the range of values I use -- outliers get calculated at my limits. It seems to work for "normal" cases

Alternate solutions anyone?

I disagree with your statement:
the other end of the scale approaches, yet never reaches zero
If earning are always positive, then returns should be positive eventually.

If earnings are negative, then CAPE10 would be negative, and predict negative returns.

Though it has not had a good track record, 1/CAPE10 was supposed to be a predictor of average 10 year returns, not 1 year returns.

GAAP
Posts: 591
Joined: Fri Apr 08, 2016 12:41 pm

Re: 1/CAPE10 at the extremes

Post by GAAP » Tue Oct 09, 2018 5:49 pm

MathWizard wrote:
Tue Oct 09, 2018 5:27 pm
GAAP wrote:
Thu Jun 28, 2018 2:02 pm
I would like to hear from those who use 1/CAPE10 to predict earnings (or anything else). Since that formula is not linear, how do you handle the extremes in values?

For example, if CAPE10 is 10, 1/CAPE10 yields a 10% real expected return. In 1921, CAPE10 got down to 5.3, predicting 19% real returns -- the extreme of 1 would predict 100% returns. The other end of the scale approaches, yet never reaches zero -- but always expecting positive real returns seems unrealistic, and basing a withdrawal on near-zero returns would clearly not work for long.

I've lately been playing with limiting the range of values I use -- outliers get calculated at my limits. It seems to work for "normal" cases

Alternate solutions anyone?

I disagree with your statement:
the other end of the scale approaches, yet never reaches zero
If earning are always positive, then returns should be positive eventually.

If earnings are negative, then CAPE10 would be negative, and predict negative returns.

Though it has not had a good track record, 1/CAPE10 was supposed to be a predictor of average 10 year returns, not 1 year returns.
CAPE10 is a moving 10-year average, so earnings would need to be very negative and/or negative for a long time to actually create a negative CAPE10. I don't think we have any tools for sort of situation, short of investing in something else...

Any positive CAPE10 value will generate a positive inverse. However, I don't see a lot of practical difference between 1/100 and 1/200 -- a 0.5% real difference over 10-15 years is close enough to zero for me.

I don't believe I said anywhere that CAPE10 predicts 1 year returns -- and I've said elsewhere that I use it for 10-15 years estimates, knowing that there is still a degree of inaccuracy.

petulant
Posts: 327
Joined: Thu Sep 22, 2016 1:09 pm

Re: 1/CAPE10 at the extremes

Post by petulant » Tue Oct 09, 2018 6:42 pm

willthrill81 wrote:
Sat Jun 30, 2018 9:03 am
letsgobobby wrote:
Sat Jun 30, 2018 12:47 am
willthrill81 wrote:
Fri Jun 29, 2018 9:37 am
letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
For example, at the extreme of an elevated CAPE, to paraphrase the article and my understanding, the median expected return shifts from, say, +10% with a +/-5% 90% CI (completely made up numbers), to +3% with a strong left skew. Given this, at extremely high valuations in CAPE, I could save more, defer retirement, sell equities, etc. So I find it very helpful.
I suppose that I'm just not convinced that CAPE alone can accurately predict that the median expected return would shift downward from something like 10% to 3%. Kitces found that historically low and high starting CAPE values impacted the subsequent 30 year returns by only +/- 1%, and I'm more concerned about what happens over that time frame than the next decade.
Can you link the Kitces study?

User avatar
willthrill81
Posts: 5823
Joined: Thu Jan 26, 2017 3:17 pm
Location: USA

Re: 1/CAPE10 at the extremes

Post by willthrill81 » Tue Oct 09, 2018 9:22 pm

petulant wrote:
Tue Oct 09, 2018 6:42 pm
willthrill81 wrote:
Sat Jun 30, 2018 9:03 am
letsgobobby wrote:
Sat Jun 30, 2018 12:47 am
willthrill81 wrote:
Fri Jun 29, 2018 9:37 am
letsgobobby wrote:
Thu Jun 28, 2018 10:18 pm
There is another recent thread about Larry's article describing how CAPE predicts a probability of a range of returns. That's how I use it, to the extent I use it at all.
I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
For example, at the extreme of an elevated CAPE, to paraphrase the article and my understanding, the median expected return shifts from, say, +10% with a +/-5% 90% CI (completely made up numbers), to +3% with a strong left skew. Given this, at extremely high valuations in CAPE, I could save more, defer retirement, sell equities, etc. So I find it very helpful.
I suppose that I'm just not convinced that CAPE alone can accurately predict that the median expected return would shift downward from something like 10% to 3%. Kitces found that historically low and high starting CAPE values impacted the subsequent 30 year returns by only +/- 1%, and I'm more concerned about what happens over that time frame than the next decade.
Can you link the Kitces study?
Here it is.

Image
https://www.kitces.com/blog/should-equi ... valuation/
“It's a dangerous business, Frodo, going out your door. You step onto the road, and if you don't keep your feet, there's no knowing where you might be swept off to.” J.R.R. Tolkien,The Lord of the Rings

petulant
Posts: 327
Joined: Thu Sep 22, 2016 1:09 pm

Re: 1/CAPE10 at the extremes

Post by petulant » Thu Oct 11, 2018 9:49 pm

willthrill81 wrote:
Tue Oct 09, 2018 9:22 pm
petulant wrote:
Tue Oct 09, 2018 6:42 pm
willthrill81 wrote:
Sat Jun 30, 2018 9:03 am
letsgobobby wrote:
Sat Jun 30, 2018 12:47 am
willthrill81 wrote:
Fri Jun 29, 2018 9:37 am


I just don't see how still having a large range of expected returns, assuming that CAPE will continue to be as predictive as it appears to have been in the past, is helpful for investors. If I thought I would only get 2% returns, I would probably make different decisions than if I would get 12% returns.
For example, at the extreme of an elevated CAPE, to paraphrase the article and my understanding, the median expected return shifts from, say, +10% with a +/-5% 90% CI (completely made up numbers), to +3% with a strong left skew. Given this, at extremely high valuations in CAPE, I could save more, defer retirement, sell equities, etc. So I find it very helpful.
I suppose that I'm just not convinced that CAPE alone can accurately predict that the median expected return would shift downward from something like 10% to 3%. Kitces found that historically low and high starting CAPE values impacted the subsequent 30 year returns by only +/- 1%, and I'm more concerned about what happens over that time frame than the next decade.
Can you link the Kitces study?
Here it is.

Image
https://www.kitces.com/blog/should-equi ... valuation/
Thanks

Post Reply