Valuation-based market timing with PE10 can improve returns?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
fredflinstone
Posts: 2066
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Postby fredflinstone » Mon Jun 06, 2011 1:52 pm

Mel Lindauer wrote:
fredflinstone wrote:I'm starting to draft a revised IPS. the formula I am tinkering with determines the percent in equities based on three factors: age, PE10, and the 5-year TIPS real yield:

min(100,(max(10,(144-(1.25*AGE)-(2*PE10)-(15*TIPS)))))

in the equation above, % equities can never be higher than 100 and never can be lower than 10. Keep in mind that I am somewhat more risk-averse than most investors and my need for growth is fairly low. Here are some sample values:

Code: Select all

PE10   age   TIPS     % equity
10     20       0           99
15     20       0           89
20     20       0           79
25     20       0           69
30     20       0           59
35     20       0           49
40     20       0           39   
      
10     40       0           74
15     40       0           64
20     40       0           54
25     40       0           44
30     40       0           34
35     40       0           24
40     40       0           14
      
10     60       0           49
15     60       0           39
20     60       0           29
25     60       0           19
30     60       0           10
35     60       0           10
40     60       0           10


any constructive criticism is welcome.


I notice that your formula would never have an investor in TIPS, even at 60, right before retirement. I don't think that makes sense. Look at all the great TIPS returns you would have missed.


Sorry, I wasn't clear. This formula is used to determine the percent of liquid assets in EQUITIES. So you look at the 5-year TIPS yield and if it is zero then you determine the % equities accordingly. If the 5-year TIPS yield were 1 percent then all of the above numbers would be reduced by 15. In others words TIPS is one of the independent variables, not the dependent variable.

The rest of my money (anything not invested in equities) could be in TIPS, cash, munis, or whatever the IPS says. In general, I always have 100 percent of my retirement funds in long-duration TIPS and that will not change.

User avatar
floydtime
Posts: 410
Joined: Sat Feb 26, 2011 10:09 pm
Location: A book

Postby floydtime » Mon Jun 06, 2011 2:41 pm

min(100,(max(10,(144-(1.25*AGE)-(2*PE10)-(15*TIPS)))))


Personally, that "feels" overly complex to me (but then I am used to AGE+5 in Bonds :) ).

*If* I were to pay attention to PE (and I must admit I am debating it), then it would probably be something as simple as...

Bond allocation = AGE + (PE10 - 20)

Where my "20" can be adjusted based on one's risk tolerance. Higher of course is more risky.
"Do not value money for any more nor any less than its worth; it is a good servant but a bad master" - Alexandre Dumas

User avatar
fredflinstone
Posts: 2066
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Postby fredflinstone » Mon Jun 06, 2011 2:49 pm

floydtime wrote:
min(100,(max(10,(144-(1.25*AGE)-(2*PE10)-(15*TIPS)))))


Personally, that "feels" overly complex to me.

*If* I were to pay attention to PE (and I must admit I am debating it), then it would probably be something as simple as...

Bond allocation = AGE + (PE10 - 20)

Where my "20" can be adjusted based on one's risk tolerance. Higher of course is more risky.


Simpler is good but I would prefer to ratchet down my equity exposure by more than 1 percentage point for every 1 percentage point increase in PE. I also think that the TIPS yield should somehow be taken into account. Aren't equities relatively more attractive if the TIPS yield is zero (or negative!) than if the TIPS yield is high?

User avatar
floydtime
Posts: 410
Joined: Sat Feb 26, 2011 10:09 pm
Location: A book

Postby floydtime » Mon Jun 06, 2011 3:12 pm

fredflinstone wrote:
floydtime wrote:
min(100,(max(10,(144-(1.25*AGE)-(2*PE10)-(15*TIPS)))))


Personally, that "feels" overly complex to me.

*If* I were to pay attention to PE (and I must admit I am debating it), then it would probably be something as simple as...

Bond allocation = AGE + (PE10 - 20)

Where my "20" can be adjusted based on one's risk tolerance. Higher of course is more risky.


Simpler is good but I would prefer to ratchet down my equity exposure by more than 1 percentage point for every 1 percentage point increase in PE. I also think that the TIPS yield should somehow be taken into account. Aren't equities relatively more attractive if the TIPS yield is zero (or negative!) than if the TIPS yield is high?


Makes sense. I am still thinking about it too. Thanks for starting such an interesting thread.

Minor quibble - unless I misunderstand Shiller's number, it's not a percentage point increase in PE, but a nominal increase (which, unless PE is 100, is way more than 1 percantage point, per 1 PE). This makes your point here stronger though. :wink:
"Do not value money for any more nor any less than its worth; it is a good servant but a bad master" - Alexandre Dumas

User avatar
Scott S
Posts: 1388
Joined: Mon Nov 24, 2008 3:28 am
Location: SW CR IA US NA PE

Postby Scott S » Mon Jun 06, 2011 3:31 pm

For his part, doesn't Schiller just recommend making small changes in AA due to valuation? There was a TV interview with him and Siegel, and it seemed like Schiller recommended making no more than a 5% shift if valuations looked higher or lower than average.

- Scott
My Plan: * Age-10 in bonds until I reach age 60, 50/50 thereafter. * Equity split: 50/50 US/Int'l, Bond split: 50/50 TBM/TIPS. * Everything over 2 months' expenses gets invested.

yobria
Posts: 5978
Joined: Mon Feb 19, 2007 11:58 pm
Location: SF CA USA

Postby yobria » Mon Jun 06, 2011 4:01 pm

letsgobobby wrote:then, you had exactly the same percentage stocks in 1999 (PE10 40) as 2009 (PE10 10)? I just have a hard time seeing how that makes any sense.


Can you site the source of these numbers?

Nick

User avatar
wade
Posts: 539
Joined: Fri Sep 17, 2010 12:38 am
Location: Main Line
Contact:

Postby wade » Mon Jun 06, 2011 4:07 pm

yobria wrote:Right, problem is you don't have that time machine. What you have is a vague pattern, that may take decades to emerge again, that doesn't work in other countries, and that may or may not work in the US going forward. So you really don't have much.


I'm starting to do some catching up on this thread, and this caught my eye.

When you write that this hasn't worked in other countries, do you have some evidence of that? I would really like to see it.

I have looked at the strategy in Japan, which I think represents the most extreme case for testing PE10. By the start of 1990, PE10 was up to the mid-90s according to my calculations. Though, because earnings may be calculated differently in the US and Japan, I don't exactly know if a PE10 of 95 in Japan means the same thing as it would in the US.

Nonetheless, while people finishing their investment periods in the time around this boom would trail the fixed allocation strategy as can only be expected, it is hard to judge the strategy as a failure in Japan. Here is some evidence:

Image

User avatar
fredflinstone
Posts: 2066
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Postby fredflinstone » Mon Jun 06, 2011 5:03 pm

floydtime wrote:
fredflinstone wrote:
floydtime wrote:
min(100,(max(10,(144-(1.25*AGE)-(2*PE10)-(15*TIPS)))))


Personally, that "feels" overly complex to me.

*If* I were to pay attention to PE (and I must admit I am debating it), then it would probably be something as simple as...

Bond allocation = AGE + (PE10 - 20)

Where my "20" can be adjusted based on one's risk tolerance. Higher of course is more risky.


Simpler is good but I would prefer to ratchet down my equity exposure by more than 1 percentage point for every 1 percentage point increase in PE. I also think that the TIPS yield should somehow be taken into account. Aren't equities relatively more attractive if the TIPS yield is zero (or negative!) than if the TIPS yield is high?


Makes sense. I am still thinking about it too. Thanks for starting such an interesting thread.

Minor quibble - unless I misunderstand Shiller's number, it's not a percentage point increase in PE, but a nominal increase (which, unless PE is 100, is way more than 1 percantage point, per 1 PE). This makes your point here stronger though. :wink:


You are correct.

User avatar
jamacq
Posts: 162
Joined: Sun Mar 04, 2007 9:28 am
Location: Virginia

Postby jamacq » Mon Jun 06, 2011 5:25 pm

I've been watching this thread and thinking about the fact that Jack Bogle himself indicated that valuation can be used when considering allocation. He cautioned however that it should be very rare.

I attended Bogleheads VI in DC in 2007 and took some notes and posted them on this forum back in June 2007.

viewtopic.php?p=39309&highlight=#39309

In my post, I reported that "-In response to the subject of changing one's allocation due to current valuations, he noted that maybe six times in a lifetime the market is way over or under priced. 'Markets make us want to do the wrong thing.' He stated that he believed it was ok to change one's allocation when one of these conditions happens (again very rarely) and suggested perhaps when the S&P PE is >=35. Today's PE is around 18 and doesn't warrant change. "

He wasn't referring to the practicality of using PE10 but did certainly note that valuation levels can be used.

I remembered his words during the 08/09 crash and decided to increase my equity allocation.

Jeff

letsgobobby
Posts: 10213
Joined: Fri Sep 18, 2009 1:10 am

Postby letsgobobby » Mon Jun 06, 2011 7:55 pm

yobria wrote:
letsgobobby wrote:then, you had exactly the same percentage stocks in 1999 (PE10 40) as 2009 (PE10 10)? I just have a hard time seeing how that makes any sense.


Can you site the source of these numbers?

Nick


Shiller posts them monthly at his website:

http://www.econ.yale.edu/~shiller/data.htm

PE10 went over 20 in 1995
it went over 30 in 1997
it went over 40 in 1999
it went under 30 in 2001
it went under 20 in Oct 2008
it bottomed end Feb 2009 at 14 BUT it fell 7% from there til Mar 9 so call it 13.

In this article that I read at the time:

http://economix.blogs.nytimes.com/2009/ ... p-e-ratio/

the PE10 was calculated at 13.5 on Feb 20, and I recalled stocks fell about 15% from there, so I took it down mentally to about 11 or 11.5 at the lows.

hope that helps.

User avatar
Pres
Posts: 230
Joined: Thu Aug 07, 2008 4:25 pm
Location: Eurozone

Postby Pres » Mon Jun 06, 2011 8:55 pm

yobria wrote:The problem is, stocks have a PE of 5 or 10 or 30 for a reason. For example, last time stocks had really low P/Es (30 years ago), bonds were yielding something like 15%. Not the best time to dump bonds.


Imagine indeed, that one day stocks are again at a really low PE10 and bonds yield 15%.

I know, this isn't a Boglehead question, but what would be the most interesting allocation to own?
Mostly stocks? Mostly bonds? 50/50 because both might end badly?

letsgobobby
Posts: 10213
Joined: Fri Sep 18, 2009 1:10 am

Postby letsgobobby » Mon Jun 06, 2011 9:27 pm

Pres wrote:
yobria wrote:The problem is, stocks have a PE of 5 or 10 or 30 for a reason. For example, last time stocks had really low P/Es (30 years ago), bonds were yielding something like 15%. Not the best time to dump bonds.


Imagine indeed, that one day stocks are again at a really low PE10 and bonds yield 15%.

I know, this isn't a Boglehead question, but what would be the most interesting allocation to own?
Mostly stocks? Mostly bonds? 50/50 because both might end badly?


at that point you might say 50/50 because both will end well.

User avatar
fredflinstone
Posts: 2066
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Postby fredflinstone » Mon Jun 06, 2011 10:23 pm

jamacq wrote:In my post, I reported that "-In response to the subject of changing one's allocation due to current valuations, he noted that maybe six times in a lifetime the market is way over or under priced. 'Markets make us want to do the wrong thing.' He stated that he believed it was ok to change one's allocation when one of these conditions happens (again very rarely) and suggested perhaps when the S&P PE is >=35. Today's PE is around 18 and doesn't warrant change."


I don't understand why an investor should "stay the course" at a PE10 of 34.9 but reduce equity exposure at a PE10 of 35.0. I much prefer a gentle, gradual reduction in equity exposure as PE10 climbs higher.

tioga
Posts: 21
Joined: Thu May 12, 2011 9:10 pm

Postby tioga » Tue Jun 07, 2011 10:40 am

louis c wrote:
Finally, I did not find in the paper where it details the 50/50 approach. Since it is the basis of comparison for the thesis, is it not meaningful to know whether the 50/50 allocation is rebalanced daily, weekly, annually, or bi-annually?



(newbie, so system does not let me paste link to the paper) mpra.ub.uni-muenchen.de/29448/1/MPRA_paper_29448.pdf
Bottom of page 4, top of page 5:

Methodology and Data
As with Fisher and Statman, I chart the nominal wealth accumulation of $1 invested at the start of 1871. The buy-and-hold strategy is represented by 100 percent large-capitalization stocks (S&P 500). I consider a fixed allocation strategy of 50 percent stocks and 50 percent Treasury bills (one-year yields) as well when discussing risk and the appropriate benchmark for market timing. The 50/50 strategy is not strictly “buy-and-hold,” as I assume the investor rebalances to meet this target asset allocation at the start of each year.

tioga
Posts: 21
Joined: Thu May 12, 2011 9:10 pm

Postby tioga » Tue Jun 07, 2011 1:32 pm

So here is something that has been bothering me while reading this thread. As many here have indicated, a true Boglehead should (1) identify their risk tolerance, (2) pick an asset allocation, and (3) rebalance periodically so as to meet that asset allocation. But isn't rebalancing to the chosen asset allocation assuming that at the time of rebalancing, the chosen asset allocation truly reflects the investor's risk tolerance?

This troubles me. Because when valuations (PE, PE10) are low, I would expect the risk of investing in equities to be qualitatively lower than when valuations are high. Doesn't it therefore follow that asset allocations should be changed not just with investor-centric attributes (such as life events or age) but also with the then-current risk of investing in equities vs. bonds?

Furthermore, extending Wade Pfau's logic, wouldn't it make sense to use valuations not just to determine one's equity-bond split, but also as what the total asset allocation (e.g. 10% in class X, 15% in class Y, etc.) should be? Why wouldn't we extend Wade's line of thinking to bias towards whatever is cheapest at that point in time? I'd very much like to see how well the PE10 of the S&P500 compares to the PE10 of international small value equities.

Finally, I have just "woken up" in the investing sense and have a nagging feeling that in a year, I will cringe when re-reading this posting, much as fredfintstone said two months or so ago. Please be gentle when you flame this post :-)

User avatar
bob90245
Posts: 6511
Joined: Mon Feb 19, 2007 8:51 pm

Postby bob90245 » Tue Jun 07, 2011 2:03 pm

tioga wrote:Because when valuations (PE, PE10) are low, I would expect the risk of investing in equities to be qualitatively lower than when valuations are high.

Don't remember if I answered this very question here or on another thread. But if you were to buy when PE10 is low, that would put you in or around March 2009. Was that a "low risk" time to buy? In case I have to refresh your memory, many experts were saying that the light at the end of the tunnel was a freight train racing toward you and heading everyone on a collision course to the Second Great Depression.
Ignore the market noise. Keep to your rebalancing schedule whether that is semi-annual, annual or trigger bands.

User avatar
floydtime
Posts: 410
Joined: Sat Feb 26, 2011 10:09 pm
Location: A book

Postby floydtime » Tue Jun 07, 2011 10:02 pm

tioga, that was a great post imho.

The problem I have is that...okay, we can tell how expensive or inexpensive equities are relative to their historical norm.

But as far as I know, there is no definitive way to tell if they are more or less expensive than bonds. Let's say PE10 hits 35. Yikes, time to sell some stocks and buy more...what?...bonds?

How do we know bonds aren't even more expensive? By comparing today's yields against the "historical norm"? What is their "historical norm"? Yields have been drifting downward now for what, 30 years?

To summarize - does it do us any good to "know" that stocks are expensive or cheap if we can't really tell if bonds are expensive or cheap?

I am still learning too, so forgive me if this is daft. :?
"Do not value money for any more nor any less than its worth; it is a good servant but a bad master" - Alexandre Dumas

User avatar
tadamsmar
Posts: 7219
Joined: Mon May 07, 2007 12:33 pm

Re: Valuation-based market timing with PE10 can improve retu

Postby tadamsmar » Tue Jun 14, 2011 12:27 pm

fredflinstone wrote:That's what Professor Wade Pfau says:

[M]arket timing [using Schiller's PE10 measure of valuation] provides comparable risks and the same average asset allocation as a 50/50 fixed allocation strategy, but with much higher returns.


http://mpra.ub.uni-muenchen.de/29448/1/ ... _29448.pdf

I have not yet read the paper closely. I am going to do so now. This seems to me to be a fundamental challenge to some of the most basic tenets of the Boglehead paradigm.

Note that Professor Pfau acknowledges this Forum on the first page of his paper.


I didn't read it closely either. I just searched "taxes" and read that part closely. He ignored taxes and fees.

As the author says "More research is needed about this." I agree.

Maybe it's not obvious to all that when you are doing long-term historical backtesting you can't assume they were using tax-deferred accounts.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Postby Rodc » Fri Jun 17, 2011 6:30 pm

dmcmahon wrote:
Rodc wrote:At P/E (or P/E10) of 5 the earnings yield is 20%. Unless bond yields are very high that is going to look mighty attractive. May take folks a little time to overcome the fear that got yield so high, but eventually that yield will be too high to resist and dollars will flow back into stocks, raising P/E (P/E10, etc.)

Likewise, at a P/E (P/E10) level of 25 the earnings yield on stocks is only 4%, and unless bond yields are extremely low that will not be attractive for long.


And what about when stocks are at P/E10 of 25+, while at the same time bond yields are 3% (10-year) ranging to 4.2% (30-year)? In other words, when both stocks and bonds are equally unattractive?


The answer is contained in the quote you provided (bold to help).
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

SmithIndex
Posts: 25
Joined: Wed Jun 01, 2011 9:50 am

Postby SmithIndex » Sat Jun 18, 2011 2:46 pm

What I find interesting is the concept of using PE10 as an indicator of the next year's market return. This paper rolls this prediction up into a portfolio, and concludes on average over the past 100 years PE10 had predictive powers.

I like the use of rolling median PE10. This allows the analysis to be extended to other data sets. There remain issues of startup and lookback period. (Using median PE10 is better than mean PE10. Earnings yield (1/PE10) might be a better metric.)

The fair PE10 value will change over time as the market changes. The value will depend on investment alternatives, amount of capital available to invest, company prospects, tax and regulatory environment. So, some sort of lookback period is needed to allow for changes in the market environment. A lookback period of 30 years could be used, with 1901 as the first year invested.

SmithIndex
Posts: 25
Joined: Wed Jun 01, 2011 9:50 am

Postby SmithIndex » Sat Jun 18, 2011 4:18 pm

Looks to me like any predictive value of PE10 has decreased over time. Maybe this worked from 1900 to 1950. Probably didn't work 1960 to 2010.

PE10 timing gave lower returns recently. Say you use a limit of 25. Then out of market: 1996 to 2002, 2004 to 2007.
Sharpe Ratio, CAGR (for past 15 years 1996 to 2010)
total stock market: 0.29, 7%
PE10 timing: 0.21, 5%

Just looking at one market, there is not enough data to make any meaningful conclusions. There are too few timing signals to make a valid statistical analysis.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Postby Rodc » Sat Jun 18, 2011 6:38 pm

median PE10


Median price and median of average earnings?

What kind of statistic is median of average and what sort of properties does it have?
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Postby Rodc » Sat Jun 18, 2011 6:41 pm

Looks to me like any predictive value of PE10 has decreased over time. Maybe this worked from 1900 to 1950. Probably didn't work 1960 to 2010.


FWIW this is basically true for Tobin's q which is not all that much different from P/E10.

http://home.comcast.net/~rodec/finance/ ... mingV2.pdf

I missed the point at the time that q worked to some degree as a timing signal on the data used to develop q, then stopped working thereafter.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Postby Rodc » Sat Jun 18, 2011 8:01 pm

To summarize - does it do us any good to "know" that stocks are expensive or cheap if we can't really tell if bonds are expensive or cheap?


That is exactly correct.

You can look at earnings yield, in this case E10/P. So P/E10 of 20 is a earnings yield of 5%. If say TMB is yielding 3% you get one story. If yielding 6% you get an entirely different story.

Stock risk is up now as P/E10 is on the high side.

Bond risk is up as well as more room for rising yields that dropping yields.

But in both cases we don't know what we will get.

So what to do?

It makes absolutely no sense to look at stock valuations in isolation, but that is just what most people discussing P/E10 are doing.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

User avatar
floydtime
Posts: 410
Joined: Sat Feb 26, 2011 10:09 pm
Location: A book

Postby floydtime » Sat Jun 18, 2011 10:40 pm

Rodc wrote:
To summarize - does it do us any good to "know" that stocks are expensive or cheap if we can't really tell if bonds are expensive or cheap?


That is exactly correct.

You can look at earnings yield, in this case E10/P. So P/E10 of 20 is a earnings yield of 5%. If say TMB is yielding 3% you get one story. If yielding 6% you get an entirely different story.

Stock risk is up now as P/E10 is on the high side.

Bond risk is up as well as more room for rising yields that dropping yields.

But in both cases we don't know what we will get.

So what to do?

It makes absolutely no sense to look at stock valuations in isolation, but that is just what most people discussing P/E10 are doing.


Thank you for helping confirm my suspicion.

Without fully understanding the *relative* value of the various assets I own (nor their future value relationships), I'll just stay the course.
"Do not value money for any more nor any less than its worth; it is a good servant but a bad master" - Alexandre Dumas

cjking
Posts: 1714
Joined: Mon Jun 30, 2008 4:30 am

T

Postby cjking » Sun Jun 19, 2011 5:09 am

Rodc wrote:FWIW this is basically true for Tobin's q which is not all that much different from P/E10.

http://home.comcast.net/~rodec/finance/ ... mingV2.pdf

I missed the point at the time that q worked to some degree as a timing signal on the data used to develop q, then stopped working thereafter.


I do get irritated every time you repeat this claim about q not working. There's a big difference between finding it didn't reliably add value in one particular scenario you defined, and refuting the work of Smithers & Wright.

If I may speak metaphorically, if q is a hammer and your paper is your experience of doing some DIY in which a hammer might possibly have been a useful tool, you have found that it didn't help, and now you go around telling everyone there's no point to having a hammer.

This may not be a fair characterisation - it's just how I feel.

No need to respond to all the above - I'm just letting off steam.

In fairness I must point out that I haven't re-read your paper since (probably) 2008, and even then I lost interest (hence concentration) as I read because the investing paradigm was to distant from anything that I might use. So I may well have missed something.

I think you did usefully prove that for a 50:50 portfolio, rebalancing probably does as good a job of protecting against the consequences of bubbles as timing might do in a more complex strategy. Unfortunately I don't expect ever to have a 50:50 (or similar) portfolio. (I could say why but that would be another discussion.)

cjking
Posts: 1714
Joined: Mon Jun 30, 2008 4:30 am

Postby cjking » Sun Jun 19, 2011 5:15 am

Rodc wrote:It makes absolutely no sense to look at stock valuations in isolation, but that is just what most people discussing P/E10 are doing.


I agree that PE10 should not be used in isolation. I did similar (but much less comprehensive) studies to Wade's, proving that PE10 (and q) did add value when used in isolation on historical data, but I long ago concluded that wasn't the way to use them, going forward. I do use return estimates for all asset classes as input to asset allocation decisons.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Re: T

Postby Rodc » Sun Jun 19, 2011 8:48 am

cjking wrote:
Rodc wrote:FWIW this is basically true for Tobin's q which is not all that much different from P/E10.

http://home.comcast.net/~rodec/finance/ ... mingV2.pdf

I missed the point at the time that q worked to some degree as a timing signal on the data used to develop q, then stopped working thereafter.


I do get irritated every time you repeat this claim about q not working. There's a big difference between finding it didn't reliably add value in one particular scenario you defined, and refuting the work of Smithers & Wright.

If I may speak metaphorically, if q is a hammer and your paper is your experience of doing some DIY in which a hammer might possibly have been a useful tool, you have found that it didn't help, and now you go around telling everyone there's no point to having a hammer.

This may not be a fair characterisation - it's just how I feel.

No need to respond to all the above - I'm just letting off steam.

In fairness I must point out that I haven't re-read your paper since (probably) 2008, and even then I lost interest (hence concentration) as I read because the investing paradigm was to distant from anything that I might use. So I may well have missed something.

I think you did usefully prove that for a 50:50 portfolio, rebalancing probably does as good a job of protecting against the consequences of bubbles as timing might do in a more complex strategy. Unfortunately I don't expect ever to have a 50:50 (or similar) portfolio. (I could say why but that would be another discussion.)


I accept your complaint. I made an overly broad statement.

I very specifically tried not to data mine. My concern was that while I am pretty certain one can find a way to make this sort of timing work on past data if you try hard enough, if one works too hard to do so that increases the chance of simply over-fitting the data and decreases the chance that the scheme will work in the future.

So I looked at what I felt a reasonable person might do it they thought about the problem, looked at some data up front but did not force a methodology tightly on the data.

Someone else under those conditions might have made other choices.

But, and I think this is very important, if one really has to carefully choose how to use valuations to make this work on past data, then the use of valuations is not robust. And thus one should be very skeptical.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

User avatar
fredflinstone
Posts: 2066
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Postby fredflinstone » Sun Jun 19, 2011 11:07 am

Rodc wrote:You can look at earnings yield, in this case E10/P. So P/E10 of 20 is a earnings yield of 5%. If say TMB is yielding 3% you get one story. If yielding 6% you get an entirely different story.


Ok, I get that.

So a few questions:

1) if E10/P is less than, say, the 10-year treasury yield would you agree that it makes sense to reduce one's allocation to stocks and move that money toward bonds?

2) what is the appropriate bond benchmark for such comparisons? TBM? The 10-year treasury? The 30-year treasury? Muni bond yields? Where do TIPS yields fit into all this? I would think that would the most important benchmark of all.

3) assuming that stocks are more tax-efficient than bonds (i.e., because unrealized capital gains do not result in taxes), how does that factor into allocation decisions for those of us who are investing in taxable accounts?

4) what if E10/P and bond yields are BOTH much lower than the historical average? Does it make sense then to increase the allocation to cash?

I'm having trouble wrapping my head around the above. After reading Professor Pfau's paper, I'm sure I agree with Jack Bogle that at least "gentle" market timing at extreme valuations makes sense. But Professor Pfau's paper doesn't really address any of the above. I would like to create an allocation formula that takes both bond yields and stock valuations into account, but I am having trouble doing so due to the gaps in my knowledge.

Bongleur
Posts: 2037
Joined: Fri Dec 03, 2010 10:36 am

Postby Bongleur » Sun Jun 19, 2011 12:17 pm

It may be that there is a statistically valid, forward-predicting methodology BUT that it is only valid for a small set of investments. ie, maybe it only works for a 50:50 portfolio, or only if you use TBS instead of LT Treasuries for the bond portion...

I'm more interested in finding a forward-predicting methodology that works for MY situation than in a general theory of prediction.

A method that reduces the negative tail and still makes "enough" yield is what I'd be happy with...
Seeking Iso-Elasticity. | Tax Loss Harvesting is an Asset Class. | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

User avatar
grayfox
Posts: 4492
Joined: Sat Sep 15, 2007 4:30 am
Location: Earth

Postby grayfox » Sun Jun 19, 2011 12:24 pm

fredflinstone wrote:1) if E10/P is less than, say, the 10-year treasury yield would you agree that it makes sense to reduce one's allocation to stocks and move that money toward bonds?

2) what is the appropriate bond benchmark for such comparisons? TBM? The 10-year treasury? The 30-year treasury? Muni bond yields? Where do TIPS yields fit into all this? I would think that would the most important benchmark of all.



You should compare E10/P to TIPS yield because they are both real yields.
For instance today, 2041 TIPS YTM is 1.815% and E10/P is 4.45%
So there is an equity premium over the 30-year TIPS

http://online.wsj.com/mdc/public/page/2 ... nav_2_3020
http://www.multpl.com/

User avatar
BlueEars
Posts: 3471
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Postby BlueEars » Sun Jun 19, 2011 12:44 pm

Here is what I recall (may be off on this) after reading a copy of Smither's Wall Street Revalued. He concluded that both PE10 and Q had a similar story to tell. He also postulated that the forward use of both to predict valuation would be less robust because of market acceptance of these concepts.

Personally I look at the last 8 months of PE10 data and if it's over 22, adjust my algorithms for AA to reduce risk. This is only one of a few valuation measure I use, FWIW.

SP-diceman
Posts: 3968
Joined: Sun Oct 05, 2008 9:17 am

Postby SP-diceman » Sun Jun 19, 2011 1:18 pm

Rodc wrote:It makes absolutely no sense to look at stock valuations in isolation, but that is just what most people discussing P/E10 are doing.


Isnt that what Bogleheads do?
Bringing bonds into the picture is a whole bunch of noise.


Thanks
SP-diceman

User avatar
fredflinstone
Posts: 2066
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Postby fredflinstone » Sun Jun 19, 2011 3:36 pm

grayfox wrote:You should compare E10/P to TIPS yield because they are both real yields.
For instance today, 2041 TIPS YTM is 1.815% and E10/P is 4.45%
So there is an equity premium over the 30-year TIPS


OK, thanks; that is helpful. Why is 30-year TIPS the relevant benchmark rather than, say, 5 year TIPS?

User avatar
grayfox
Posts: 4492
Joined: Sat Sep 15, 2007 4:30 am
Location: Earth

Postby grayfox » Sun Jun 19, 2011 4:10 pm

fredflinstone wrote:
grayfox wrote:You should compare E10/P to TIPS yield because they are both real yields.
For instance today, 2041 TIPS YTM is 1.815% and E10/P is 4.45%
So there is an equity premium over the 30-year TIPS


OK, thanks; that is helpful. Why is 30-year TIPS the relevant benchmark rather than, say, 5 year TIPS?


Well, there's term premium between a 5-year and 30-year TIPS. (-0.451 to 1.815) We'd like to know just the pure equity premium.

Stocks have infinite maturity and duration P/D. So use the bond with the longest maturity and duration you can find. It's too bad there weren't perpetual TIPS, because there would probably be a some term premium between 30-years and perpetuity. In other words, part of what we're calling equity premium is actually term premium for longer duration of about 50 years for S&P 500.

User avatar
BlueEars
Posts: 3471
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Postby BlueEars » Sun Jun 19, 2011 4:26 pm

When I think of stocks and bonds, I'll generally compare stock PE's to short to intermediate Treasury bonds. The so called Fed model popularized by Yardeni (which the Fed never really used) compared stock PE's to the 10yr Treasury. So I'm not sure I'd agree with using 30yr TIPS to compare to current stock valuations.

Having said this, I'm not sure there is a lot of evidence for a great stock/bond valuation metric. But I'm willing to be convinced.

avalpert
Posts: 5130
Joined: Sat Mar 22, 2008 4:58 pm

Postby avalpert » Sun Jun 19, 2011 4:50 pm

Les wrote:When I think of stocks and bonds, I'll generally compare stock PE's to short to intermediate Treasury bonds. The so called Fed model popularized by Yardeni (which the Fed never really used) compared stock PE's to the 10yr Treasury. So I'm not sure I'd agree with using 30yr TIPS to compare to current stock valuations.

Having said this, I'm not sure there is a lot of evidence for a great stock/bond valuation metric. But I'm willing to be convinced.


Actually, they originally used 30-year notes, they only started using 10-year notes in 2001 when 30-year notes ceased to be issued in preparation for surpluses as far as the eye could see :)

And Siegel argued that inflation-protected bonds made the model more believable (and most agreed, Asness said something to the effect of Siegel changed the model from garbage to legitimate model). So I think using the rate on the longest dated inflation protected bond available is the best choice.

User avatar
BlueEars
Posts: 3471
Joined: Sat Mar 10, 2007 12:15 am
Location: West Coast

Postby BlueEars » Sun Jun 19, 2011 5:59 pm

avalpert wrote:
Les wrote:When I think of stocks and bonds, I'll generally compare stock PE's to short to intermediate Treasury bonds. The so called Fed model popularized by Yardeni (which the Fed never really used) compared stock PE's to the 10yr Treasury. So I'm not sure I'd agree with using 30yr TIPS to compare to current stock valuations.

Having said this, I'm not sure there is a lot of evidence for a great stock/bond valuation metric. But I'm willing to be convinced.


Actually, they originally used 30-year notes, they only started using 10-year notes in 2001 when 30-year notes ceased to be issued in preparation for surpluses as far as the eye could see :)

And Siegel argued that inflation-protected bonds made the model more believable (and most agreed, Asness said something to the effect of Siegel changed the model from garbage to legitimate model). So I think using the rate on the longest dated inflation protected bond available is the best choice.

One source for the Fed Model info is here: http://en.wikipedia.org/wiki/Fed_model And that's what I've seen in discussions, the use of the 10yr Treasury. But I don't read many academic papers so I'm not going to set myself up as an authority on this.

I looked up what Siegel (in 2002) said about the Fed Model in his Stocks For the Long Run, pp 104-104. There was a 1997 Fed paper documenting the correspondence between 30 Treasuries (not TIPS) and SP500 earnings yields. Siegel goes on to show a chart from 1926-2001. The Fed Model didn't seem to work so well from 1926 to 1970. Siegel states, "both history and theory suggest the Fed model breaks down when inflation is low or when consumer prices are stagnant and deflation threatens". And he further says, "Unless inflation heads upward again, it is unlikely that the tight correlation experienced over the past two decades between these two yields will continue". Interesting!

On a "common sense" basis, I'd think that few buy stocks with a 30 year outlook on company dividends, growth and earnings. Many active fund managers will say they buy for the long term and might refer to at least 3-5 years. Whenever I bought stocks in the past (don't do individual stocks now), I'd not have expectations for products out more then maybe 5 years. Certainly never 30 years.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Postby Rodc » Sun Jun 19, 2011 7:01 pm

grayfox wrote:
fredflinstone wrote:1) if E10/P is less than, say, the 10-year treasury yield would you agree that it makes sense to reduce one's allocation to stocks and move that money toward bonds?

2) what is the appropriate bond benchmark for such comparisons? TBM? The 10-year treasury? The 30-year treasury? Muni bond yields? Where do TIPS yields fit into all this? I would think that would the most important benchmark of all.



You should compare E10/P to TIPS yield because they are both real yields.
For instance today, 2041 TIPS YTM is 1.815% and E10/P is 4.45%
So there is an equity premium over the 30-year TIPS

http://online.wsj.com/mdc/public/page/2 ... nav_2_3020
http://www.multpl.com/


Thank you.

Sorry for that slip. I should not have used TBM.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Postby Rodc » Sun Jun 19, 2011 7:01 pm

SP-diceman wrote:
Rodc wrote:It makes absolutely no sense to look at stock valuations in isolation, but that is just what most people discussing P/E10 are doing.


Isnt that what Bogleheads do?
Bringing bonds into the picture is a whole bunch of noise.


Thanks
SP-diceman


If they do they shouldn't.

It is always about the portfolio. Not about individual slices.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Postby Rodc » Sun Jun 19, 2011 7:45 pm

Fred,

1) if E10/P is less than, say, the 10-year treasury yield would you agree that it makes sense to reduce one's allocation to stocks and move that money toward bonds?


As noted above TIPS. As to what to do that is the challenge. When to change and by how much? some say make small gentle changes. That is ok, but unless you make lots of small changes, adding to a large change, it just does not matter very much. (a 50% loss if you have 60% stocks is a 30% loss. If you have 55% stocks it is a 27.5% loss. Is that materially better?)

Given all the market noise, uncertainty, and behavioral issues (if you make a plan will you really follow it?) this is much harder than it at first seems.

Someone has a great quote: In theory, theory and practice are the same. In practice they aren't. :)

It is quite unclear to me that in practice that doing something about valuations is a good idea. If one avoids too much risk in the first place, and just puts away money month by month for 30 or 40 years they may or may not have the very best returns, but it is a very good bet they will have a nice retirement. That is really what matters.

4) what if E10/P and bond yields are BOTH much lower than the historical average? Does it make sense then to increase the allocation to cash?


I was thinking about this some today, and rather than trying to improve returns, attacking risk by simply upping cash may indeed be the most useful thing you can do. You have to settle for less in returns, but that still may be the thing to do to avoid the risk.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

Bongleur
Posts: 2037
Joined: Fri Dec 03, 2010 10:36 am

Postby Bongleur » Sun Jun 19, 2011 8:26 pm

This thread is relevent to this discussion:

Time Diversification actually DOES reduce stock risk

viewtopic.php?t=76452


If you maintain a constant dollar amount in stocks (as opposed to the usual percentage rule) then you are automatically holding a higher percentage when stocks are cheap (ie PE10 is low) and a lower percentage when stocks are expensive (ie PE10 is high).

This might give a better risk-adjusted return than the usual fixed % method.

You could think about an age-adusted fixed dollar strategy, where the starting amount is some percentage of the starting portfolio value at various ages. Maybe change it every 5 years, with adjustments if you are above/below your goal.

Maybe when thinking about the usefulness of PE10, instead of working with the actual PE10 numbers, just maintain various dollar amounts equalling various percentages of the starting portfolio value. Might make first-order calculations easier to perform.
Seeking Iso-Elasticity. | | Tax Loss Harvesting is an Asset Class. | | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

User avatar
kenyan
Posts: 2844
Joined: Thu Jan 13, 2011 12:16 am

Postby kenyan » Sun Jun 19, 2011 9:58 pm

"The greatest enemy of a good plan is the dream of a perfect plan."

Though I agree with the principle that stock valuations matter, I think that taking advantage of them consistently and effectively is probably only possible in retrospect. All of this market timing can lead one seriously off-course.

That said, perhaps I'll be doing a lot of questioning myself - with good reason? - the next time we hit PE10 of 30+.

User avatar
fredflinstone
Posts: 2066
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Postby fredflinstone » Sun Jun 19, 2011 10:59 pm

grayfox wrote:You should compare E10/P to TIPS yield because they are both real yields.
For instance today, 2041 TIPS YTM is 1.815% and E10/P is 4.45%
So there is an equity premium over the 30-year TIPS


does anyone know how the current equity risk premium of 2.6 percentage points compares to the historical average?

just wondering: was the equity risk premium negative in early 2000?

User avatar
fredflinstone
Posts: 2066
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Postby fredflinstone » Sun Jun 19, 2011 11:02 pm

grayfox wrote:
fredflinstone wrote:
grayfox wrote:You should compare E10/P to TIPS yield because they are both real yields.
For instance today, 2041 TIPS YTM is 1.815% and E10/P is 4.45%
So there is an equity premium over the 30-year TIPS


OK, thanks; that is helpful. Why is 30-year TIPS the relevant benchmark rather than, say, 5 year TIPS?


Well, there's term premium between a 5-year and 30-year TIPS. (-0.451 to 1.815) We'd like to know just the pure equity premium.

Stocks have infinite maturity and duration P/D. So use the bond with the longest maturity and duration you can find. It's too bad there weren't perpetual TIPS, because there would probably be a some term premium between 30-years and perpetuity. In other words, part of what we're calling equity premium is actually term premium for longer duration of about 50 years for S&P 500.


makes perfect sense. Thanks for your helpful response.

Bongleur
Posts: 2037
Joined: Fri Dec 03, 2010 10:36 am

Postby Bongleur » Sun Jun 19, 2011 11:45 pm

Is PE becoming unreliable because management is gaming the financial report system?

VALUE INVESTING
viewtopic.php?t=52397

Walt in AZ said
>
Has the quality of financial reporting deteriorated over time making it more difficult to find truly undervalued companies? In other words, is the balance sheet, income statement, statement of cash flows, GAAP, etc. the same today as it was in Graham and Dodd's time? I think one would have to be especially savvy (and cynical) in today's world of creative accounting and deceit to be able to glean information from public reports. I also wonder if some of the top executives have an incentive to distort the financial picture to take advantage of thier compensation package. Heck, a basic discussion of the SP500 P/E ratio and the numbers behind it gets murky and confusing very quickly.
>

Robert T said
>
On valuation metrics: One of the “lessons learned in the decade” from Grantham’s January 2010 Quarterly letter is:

“The two time-tested investment tools, value (P/E ratios and P/B ratios) and price momentum, are now much more heavily used and not so reliable as they once were, say from 1977 to 1997.”

Glenn Greenberg in his commentary in Security Analysis says:

“We and other investors today tend to focus on cash flow after capital expenditures (free cash flow), instead of earnings, to evaluate investment merits of a business. One advantage of this approach is that it helps shortcut a good many games that management can play in reporting profits.”

From Klarman:

“Most value investors today analyze free cash flow.”

And from Berkowitz:

“To value equities, we at Fairholme begin by calculating free cash flow."
>
Seeking Iso-Elasticity. | | Tax Loss Harvesting is an Asset Class. | | A well-planned presentation creates a sense of urgency. If the prospect fails to act now, he will risk a loss of some sort.

letsgobobby
Posts: 10213
Joined: Fri Sep 18, 2009 1:10 am

Postby letsgobobby » Mon Jun 20, 2011 2:25 am

That could be very important. Bongleur, your other point about "attacking" this from the risk side is a good one, too. I think a PE10 approach does improve returns but more importantly I think it does so at lower risk levels. Just in the last 15 years, it would have largely avoided the tech bubble and bust, and the final blow off top of the real estate-associated credit bubble and subsequent financial crisis of 2007-09.

re: constant dollar allocation to stocks... doesn't seem like that would be helpful in the real world where most investors are either accumulators (constantly adding to their portfolio) or decumulators (constantly withdrawing from).

User avatar
fredflinstone
Posts: 2066
Joined: Mon Mar 29, 2010 7:35 am
Location: Bedrock

Postby fredflinstone » Mon Jun 20, 2011 5:33 am

kenyan wrote:"The greatest enemy of a good plan is the dream of a perfect plan."

Though I agree with the principle that stock valuations matter, I think that taking advantage of them consistently and effectively is probably only possible in retrospect. All of this market timing can lead one seriously off-course.


I am probably in the minority here, but for me personally I think staying the course is easier--not harder--if valuation is taken into account in asset allocation decisions. This obviously depends on the individual.

Roy
Posts: 970
Joined: Wed Sep 10, 2008 9:34 am

Postby Roy » Mon Jun 20, 2011 6:49 am

fredflinstone wrote:I am probably in the minority here, but for me personally I think staying the course is easier--not harder--if valuation is taken into account in asset allocation decisions. This obviously depends on the individual.


Don't know the demographics on this but I can see it both ways, and I think you correct about it being an individual issue. Many supposedly disciplined investors could not even rebalance during the prior lowpoint (some even contemplated Plan B), let alone buy yet more equities as a PE10-inspired method might suggest they do. Investing being simple but not easy is always easily discussed in the calm of ex ante but demonstrated in the cauldron of the moment.

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Postby Rodc » Mon Jun 20, 2011 10:16 am

fredflinstone wrote:
kenyan wrote:"The greatest enemy of a good plan is the dream of a perfect plan."

Though I agree with the principle that stock valuations matter, I think that taking advantage of them consistently and effectively is probably only possible in retrospect. All of this market timing can lead one seriously off-course.


I am probably in the minority here, but for me personally I think staying the course is easier--not harder--if valuation is taken into account in asset allocation decisions. This obviously depends on the individual.


Know thy self.

It is important to understand our own personalities so we can pick from the "good enough" strategies one that fits our needs and abilities.

I have from time to time made choices that are not mathematically optimal for reasons of what I think is most likely to work for our family.
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.


Return to “Investing - Theory, News & General”

Who is online

Users browsing this forum: asset_chaos, Bing [Bot], BlueEars, Dandy, indexonlyplease, Intheheadlights, jaj2276, lawrun21, Naismith, TBillT, Texanbybirth, user76, zwzhang and 99 guests