CAPE: A much stronger predictor of stock returns than many think

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

HomerJ wrote: Wed Dec 16, 2020 4:11 pm This goes back to academic vs real life.
Imagine an investor who must decide at the beginning of each year how much to invest for their retirement in 30 years. Call this the savings rate. For the sake of simplicity, let's assume they are 100% invested in equities and that each year they make the decision mechanically using as inputs three numbers: the current balance in the account, the number of years remaining until retirement, and an estimate of future real returns.
No one does this.

The problem is that, in real life, no one knows when they are going to retire, and no one even knows how much they are going to need in retirement.

No 30 year-old can calculate how large her nest egg should be in 30 years. She is still building her career, she may get married, she may have kids, she may acquire new taste for luxuries, she may decide she likes a simple life...

No one is or really should be changing their saving rate based on CAPE.
I'm really surprised that you would try to argue that the serendipity of life makes long-term retirement planning a rather useless endeavor.
The Sensible Steward
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: CAPE: A much stronger predictor of stock returns than many think

Post by vineviz »

HomerJ wrote: Wed Dec 16, 2020 4:11 pm The problem is that, in real life, no one knows when they are going to retire, and no one even knows how much they are going to need in retirement.
Not knowing the future is no excuse to avoid planning for the future.

I don't KNOW when I'm going to retire but that doesn't stop me from setting a retirement goal and working to achieve it.

And because retirement is not my ONLY goal, I'm regularly put in the position of having to make a tradeoff between saving and spending (e.g. replacing the bathroom sink, traveling with my children while they are still young, fixing the roof this year or trying to wait one more year, etc.). Any information that can help me make better decisions in the face of uncertainty is valuable information, and I suspect most investors feel the same way. That's the very definition of "real life".
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

marcopolo wrote: Wed Dec 16, 2020 4:14 pm So what is the actionable thing?

You say don't use it for AA.
You say no one uses it for deciding how much to invest.
Maybe you are suggesting they should? Do You?

What do you suggest, specifically, not some hand waving theoretical, but what specific action should an investor take based on CAPE? Both, when it is high and low.
As I told HomerJ, I think that a very plausible means of using CAPE is to use the lower of a historically conservative return estimate for stocks (e.g. 5% real) or a CAPE-derived estimate. I see no need for such a plan to ruffle any feathers.
The Sensible Steward
User avatar
HomerJ
Posts: 21281
Joined: Fri Jun 06, 2008 12:50 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by HomerJ »

willthrill81 wrote: Wed Dec 16, 2020 4:08 pm
It sounds like you're moving the goalposts.

vineviz demonstrated one effective means of using CAPE
Well, the retirement example was done starting in 1966 and CAPE hadn't been discovered yet.

CAPE predictions started diverging from reality once it was discovered.

I'd be interested in seeing a retirement example done starting in 1989.

But it has to be done using data known at THAT time, not data from 1926-2020. I'm not sure if vineviz understands that point.

Decisions made in 1989 would have to use 1926-1988 data, and decisions made in 1990 would have to use data from 1926-1989, etc.
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
GAAP
Posts: 2556
Joined: Fri Apr 08, 2016 12:41 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by GAAP »

vineviz wrote: Wed Dec 16, 2020 4:19 pm
HomerJ wrote: Wed Dec 16, 2020 4:11 pm The problem is that, in real life, no one knows when they are going to retire, and no one even knows how much they are going to need in retirement.
Not knowing the future is no excuse to avoid planning for the future.

I don't KNOW when I'm going to retire but that doesn't stop me from setting a retirement goal and working to achieve it.

And because retirement is not my ONLY goal, I'm regularly put in the position of having to make a tradeoff between saving and spending (e.g. replacing the bathroom sink, traveling with my children while they are still young, fixing the roof this year or trying to wait one more year, etc.). Any information that can help me make better decisions in the face of uncertainty is valuable information, and I suspect most investors feel the same way. That's the very definition of "real life".
I'll bet that you are even able to do this without requiring a "rational optimization model"...
“Adapt what is useful, reject what is useless, and add what is specifically your own.” ― Bruce Lee
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

Uncorrelated wrote: Wed Dec 16, 2020 4:15 pm
willthrill81 wrote: Wed Dec 16, 2020 3:53 pm
HomerJ wrote: Wed Dec 16, 2020 3:42 pm The title of this thread is that CAPE is a STRONG predictor of stock returns.

When the opposite has been true in real life.
In the terminology of regression analysis, which was used in the OP, the terms 'independent variable' and 'predictor' are interchangeable.

A variable being a strong predictor of another variable in this type of analysis merely means that there was a strong linear correlation between the variables in the data being analyzed. Nothing more, nothing less.

And from 1989-2010, forward 10 year returns were very strongly correlated with the starting CAPE. That was 'real life' and is not up for debate.
I tried to reproduce your graphs but failed.
I used annual data rather than monthly.

The p-value for CAPE as a linear predictor of forward 10 year returns was <.001.
The Sensible Steward
User avatar
Uncorrelated
Posts: 1237
Joined: Sun Oct 13, 2019 3:16 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Uncorrelated »

GAAP wrote: Wed Dec 16, 2020 4:05 pm
Uncorrelated wrote: Wed Dec 16, 2020 3:46 pm
GAAP wrote: Wed Dec 16, 2020 2:28 pm
HomerJ wrote: Sat Dec 12, 2020 7:00 pm Anyone who changed their AA based on valuations over the past 30 years has LESS money than those of us who ignored valuations.
Seasonal wrote: Wed Dec 16, 2020 10:41 am Please provide evidence that CAPE usually produced better outcomes regarding savings rate and asset allocation than ignoring it.
dont_know_mind wrote: Wed Dec 16, 2020 3:34 am From what I have seen (but I could be wrong), if you are going to tilt or market-time or take a positional view (or why else would CAPE be useful) you get completely caned using CAPE or at least I did when I tried to apply it.
So AA changes, market-timing, and market-tilting are the only reasons to estimate future returns? 465 replies so far, and still ignoring other reasons to estimate future returns.

Your own personal need/desire for future returns may only be for those three reasons -- that doesn't mean that CAPE-based estimates are not useful for other reasons. Quite frankly, unless the naysayers explain why CAPE used for other reasons is not useful, this thread is also not useful anymore.
Provided that the CAPE10 has a "true" R^2 of at least 0.5% on a monthly horizon and there are no other indicators that dominate CAPE10, it would be useful and economically significant for all those things. This is recognized in many papers even those critical of economical indicators, Campbell and Thompson even have a direct formula to calculate the expected sharpe ratio increase from R^2.

However, the fact that CAPE10 has failed to actually obtain superior performance in the past casts doubt that the true R^2 is indeed that high, and suggests that these high R^2's are the result of statistical errors rather than a genuine relationship.

Some suggest that an indicator can be useful for long term planning but not market timing. I simply don't see how this is even possible within any rational optimization model (i.e. merton's portfolio model). But the opposite, where an indicator can be used for market timing but not long-term planning, is possible. For example, if you have insider information that has not yet reached the market.
My question was not if CAPE was useful for those things, but whether it is useful for other things. I have no interest in using CAPE for market timing, AA changes in general, or tilting. Does that mean that I can't possibly use CAPE for something else?

In my limited reading, Merton's portfolio model requires known constant values for the risk-free rate, expected return, and volatility -- is that correct?

If so, it doesn't sound at all rational to me.

In any case, why does a planning methodology -- which if done at all well, will be regularly updated to adjust for reality -- require an optimization model at all?
Merton's portfolio model doesn't require a constant risk free rate, expected return or volatility. You can plug in a time-varying expected return and perform the optimization, but this can be difficult. Unless you're looking at the most trivial problem, you'll probably need machine learning techniques to find an approximate solution.

The error margin between successfully timing the market and shooting yourself in the foot is razor thin. I don't think there is any chance using CAPE is doable for non-experts. If you are a non-expert, buy a target date fund.
GAAP
Posts: 2556
Joined: Fri Apr 08, 2016 12:41 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by GAAP »

HomerJ wrote: Wed Dec 16, 2020 4:22 pm I'd be interested in seeing a retirement example done starting in 1989.

But it has to be done using data known at THAT time, not data from 1926-2020. I'm not sure if vineviz understands that point.

Decisions made in 1989 would have to use 1926-1988 data, and decisions made in 1990 would have to use data from 1926-1989, etc.
Am I missing something? How would this be any different from most backtesting efforts? That's certainly the way I always did things.

He says in the explanation:
There are two options for estimating future real returns. Option 1 is to use a CAPE regression which incorporates only the data available as of 12/31 of the prior year (i.e. no look-ahead bias). Option 2 is to just use the long-run average return as of 12/31 of the prior year. I compared the outcomes for all 81 rolling 30-year periods starting in 1910. Option 1 (using CAPE) produced a higher retirement wealth on average (about 6% higher) with smoother contributions (about 8% less volatility) than not using CAPE.
“Adapt what is useful, reject what is useless, and add what is specifically your own.” ― Bruce Lee
GAAP
Posts: 2556
Joined: Fri Apr 08, 2016 12:41 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by GAAP »

Uncorrelated wrote: Wed Dec 16, 2020 4:26 pm
GAAP wrote: Wed Dec 16, 2020 4:05 pm My question was not if CAPE was useful for those things, but whether it is useful for other things. I have no interest in using CAPE for market timing, AA changes in general, or tilting. Does that mean that I can't possibly use CAPE for something else?

In my limited reading, Merton's portfolio model requires known constant values for the risk-free rate, expected return, and volatility -- is that correct?

If so, it doesn't sound at all rational to me.

In any case, why does a planning methodology -- which if done at all well, will be regularly updated to adjust for reality -- require an optimization model at all?
Merton's portfolio model doesn't require a constant risk free rate, expected return or volatility. You can plug in a time-varying expected return and perform the optimization, but this can be difficult. Unless you're looking at the most trivial problem, you'll probably need machine learning techniques to find an approximate solution.

The error margin between successfully timing the market and shooting yourself in the foot is razor thin. I don't think there is any chance using CAPE is doable for non-experts. If you are a non-expert, buy a target date fund.
Once again: NOT FOR MARKET-TIMING.

You're ignoring the question.
Last edited by GAAP on Wed Dec 16, 2020 4:30 pm, edited 1 time in total.
“Adapt what is useful, reject what is useless, and add what is specifically your own.” ― Bruce Lee
User avatar
Uncorrelated
Posts: 1237
Joined: Sun Oct 13, 2019 3:16 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Uncorrelated »

willthrill81 wrote: Wed Dec 16, 2020 4:24 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:15 pm
willthrill81 wrote: Wed Dec 16, 2020 3:53 pm
HomerJ wrote: Wed Dec 16, 2020 3:42 pm The title of this thread is that CAPE is a STRONG predictor of stock returns.

When the opposite has been true in real life.
In the terminology of regression analysis, which was used in the OP, the terms 'independent variable' and 'predictor' are interchangeable.

A variable being a strong predictor of another variable in this type of analysis merely means that there was a strong linear correlation between the variables in the data being analyzed. Nothing more, nothing less.

And from 1989-2010, forward 10 year returns were very strongly correlated with the starting CAPE. That was 'real life' and is not up for debate.
I tried to reproduce your graphs but failed.
I used annual data rather than monthly.

The p-value for CAPE as a linear predictor of forward 10 year returns was <.001.
I'm unable to take this statement seriously because you have yet to provide a detailed description of your methodology. Goyal and Welch find R^2 values in the negatives and you find p < .001. Obviously one of you must be wrong, and I don't see any errors in Goyal and Welch' methodology. I didn't see spotting any errors in their methodology either.
marcopolo
Posts: 8445
Joined: Sat Dec 03, 2016 9:22 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by marcopolo »

willthrill81 wrote: Wed Dec 16, 2020 4:21 pm
marcopolo wrote: Wed Dec 16, 2020 4:14 pm So what is the actionable thing?

You say don't use it for AA.
You say no one uses it for deciding how much to invest.
Maybe you are suggesting they should? Do You?

What do you suggest, specifically, not some hand waving theoretical, but what specific action should an investor take based on CAPE? Both, when it is high and low.
As I told HomerJ, I think that a very plausible means of using CAPE is to use the lower of a historically conservative return estimate for stocks (e.g. 5% real) or a CAPE-derived estimate. I see no need for such a plan to ruffle any feathers.
Again, you have only addressed what you suggest when CAPE is high.

Many if us, Homer included I believe, already use a fairly low forward return expectation (I use much lower than 5% real), regardless of CAPE. Because I believe we can always have poor returns, even when CAPE is "low".

So, I would not increase my expected returns in my planning in response to CAPE being low. Do you, or is it just a one sided message?
Once in a while you get shown the light, in the strangest of places if you look at it right.
marcopolo
Posts: 8445
Joined: Sat Dec 03, 2016 9:22 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by marcopolo »

vineviz wrote: Wed Dec 16, 2020 4:19 pm
HomerJ wrote: Wed Dec 16, 2020 4:11 pm The problem is that, in real life, no one knows when they are going to retire, and no one even knows how much they are going to need in retirement.
Not knowing the future is no excuse to avoid planning for the future.

I don't KNOW when I'm going to retire but that doesn't stop me from setting a retirement goal and working to achieve it.

And because retirement is not my ONLY goal, I'm regularly put in the position of having to make a tradeoff between saving and spending (e.g. replacing the bathroom sink, traveling with my children while they are still young, fixing the roof this year or trying to wait one more year, etc.). Any information that can help me make better decisions in the face of uncertainty is valuable information, and I suspect most investors feel the same way. That's the very definition of "real life".

You actually use CAPE to decide whether to take vacations with your kids?!?

I am sure glad I did not know about CAPE when my kids were younger.
Once in a while you get shown the light, in the strangest of places if you look at it right.
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

Uncorrelated wrote: Wed Dec 16, 2020 4:30 pm
willthrill81 wrote: Wed Dec 16, 2020 4:24 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:15 pm
willthrill81 wrote: Wed Dec 16, 2020 3:53 pm
HomerJ wrote: Wed Dec 16, 2020 3:42 pm The title of this thread is that CAPE is a STRONG predictor of stock returns.

When the opposite has been true in real life.
In the terminology of regression analysis, which was used in the OP, the terms 'independent variable' and 'predictor' are interchangeable.

A variable being a strong predictor of another variable in this type of analysis merely means that there was a strong linear correlation between the variables in the data being analyzed. Nothing more, nothing less.

And from 1989-2010, forward 10 year returns were very strongly correlated with the starting CAPE. That was 'real life' and is not up for debate.
I tried to reproduce your graphs but failed.
I used annual data rather than monthly.

The p-value for CAPE as a linear predictor of forward 10 year returns was <.001.
I'm unable to take this statement seriously because you have yet to provide a detailed description of your methodology.
I used annual (1/1/XXXX) data for CAPE from Shiller's website. I used forward 10 year returns from Portfolio Visualizer.
The Sensible Steward
Walkure
Posts: 1023
Joined: Tue Apr 11, 2017 9:59 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Walkure »

GAAP wrote: Wed Dec 16, 2020 4:11 pm
marcopolo wrote: Wed Dec 16, 2020 4:02 pm That is a nice academic excersize.

While I have never said CAPE is useless, this is precisely why I keep questioning how actionable it really is.

How many proponents of valuation models were telling people in the depths of the 1999/2000 recession "Hey this is great, with valuations this low, you don't need to bother saving anything for retirement!"

Most people I know just saved what they could based on income, expenses, and life circumstances, not some measure of valuation.

Do you know anyone that actually does that in real life?
Did I say that? No.

Did I save as much as I could as often as I could? Yes. Did I also use CAPE to estimate when I might be able to retire? Yes.

Do I use CAPE to set a withdrawal estimate that drives a budget now that I am retired? Yes.

AFAIK, I'm living a real life.
This is exactly the problem with finding ways to make CAPE actionable. It is not useful for determining how much to save, but having saved whatever you could, and since nobody picks a year 30 years out and then adjusts spending on an annual basis to hit it, you found it useful for estimating how soon you'd be able to retire?
I can see how it might be marginally relevant for budgeting in the withdrawal phase, but if your spending is really that finely tuned to short term fluctuations I'd say your "number" was too small to begin with.
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

marcopolo wrote: Wed Dec 16, 2020 4:31 pm
willthrill81 wrote: Wed Dec 16, 2020 4:21 pm
marcopolo wrote: Wed Dec 16, 2020 4:14 pm So what is the actionable thing?

You say don't use it for AA.
You say no one uses it for deciding how much to invest.
Maybe you are suggesting they should? Do You?

What do you suggest, specifically, not some hand waving theoretical, but what specific action should an investor take based on CAPE? Both, when it is high and low.
As I told HomerJ, I think that a very plausible means of using CAPE is to use the lower of a historically conservative return estimate for stocks (e.g. 5% real) or a CAPE-derived estimate. I see no need for such a plan to ruffle any feathers.
Again, you have only addressed what you suggest when CAPE is high.

Many if us, Homer included I believe, already use a fairly low forward return expectation (I use much lower than 5% real), regardless of CAPE. Because I believe we can always have poor returns, even when CAPE is "low".

So, I would not increase my expected returns in my planning in response to CAPE being low. Do you, or is it just a one sided message?
Apparently you don't understand the method I described.

Even if CAPE is low, implying relatively high forward returns, I would cap the expected return for planning purposes at something like 5% real.
The Sensible Steward
GAAP
Posts: 2556
Joined: Fri Apr 08, 2016 12:41 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by GAAP »

marcopolo wrote: Wed Dec 16, 2020 4:31 pm
Again, you have only addressed what you suggest when CAPE is high.

Many if us, Homer included I believe, already use a fairly low forward return expectation (I use much lower than 5% real), regardless of CAPE. Because I believe we can always have poor returns, even when CAPE is "low".

So, I would not increase my expected returns in my planning in response to CAPE being low. Do you, or is it just a one sided message?
So, how do you decide on expected return for planning purposes? Does it ever change? If so, under what conditions? If not, why not? Would it have ever changed under any set of historical conditions that we have reasonable ways of quantifying? IOW, could I test it?

I'm open to anything that is better than CAPE -- but I have yet to find it.
“Adapt what is useful, reject what is useless, and add what is specifically your own.” ― Bruce Lee
User avatar
HomerJ
Posts: 21281
Joined: Fri Jun 06, 2008 12:50 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by HomerJ »

willthrill81 wrote: Wed Dec 16, 2020 4:18 pm
HomerJ wrote: Wed Dec 16, 2020 4:11 pm This goes back to academic vs real life.
Imagine an investor who must decide at the beginning of each year how much to invest for their retirement in 30 years. Call this the savings rate. For the sake of simplicity, let's assume they are 100% invested in equities and that each year they make the decision mechanically using as inputs three numbers: the current balance in the account, the number of years remaining until retirement, and an estimate of future real returns.
No one does this.

The problem is that, in real life, no one knows when they are going to retire, and no one even knows how much they are going to need in retirement.

No 30 year-old can calculate how large her nest egg should be in 30 years. She is still building her career, she may get married, she may have kids, she may acquire new taste for luxuries, she may decide she likes a simple life...

No one is or really should be changing their saving rate based on CAPE.
I'm really surprised that you would try to argue that the serendipity of life makes long-term retirement planning a rather useless endeavor.
I'm just not sure how someone on the Internet telling me when I was 40 with two kids, and looking for that next promotion, that 10-year returns are 40% likely to be 5% plus/minus 8% instead of 7% plus/minus 8% would have changed my savings rate.
"The best tools available to us are shovels, not scalpels. Don't get carried away." - vanBogle59
User avatar
Uncorrelated
Posts: 1237
Joined: Sun Oct 13, 2019 3:16 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Uncorrelated »

GAAP wrote: Wed Dec 16, 2020 4:30 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:26 pm
GAAP wrote: Wed Dec 16, 2020 4:05 pm My question was not if CAPE was useful for those things, but whether it is useful for other things. I have no interest in using CAPE for market timing, AA changes in general, or tilting. Does that mean that I can't possibly use CAPE for something else?

In my limited reading, Merton's portfolio model requires known constant values for the risk-free rate, expected return, and volatility -- is that correct?

If so, it doesn't sound at all rational to me.

In any case, why does a planning methodology -- which if done at all well, will be regularly updated to adjust for reality -- require an optimization model at all?
Merton's portfolio model doesn't require a constant risk free rate, expected return or volatility. You can plug in a time-varying expected return and perform the optimization, but this can be difficult. Unless you're looking at the most trivial problem, you'll probably need machine learning techniques to find an approximate solution.

The error margin between successfully timing the market and shooting yourself in the foot is razor thin. I don't think there is any chance using CAPE is doable for non-experts. If you are a non-expert, buy a target date fund.
Once again: NOT FOR MARKET-TIMING.

You're ignoring the question.
No, you cannot use CAPE within Merton's portfolio problem if you don't want to perform market timing. The only way to avoid market timing with this optimization problem is to use constant parameters for everything, but if you do that then obviously you can't use CAPE either. If you do use CAPE to estimate a time-varying risk premium or expected return, then that would be market timing.
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

HomerJ wrote: Wed Dec 16, 2020 4:39 pm
willthrill81 wrote: Wed Dec 16, 2020 4:18 pm
HomerJ wrote: Wed Dec 16, 2020 4:11 pm This goes back to academic vs real life.
Imagine an investor who must decide at the beginning of each year how much to invest for their retirement in 30 years. Call this the savings rate. For the sake of simplicity, let's assume they are 100% invested in equities and that each year they make the decision mechanically using as inputs three numbers: the current balance in the account, the number of years remaining until retirement, and an estimate of future real returns.
No one does this.

The problem is that, in real life, no one knows when they are going to retire, and no one even knows how much they are going to need in retirement.

No 30 year-old can calculate how large her nest egg should be in 30 years. She is still building her career, she may get married, she may have kids, she may acquire new taste for luxuries, she may decide she likes a simple life...

No one is or really should be changing their saving rate based on CAPE.
I'm really surprised that you would try to argue that the serendipity of life makes long-term retirement planning a rather useless endeavor.
I'm just not sure how someone on the Internet telling me when I was 40 with two kids, and looking for that next promotion, that 10-year returns are 40% likely to be 5% plus/minus 8% instead of 7% plus/minus 8% would have changed my savings rate.
First, that's not a correct interpretation of the data that anyone has put forward. There is no '40% likelihood'; the confidence interval captures that standard error of the prediction.

Second, I don't see what's difficult to understand about how to use a lower expected return than you had been using for planning purposes. If you believe that forward returns will be 1% instead of 4%, for instance, then it means that you'll have to save more, work longer, or both. If you're in retirement, then it means that you should consider withdrawing less.
The Sensible Steward
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: CAPE: A much stronger predictor of stock returns than many think

Post by vineviz »

Uncorrelated wrote: Wed Dec 16, 2020 4:40 pm
GAAP wrote: Wed Dec 16, 2020 4:30 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:26 pm
GAAP wrote: Wed Dec 16, 2020 4:05 pm My question was not if CAPE was useful for those things, but whether it is useful for other things. I have no interest in using CAPE for market timing, AA changes in general, or tilting. Does that mean that I can't possibly use CAPE for something else?

In my limited reading, Merton's portfolio model requires known constant values for the risk-free rate, expected return, and volatility -- is that correct?

If so, it doesn't sound at all rational to me.

In any case, why does a planning methodology -- which if done at all well, will be regularly updated to adjust for reality -- require an optimization model at all?
Merton's portfolio model doesn't require a constant risk free rate, expected return or volatility. You can plug in a time-varying expected return and perform the optimization, but this can be difficult. Unless you're looking at the most trivial problem, you'll probably need machine learning techniques to find an approximate solution.

The error margin between successfully timing the market and shooting yourself in the foot is razor thin. I don't think there is any chance using CAPE is doable for non-experts. If you are a non-expert, buy a target date fund.
Once again: NOT FOR MARKET-TIMING.

You're ignoring the question.
No, you cannot use CAPE within Merton's portfolio problem if you don't want to perform market timing. The only way to avoid market timing with this optimization problem is to use constant parameters for everything, but if you do that then obviously you can't use CAPE either. If you do use CAPE to estimate a time-varying risk premium or expected return, then that would be market timing.
I don’t think that you understand “market timing” in the same way other people do.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
Uncorrelated
Posts: 1237
Joined: Sun Oct 13, 2019 3:16 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Uncorrelated »

willthrill81 wrote: Wed Dec 16, 2020 4:36 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:30 pm
willthrill81 wrote: Wed Dec 16, 2020 4:24 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:15 pm
willthrill81 wrote: Wed Dec 16, 2020 3:53 pm

In the terminology of regression analysis, which was used in the OP, the terms 'independent variable' and 'predictor' are interchangeable.

A variable being a strong predictor of another variable in this type of analysis merely means that there was a strong linear correlation between the variables in the data being analyzed. Nothing more, nothing less.

And from 1989-2010, forward 10 year returns were very strongly correlated with the starting CAPE. That was 'real life' and is not up for debate.
I tried to reproduce your graphs but failed.
I used annual data rather than monthly.

The p-value for CAPE as a linear predictor of forward 10 year returns was <.001.
I'm unable to take this statement seriously because you have yet to provide a detailed description of your methodology.
I used annual (1/1/XXXX) data for CAPE from Shiller's website. I used forward 10 year returns from Portfolio Visualizer.
There are a billion ways this can go wrong. Please describe precisely which statistical test you have used, how many lags of netwey-west you used, how you corrected for persistence in the right hand side of the equation, how you corrected for persistence in the left hand side of the equation, how you corrected from the fact price appears on both sides of the equation, which robustness tests you have used, etc.

Or you could just refer to a paper which has done all this, such as the one by Goyal and Welch.
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

Uncorrelated wrote: Wed Dec 16, 2020 4:47 pm
willthrill81 wrote: Wed Dec 16, 2020 4:36 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:30 pm
willthrill81 wrote: Wed Dec 16, 2020 4:24 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:15 pm

I tried to reproduce your graphs but failed.
I used annual data rather than monthly.

The p-value for CAPE as a linear predictor of forward 10 year returns was <.001.
I'm unable to take this statement seriously because you have yet to provide a detailed description of your methodology.
I used annual (1/1/XXXX) data for CAPE from Shiller's website. I used forward 10 year returns from Portfolio Visualizer.
There are a billion ways this can go wrong. Please describe precisely which statistical test you have used
Linear regression.
The Sensible Steward
GAAP
Posts: 2556
Joined: Fri Apr 08, 2016 12:41 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by GAAP »

Walkure wrote: Wed Dec 16, 2020 4:38 pm It is not useful for determining how much to save, but having saved whatever you could, and since nobody picks a year 30 years out and then adjusts spending on an annual basis to hit it, you found it useful for estimating how soon you'd be able to retire?
I used it to determine that I could retire early. How early varied over the years until it actually happened.
Walkure wrote: Wed Dec 16, 2020 4:38 pm I can see how it might be marginally relevant for budgeting in the withdrawal phase, but if your spending is really that finely tuned to short term fluctuations I'd say your "number" was too small to begin with.
A withdrawal estimate in my case is just a general target number. Spending is a cash-flow thing that occurs over the year. The budget is really just an estimate of where I think I can/will go for the next year.

I'm managing the portfolio as a perpetuity -- meaning no set endpoint. My withdrawal estimate is designed to allow for annual real income increases of 1% indefinitely (forever). The more I over-spend, the less often I actually get that real income increase. Under-spending works to support future real income increases.

I never set a "number". I retired once the portfolio value I had supported the expenses I desired to support in the conditions at the time.
“Adapt what is useful, reject what is useless, and add what is specifically your own.” ― Bruce Lee
marcopolo
Posts: 8445
Joined: Sat Dec 03, 2016 9:22 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by marcopolo »

GAAP wrote: Wed Dec 16, 2020 4:38 pm
marcopolo wrote: Wed Dec 16, 2020 4:31 pm
Again, you have only addressed what you suggest when CAPE is high.

Many if us, Homer included I believe, already use a fairly low forward return expectation (I use much lower than 5% real), regardless of CAPE. Because I believe we can always have poor returns, even when CAPE is "low".

So, I would not increase my expected returns in my planning in response to CAPE being low. Do you, or is it just a one sided message?
So, how do you decide on expected return for planning purposes? Does it ever change? If so, under what conditions? If not, why not? Would it have ever changed under any set of historical conditions that we have reasonable ways of quantifying? IOW, could I test it?

I'm open to anything that is better than CAPE -- but I have yet to find it.
Never worried about it when i was accumulating. Lived life, and saved what we could.

When deciding if we were comfortable to retire and during withdrawal now, we use a conservative fixed number for expected return.
If we get even lower than that, we can re-evaluate our plans. But, i would rather do that when the low returns actually show up rather than based on a metric that has been of questionable accuracy (even if it better than anything else we have).

Is a CAPE based plan any better, if the best estimate of the spread is something like +/-8%. Ok, you get a planning number, it is almost certain to be wrong. So, you will have to adapt anyway. So, really not that different than just using a low fixed value.
Once in a while you get shown the light, in the strangest of places if you look at it right.
GAAP
Posts: 2556
Joined: Fri Apr 08, 2016 12:41 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by GAAP »

Uncorrelated wrote: Wed Dec 16, 2020 4:40 pm No, you cannot use CAPE within Merton's portfolio problem if you don't want to perform market timing. The only way to avoid market timing with this optimization problem is to use constant parameters for everything, but if you do that then obviously you can't use CAPE either. If you do use CAPE to estimate a time-varying risk premium or expected return, then that would be market timing.
I have no interest in using Merton's portfolio problem for anything -- not optimization, and especially not market timing.

Let me get this straight -- using CAPE to estimate my portfolio balance at the end of next year is market timing? If so, then what method would I use to generate that estimate without market timing?
“Adapt what is useful, reject what is useless, and add what is specifically your own.” ― Bruce Lee
GAAP
Posts: 2556
Joined: Fri Apr 08, 2016 12:41 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by GAAP »

vineviz wrote: Wed Dec 16, 2020 4:45 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:40 pm No, you cannot use CAPE within Merton's portfolio problem if you don't want to perform market timing. The only way to avoid market timing with this optimization problem is to use constant parameters for everything, but if you do that then obviously you can't use CAPE either. If you do use CAPE to estimate a time-varying risk premium or expected return, then that would be market timing.
I don’t think that you understand “market timing” in the same way other people do.
+1
“Adapt what is useful, reject what is useless, and add what is specifically your own.” ― Bruce Lee
User avatar
Uncorrelated
Posts: 1237
Joined: Sun Oct 13, 2019 3:16 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Uncorrelated »

willthrill81 wrote: Wed Dec 16, 2020 4:49 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:47 pm
willthrill81 wrote: Wed Dec 16, 2020 4:36 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:30 pm
willthrill81 wrote: Wed Dec 16, 2020 4:24 pm

I used annual data rather than monthly.

The p-value for CAPE as a linear predictor of forward 10 year returns was <.001.
I'm unable to take this statement seriously because you have yet to provide a detailed description of your methodology.
I used annual (1/1/XXXX) data for CAPE from Shiller's website. I used forward 10 year returns from Portfolio Visualizer.
There are a billion ways this can go wrong. Please describe precisely which statistical test you have used
Linear regression.
Thank you for confirming you made all the statistical errors that the authors of Long-Horizon Predictability: A Cautionary Tale say are common for this type of application. I guess they were right these errors are common.
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

marcopolo wrote: Wed Dec 16, 2020 4:54 pm
GAAP wrote: Wed Dec 16, 2020 4:38 pm
marcopolo wrote: Wed Dec 16, 2020 4:31 pm
Again, you have only addressed what you suggest when CAPE is high.

Many if us, Homer included I believe, already use a fairly low forward return expectation (I use much lower than 5% real), regardless of CAPE. Because I believe we can always have poor returns, even when CAPE is "low".

So, I would not increase my expected returns in my planning in response to CAPE being low. Do you, or is it just a one sided message?
So, how do you decide on expected return for planning purposes? Does it ever change? If so, under what conditions? If not, why not? Would it have ever changed under any set of historical conditions that we have reasonable ways of quantifying? IOW, could I test it?

I'm open to anything that is better than CAPE -- but I have yet to find it.
Never worried about it when i was accumulating. Lived life, and saved what we could.

When deciding if we were comfortable to retire and during withdrawal now, we use a conservative fixed number for expected return.
If we get even lower than that, we can re-evaluate our plans. But, i would rather do that when the low returns actually show up rather than based on a metric that has been of questionable accuracy (even if it better than anything else we have).

Is a CAPE based plan any better, if the best estimate of the spread is something like +/-8%. Ok, you get a planning number, it is almost certain to be wrong. So, you will have to adapt anyway. So, really not that different than just using a low fixed value.
I don't know where the idea that the confidence interval for CAPE estimates is 8% because it's been 4% or lower, depending on which data set it's based upon.

It seems that some of the latest 'anti-CAPE' arguments revolve around CAPE not being necessary for accumulators. I don't think that anyone has claimed that CAPE is necessary for anyone at any time. Rather, some have said that they have found it useful for planning purposes. Those who don't believe that CAPE is useful shouldn't use it. It's as simple as that.
The Sensible Steward
marcopolo
Posts: 8445
Joined: Sat Dec 03, 2016 9:22 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by marcopolo »

GAAP wrote: Wed Dec 16, 2020 4:55 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:40 pm No, you cannot use CAPE within Merton's portfolio problem if you don't want to perform market timing. The only way to avoid market timing with this optimization problem is to use constant parameters for everything, but if you do that then obviously you can't use CAPE either. If you do use CAPE to estimate a time-varying risk premium or expected return, then that would be market timing.
I have no interest in using Merton's portfolio problem for anything -- not optimization, and especially not market timing.

Let me get this straight -- using CAPE to estimate my portfolio balance at the end of next year is market timing? If so, then what method would I use to generate that estimate without market timing?
Even the most adamant CAPE supporters, I don't think, have claimed that CAPE can be used in any meaningful way to predict portfolio performance over the next year.

What do you do with this information anyway? You will have your answer in a year. Whatever decision you make with that info will almost certainly be wrong, and could probably wait until you know the actual answer.
Once in a while you get shown the light, in the strangest of places if you look at it right.
Wrench
Posts: 1055
Joined: Sun Apr 28, 2019 10:21 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Wrench »

HomerJ wrote: Mon Dec 07, 2020 12:54 pm
willthrill81 wrote: Mon Dec 07, 2020 12:12 pm Knowing what a plausible range (i.e. +/- 3%) the next 10 years' of stock returns are very likely to fall within can still be very useful for planning purposes. For instance, a plan that's built on stocks returning 7% real over the next decade is not plausible since the current range is about -1% to +5%.


Just always plan on low and not average historical returns and ignore valuations and all other metrics.

It's really as a simple as that.

If you get more (like we have the past 10 years), great, you adjust and get to retire earlier or spend more in retirement.

If you get less, you adjust the other direction and retire later, save more, or spend less in retirement. Starting from a low point, your adjustment won't be as painful.

Let me ask you this... If CAPE was low, would you advise people to plan around higher than average returns? Would you tell them to save less going forward because it's "likely" returns will be good? Or would you tell them, it's likely returns will be high, but they better keep saving as normal for now until we see what actually happens?

Because that's me... I deal with ACTUAL returns, not expected returns. This is real life, not academia.

Plus or minus 8% isn't that useful to plan around. In my spreadsheets, I've always loosely planned around 4%-5% nominal (low by historical standards), and adjusted. It's not like any of our plans are in stone. I have a plus or minus around my retirement age too... I have a plus or minus around how much I will spend in retirement too...

Just live below your means, save a decent amount, and adjust as you go based on actual returns.

There's no reason to make changes based on "expected" returns.
+1
This is the most actionable post in this entire thread. Plan for the worst, hope for the best, and be flexible enough to modify along the way.
RyeBourbon
Posts: 1219
Joined: Tue Sep 01, 2020 12:20 pm
Location: Delaware/Philly

Re: CAPE: A much stronger predictor of stock returns than many think

Post by RyeBourbon »

GAAP wrote: Wed Dec 16, 2020 3:31 pm
scout1 wrote: Wed Dec 16, 2020 2:37 pm There is no value in estimating future returns incorrectly, and historically the CAPE has been incorrect. Now if you could estimate future returns even a little bit accurately, there would be a LOT of value in that.
No value -- really?

Historically, if you constrained CAPE values by rounding up anything under 10.5to 10.5, you would have overestimated 1-year returns about two-thirds of the time. If your planning is revisited every year, you don't really even care about longer terms.

If a planning effort revolves around estimating a date to hit "a number", underestimation reduces disappointment and provides a much better forward path. If a planning effort revolves around a withdrawal calculation, then underestimation reduces portfolio failure and smooths withdrawals. In either case, a CAPE-based estimate that is generally pessimistic has quite a bit of value.
My advice is always underestimate.
Retired June 2023. AA = 55/35/10
User avatar
Uncorrelated
Posts: 1237
Joined: Sun Oct 13, 2019 3:16 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Uncorrelated »

GAAP wrote: Wed Dec 16, 2020 4:55 pm
Uncorrelated wrote: Wed Dec 16, 2020 4:40 pm No, you cannot use CAPE within Merton's portfolio problem if you don't want to perform market timing. The only way to avoid market timing with this optimization problem is to use constant parameters for everything, but if you do that then obviously you can't use CAPE either. If you do use CAPE to estimate a time-varying risk premium or expected return, then that would be market timing.
I have no interest in using Merton's portfolio problem for anything -- not optimization, and especially not market timing.

Let me get this straight -- using CAPE to estimate my portfolio balance at the end of next year is market timing? If so, then what method would I use to generate that estimate without market timing?
If you don't want to perform market timing, then you would need to use a method that implicitly or explicitly assumes that the underlying market parameters are constant. Using CAPE is obviously not compatible with that goal.

But I hear you saying, I merely use CAPE to estimate my future estate, I don't actually make any decisions based on it. You are now using a suboptimal allocation, which is guaranteed to be worse than market timing if you believe that CAPE has any predictive power.

I'm not saying market timing doesn't work, I'm just explaining this is market timing using the only consistent definition of market timing that I know.
User avatar
LadyGeek
Site Admin
Posts: 95691
Joined: Sat Dec 20, 2008 4:34 pm
Location: Philadelphia
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by LadyGeek »

FYI - The discussion is now focused on CAPE as a withdrawal strategy. Should this part of the discussion be split into a new topic? If it doesn't make sense to do so, we can continue here.
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.
User avatar
LadyGeek
Site Admin
Posts: 95691
Joined: Sat Dec 20, 2008 4:34 pm
Location: Philadelphia
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by LadyGeek »

^^^ A member notified the moderators that the discussion should be split into a new topic. See: CAPE as a withdrawal strategy

Let's use this thread for CAPE predictions.
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.
User avatar
canadianbacon
Posts: 677
Joined: Sun Nov 10, 2019 9:04 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by canadianbacon »

Wrench wrote: Wed Dec 16, 2020 5:01 pm
HomerJ wrote: Mon Dec 07, 2020 12:54 pm Just live below your means, save a decent amount, and adjust as you go based on actual returns.

There's no reason to make changes based on "expected" returns.
+1
This is the most actionable post in this entire thread. Plan for the worst, hope for the best, and be flexible enough to modify along the way.
There are many advanced theoreticians in this topic, but Homer's the one I'd actually let manage my money.
Bulls make money, bears make money, pigs get slaughtered.
Goldwater85
Posts: 241
Joined: Fri Nov 22, 2019 6:00 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Goldwater85 »

HomerJ wrote: Wed Dec 16, 2020 4:22 pm
Well, the retirement example was done starting in 1966 and CAPE hadn't been discovered yet.

CAPE predictions started diverging from reality once it was discovered.

I'd be interested in seeing a retirement example done starting in 1989.

But it has to be done using data known at THAT time, not data from 1926-2020. I'm not sure if vineviz understands that point.

Decisions made in 1989 would have to use 1926-1988 data, and decisions made in 1990 would have to use data from 1926-1989, etc.
CAPE is one of the oldest and most widely recognized valuation adjustments in securities analysis. Graham and Dodd, in Securities Analysis in 1934—the most widely read and influential tome among practitioners for decades—cited the concept as an important back-of-the-envelope valuation tool. They observed that a snapshot of P/E could mislead an analyst because companies at the peak of their business cycles were frequently assigned unusually low P/Es (savvy analysts understood that current earning were likely to decline) and vice versa for companies at the troughs of their cycles (in 1934 American industry was pronouncedly cyclical). To adjust for this, they advocated averaging earnings over the prior 7-10 year period (roughly corresponding to a full business cycle). Doubtless, this idea did not originate with Ben Graham.

Shiller standardized a 10-year average and advocated an inflation adjustment and applying it to market indices. But the idea, even with his tweaks, is highly intuitive. CAPE wasn’t suddenly ‘discovered’ by the market in the late ‘80s.

Better arguments against CAPE allege that the nature of the broader U.S. business cycle has changed since the 1960s, with manufacturing, transportation and heavy industry no long dominant—making this measure less useful on an economy-wide scale. Not sure I find these arguments fully persuasive.
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

Goldwater85 wrote: Fri Dec 25, 2020 7:45 am Better arguments against CAPE allege that the nature of the broader U.S. business cycle has changed since the 1960s, with manufacturing, transportation and heavy industry no long dominant—making this measure less useful on an economy-wide scale. Not sure I find these arguments fully persuasive.
The changing roles of various sectors of the stock market's composition does not seem to me at least to have a theoretical link to CAPE no longer being predictive of forward stock returns. As we've seen in recent decades, the strength of the correlation between CAPE and forward returns has been much stronger, not weaker, than in the past.
The Sensible Steward
User avatar
Forester
Posts: 2400
Joined: Sat Jan 19, 2019 1:50 pm
Location: UK

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Forester »

As risk management, I think diversifying with small cap / SCV, and owning a global portfolio, is better for guarding against rich valuations, than adjusting exposure guided by the CAPE;

https://www.starcapital.de/fileadmin/us ... imling.pdf

Image

CAPE is more useful for enterprising investors putting new money to work.

I believe CAPE-critical articles like this https://thepatientinvestor.com/index.ph ... -the-cape/ miss the point, who would use the measure as a binary in/out strategy?

Link between starting value and medium term returns is clear;

Image
Amateur Self-Taught Senior Macro Strategist
marcopolo
Posts: 8445
Joined: Sat Dec 03, 2016 9:22 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by marcopolo »

willthrill81 wrote: Sat Dec 26, 2020 8:48 pm
Goldwater85 wrote: Fri Dec 25, 2020 7:45 am Better arguments against CAPE allege that the nature of the broader U.S. business cycle has changed since the 1960s, with manufacturing, transportation and heavy industry no long dominant—making this measure less useful on an economy-wide scale. Not sure I find these arguments fully persuasive.
The changing roles of various sectors of the stock market's composition does not seem to me at least to have a theoretical link to CAPE no longer being predictive of forward stock returns. As we've seen in recent decades, the strength of the correlation between CAPE and forward returns has been much stronger, not weaker, than in the past.
But, it may partially explain the shift in CAPE that I demonstrated in a long ago post. CAPE does not take into anticipated growth rate, it only considers PE, instead of PEG.

People are willing to pay higher multiples for a dollar of earnings if the expectation is that dollar will be two dollars in a few years. This is why technology stocks command a higher PE than utility stocks.

So, if an economy shifts from low growth industries to faster growth industries, then one might expect a shift to higher PEs (and CAPE) as a result.

That does not necessarily translate to lower forward returns, because the nature of the underlying investments have changed. That is the problem with a static view of CAPE, it ignores the dynamic nature of the market. All other things being equal, higher CAPE should portend lower forward returns, but all things are rarely equal.
Once in a while you get shown the light, in the strangest of places if you look at it right.
Hydromod
Posts: 1051
Joined: Tue Mar 26, 2019 10:21 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Hydromod »

Just for grins, I updated the related approach from Philosophical Economist with data since I last looked at it in 2019. This theory was mentioned upthread here.

CAGR is calculated from the Shiller monthly database. Equity fraction is calculated with the quarterly data from FRED pointed to by the Philosophical Economist.

This is broken into pairs of estimated equity fraction and subsequent (nominal) CAGR over the following 4, 6, 8, ... years, for all available pairs starting in 1951. The equity fraction is a measure of the fraction of resources devoted to equities. The linked entry points to documentation on this idea.

Image

I placed a minimum convex polygon around the cloud of points (the gray polygon). I would argue the gray outline is the bounding range of what has been observed thus far, and would be a reasonable planning value based on macroeconomics. The cloud seems to be tighter than the CAPE estimates, at least for the 10-year comparison.

The yellow square in each figure is the equity fraction based on the most recent available data from FRED (7-1-2020). The yellow bar gives the range of CAGR based on the bounding polygon, listed in the top right corner. The expected CAGR (labeled E[CAGR]) is calculated from a regression line through all points.

I argue the tightest estimates are for 10 to 12 years ahead. Clearly there has been a very wide scatter in CAGR for relatively short periods, but there has not been a time in the past where 10-year nominal CAGR would have been more than 4.1% or less than -3.7% when the equity fraction was like July, 2020.

The d[CAGR] value listed in each figure is the change in E[CAGR] from the previous time. So in the middle right figure, a d[CAGR] of 4.4 represents E[CAGR] during years 11 though 12.

The d[CAGR] values suggest that the moving expected two-year CAGR will be declining for the next half decade or so, then start recovering after a decade or so. There's an indication that the actual path may be quite volatile, however.

When I calculated ending with the 4-1-2020 data, the 10-year E[CAGR] was 2.2 with bounds of -2.5 and 5.8. Ending with the 4-1-2019 data, 10-year E[CAGR] was 1.1 with bounds of -3.3 to 4.7. I expect that once the more recent data are available, the CAGR numbers will decline some more.
klaus14
Posts: 953
Joined: Sun Nov 25, 2018 6:43 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by klaus14 »

Hydromod wrote: Sun Jan 10, 2021 4:57 pm there has not been a time in the past where 10-year nominal CAGR would have been more than 4.1% or less than -3.7% when the equity fraction was like July, 2020.
This is awesome!
Do you think this is a better predictor than CAPE? How about a two variable model?

Latest data from FRED seems to be Q3 alloc = 45%.
Since then market went up. I'd guess it's like 47% now, which is the highest in history except Q1 2000 when it was 52%
My investment algorithm: https://www.bogleheads.org/forum/viewtopic.php?f=10&t=351899&p=6112869#p6112869
Hydromod
Posts: 1051
Joined: Tue Mar 26, 2019 10:21 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Hydromod »

klaus14 wrote: Sun Jan 10, 2021 6:57 pm
Hydromod wrote: Sun Jan 10, 2021 4:57 pm there has not been a time in the past where 10-year nominal CAGR would have been more than 4.1% or less than -3.7% when the equity fraction was like July, 2020.
This is awesome!
Do you think this is a better predictor than CAPE? How about a two variable model?

Latest data from FRED seems to be Q3 alloc = 45%.
Since then market went up. I'd guess it's like 47% now, which is the highest in history except Q1 2000 when it was 52%
I'm pretty far from claiming I know much about this stuff, but I find the approach and rationale appealing. It appears that it may have a tighter bound than the CAPE approach, which is nice.

I have no idea what a second variable might be though.
klaus14
Posts: 953
Joined: Sun Nov 25, 2018 6:43 pm

Re: CAPE: A much stronger predictor of stock returns than many think

Post by klaus14 »

Hydromod wrote: Sun Jan 10, 2021 8:20 pm
klaus14 wrote: Sun Jan 10, 2021 6:57 pm
Hydromod wrote: Sun Jan 10, 2021 4:57 pm there has not been a time in the past where 10-year nominal CAGR would have been more than 4.1% or less than -3.7% when the equity fraction was like July, 2020.
This is awesome!
Do you think this is a better predictor than CAPE? How about a two variable model?

Latest data from FRED seems to be Q3 alloc = 45%.
Since then market went up. I'd guess it's like 47% now, which is the highest in history except Q1 2000 when it was 52%
I'm pretty far from claiming I know much about this stuff, but I find the approach and rationale appealing. It appears that it may have a tighter bound than the CAPE approach, which is nice.

I have no idea what a second variable might be though.
i meant two variables: CAPE + Equity Alloc. The two are correlated so extra signal is small, but it's there.
My investment algorithm: https://www.bogleheads.org/forum/viewtopic.php?f=10&t=351899&p=6112869#p6112869
ValuationsMatter
Posts: 390
Joined: Mon Mar 09, 2020 2:12 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by ValuationsMatter »

willthrill81 wrote: Mon Nov 30, 2020 4:38 pm
jarjarM wrote: Mon Nov 30, 2020 4:26 pm P.S. As other suggested upthread, there is not out of sample verification for this so the high R square is expected. That's why the correlation for the 10 year individual sample set is 0.99, basically using the same data points to construct the model and its coefficients then testing it against the exact same data.
The analysis I conducted did not do what you claim in the last phrase whatsoever. That's not remotely close to how correlations or regression are calculated. The correlation between two sets of variables can range from -1 to +1 with 0 representing no correlation at all. As a social scientist by trade, I can tell you that correlations close to zero are quite common and that correlations close to -1 or +1 are quite rare outside of the physical sciences.

While you're correct that the specific parameters of the regression model are optimized for the data presented, which I've already said to be so, the point of the analysis in the OP was merely to illustrate that CAPE has been a very strong predictor of future market returns since Shiller proposed it. In other words, it is an out of sample test.

I'm skeptical. Valuations matter, but they don't matter THAT much. I would like to validate your analysis. Would you please provide me the data/code? I write in R and python primarily, and if you didn't use either, I can redo your analysis there, anyways. This is non-attributional to anyone as I will confide my review of your work back to you only. If interested, PM me for more details.

If you're right, I'd love to be on your side here, as this is the very thesis of my investment strategy, but even I am skeptical of any such correlation.
User avatar
Topic Author
willthrill81
Posts: 32250
Joined: Thu Jan 26, 2017 2:17 pm
Location: USA
Contact:

Re: CAPE: A much stronger predictor of stock returns than many think

Post by willthrill81 »

ValuationsMatter wrote: Sun Jan 10, 2021 10:51 pm
willthrill81 wrote: Mon Nov 30, 2020 4:38 pm
jarjarM wrote: Mon Nov 30, 2020 4:26 pm P.S. As other suggested upthread, there is not out of sample verification for this so the high R square is expected. That's why the correlation for the 10 year individual sample set is 0.99, basically using the same data points to construct the model and its coefficients then testing it against the exact same data.
The analysis I conducted did not do what you claim in the last phrase whatsoever. That's not remotely close to how correlations or regression are calculated. The correlation between two sets of variables can range from -1 to +1 with 0 representing no correlation at all. As a social scientist by trade, I can tell you that correlations close to zero are quite common and that correlations close to -1 or +1 are quite rare outside of the physical sciences.

While you're correct that the specific parameters of the regression model are optimized for the data presented, which I've already said to be so, the point of the analysis in the OP was merely to illustrate that CAPE has been a very strong predictor of future market returns since Shiller proposed it. In other words, it is an out of sample test.

I'm skeptical. Valuations matter, but they don't matter THAT much. I would like to validate your analysis. Would you please provide me the data/code? I write in R and python primarily, and if you didn't use either, I can redo your analysis there, anyways. This is non-attributional to anyone as I will confide my review of your work back to you only. If interested, PM me for more details.

If you're right, I'd love to be on your side here, as this is the very thesis of my investment strategy, but even I am skeptical of any such correlation.
I provided the source of the information in the OP. I used annual returns from Portfolio Visualizer and annual starting valuations from Shiller's site.

jarjarM's comment was indicative of not understanding how correlation or regression analysis works (e.g., no 'out of sample verification' does not equate to a high R-square value whatsoever).
The Sensible Steward
User avatar
Forester
Posts: 2400
Joined: Sat Jan 19, 2019 1:50 pm
Location: UK

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Forester »

US CAPE is now over 37, highest since the year 2000 https://www.multpl.com/shiller-pe/table/by-month

The real 10yr yield today is -0.64%, back then it was 1.77%.
Amateur Self-Taught Senior Macro Strategist
Alchemist
Posts: 638
Joined: Sat Aug 30, 2014 6:35 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Alchemist »

Cliff Asness is kind enough to provide us an easily tested real world application. In November of 2012, he used CAPE to predict the return of the S&P 500 over the following decade. The prediction still has a year an half or so to reach a full decade but we can examine how it is going so far.

Asness calculated CAPE prediction: 0.9%

Actual return of VFIAX up to March 2021: 15.36%

The prediction was off by 15.25% :oops:

To be fair to Cliff, his prediction ranged from -4.4% to +8.3% so it is a ‘mere’ 7.06% outside of the already pretty useless 12.7% error range. There also could be some horrific market crash between today and November 2022. It will be interesting to come back and check on how this prediction of his fully fleshes out.

Link for Asness prediction: https://www.aqr.com/Insights/Research/W ... Shiller-PE

Portfolio Visualizer source for VFIAX: https://www.portfoliovisualizer.com/bac ... ion1_1=100
Always passive
Posts: 1267
Joined: Fri Apr 14, 2017 4:25 am
Location: Israel

Re: CAPE: A much stronger predictor of stock returns than many think

Post by Always passive »

Has anyone done the work for foreign stocks?
am
Posts: 4233
Joined: Sun Sep 30, 2007 9:55 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by am »

Forester wrote: Sat Apr 10, 2021 5:36 am US CAPE is now over 37, highest since the year 2000 https://www.multpl.com/shiller-pe/table/by-month

The real 10yr yield today is -0.64%, back then it was 1.77%.
What to do? Probably nothing. How about if cape is in record territory or in the 50s or 60s? Does low bond yields invalidate cape?

As I approach my goal for investments years before I’m ready to retire, I wonder whether I should dial back risk from about 80/20 to 70/30 or less? On the other hand , I’ve hopefully got decades of investing left.
User avatar
CyclingDuo
Posts: 6008
Joined: Fri Jan 06, 2017 8:07 am

Re: CAPE: A much stronger predictor of stock returns than many think

Post by CyclingDuo »

Always passive wrote: Sat Apr 10, 2021 7:28 am Has anyone done the work for foreign stocks?
My previous post was deemed distracted formatting and removed, so I will provide this as a replacement.

https://siblisresearch.com/data/cape-ratios-by-country/

https://www.starcapital.de/en/research/ ... valuation/

The Star Capital data at the link above covers more countries and is as of the end of October, 2020.

The Siblis Research shown below is as of the end of December, 2020...

Image
"Save like a pessimist, invest like an optimist." - Morgan Housel | "Pick a bushel, save a peck!" - Grandpa
Post Reply