Wade Pfau on High CAPE ratios

[Lsbcal]

...
I personally check CAPE & inflation rates. Specifically I look for when yield minus inflation drops below zero. Sign to head for the exits usually.

I'm wondering where you got this thought. It doesn't seem to spring from the graph you have shown. I guess that value would be:

1/CAPE - inflation = 1/26.7 - 0.1% = 3.7% - 0.1% = 3.6%

 

[Texas Proud]

Same graph with starting time periods grouped by decade and time period (latest start is around 2002 - so you get returns up to 2012). I realize it doesn't look pretty (hey, this is a hobby!), hope it does make sense.

Red is 20s, 30s and 40s.
Green is 50s, 60s and 70s.
Blue is 80s, 90s and early 00s.

Individual shapes gives distinction between individual decades.

I made a typo in my earlier clarification: y-axis is real return including dividends (as per the title), but without re-investing said dividends.

[Edit] Please let me know if you see something unexpected -- no garantuees that I did the analysis fully correct ..

Nothing unexpected.... it was like I was thinking.... that the decades followed a certain slope... IOW, the slope for any particular decade looks similar even though the various decades do not have the same slope...


But, all going in a downward trend with higher CAPE....

 

[NW-Bound]

Just because P/E is high doesn't mean that stock prices can't go up.

 

[Totoro]

I'm wondering where you got this thought. It doesn't seem to spring from the graph you have shown. I guess that value would be:

1/CAPE - inflation = 1/26.7 - 0.1% = 3.7% - 0.1% = 3.6%

It doesn't show in the graph as the individual CAPE-10s aren't corrected for inflation. Current result in the US indeed would be around 3.6%, so it seems "safe" for this signal.

Note that this is a short term indicator. As in, get out NOW while the indicator is negative, and get back in once it is positive.

1/CAPE - inflation (yoy) has been negative in the following periods (between brackets index evolution):

• Late 1941 to mid 1945 (9.8 -> 8.6)
• Mid 1946 to mid 1948 (18.1 -> 16.4)
• First half 1951 (22.0 -> 21.6)
• Late 1968 to early 1970 (103.8 -> 87.2)
• Mid 1973 to mid 1975 (103.8 -> 92.5)
• Mid 1979 to early 1981 (107.4 -> 128.4)
• Late 1999 to mid 2001 (1318 -> 1239)
• Late 2005 to mid 2006 (1226 -> 1260)
• Late 2007 to late 2008 (1463 -> 1217)

So with the exception of 1979 and 2005 the signal worked pretty well in the past. Even in 1979 it wouldn't have been that bad. Bond yields were pretty high then too (10%) so you wouldn't have lost out much, if anything. And 2005 is basically a flat result.




Some notes:

• It doesn't miss all the dips by a long shot. It just gets you out for some dips, big and small.
• There is a strong correlation with interest inversion signals - so it's not really new.
• It is a short term signal, so if inflation is unstable it can be confusing.
• Most "avoid" periods are pretty short. You'll need some major cojones to act on it.
• It seems to get positive again too early (e.g. 1970, 2002, 2008)

I have an underlying excel -- Should make some pretty graphs with it and update for the last 3 years of data. I took most from Shiller though, and his website update makes pulling down the Excel with raw data harder to do now.



And as with everything, it might be a fluke or pattern which doesn't say anything really. Intuitively though for me it makes sense (negative real yield = sign of overpricing).



A bit short on time in the coming months also, but let me know if you're interested.

 

[Lsbcal]

Totoro, thanks for the thorough explanation. I have a model which uses the yield curve (strong dependency) and PE10 (weak dependancy) plus the SP500 returns versus bonds. Since you mention that 1/CAPE - inflation is strongly correlated with the yield curve, it might be somewhat similar. I'll look at my data and check it out.

 

[photoguy]

OK... not as good as I was thinking....

The relationship is far from exact as there is still a lot of randomness to be found in the numbers, but statistical analysis suggests that PE10 can explain about 31% of the fluctuations in real stock returns over ten-year periods.

​

That leaves a lot of explaining to do....

I think we should consider that maybe PE10 is about as good as can be done by *any method* and there's no more to be explained - the remaining 69% is noise and will never be modeled.

 

[Options]

My head hurts !!!!

Here's a great antidote:

Top 41 Rick Ferri Quotes - Fund Reference

Specifically:

Rick [Ferri]on Predictions

On media predictions:

Don’t believe what you read or hear in the media about where a stock is going or where the market will be by a certain date. The people who make these predictions have no idea what prices will be in the future. They do it just for attention. Following their advice will probably make you poorer, not richer.

On guru predictions:

Guru predictions are not investment advice — they are entertainment. Treat them as such and you will be better off for it.

On TV appearances:

Don’t ask famous market gurus if you want to know where the markets are going, and don’t listen to their advice when you see them on television or hear them on the radio.

On market timing:

If you want market timing advice, try a shiny new penny. It’s more accurate than the gurus.

On networks and predictions:

But it doesn’t matter that forecasters can’t predict. The networks aren’t keeping score. Why would any producer let bad forecasting get in the way of good theater?

On marketing:

The more things change, the more they stay the same. There weren’t any market timing experts in the 1980s and 1990s, and there haven’t been any since. There are no experts at predicting markets — there are only experts at marketing predictions.

On predictions:

People can’t predict markets but markets can predict people.

 

[NW-Bound]

I think we should consider that maybe PE10 is about as good as can be done by *any method* and there's no more to be explained - the remaining 69% is noise and will never be modeled.

Do we call "noise" a myriad of other factors, things that are too complicated to consider in a simple number like PE?

What about lack of other investment opportunities, low interest rates, low bond yields, few foreign markets with geopolitical stability, currency exchange rate, etc...?

 

[haha]

I think we should consider that maybe PE10 is about as good as can be done by *any method* and there's no more to be explained - the remaining 69% is noise and will never be modeled.

I wonder if PE10 modified by some sort of momentum factor to prevent a very early exit from a blow-off might not help? It might be that in our modern era of politically controlled markets you do better by being less than completely rational.

In any case, it is hard to tax effectively use these methods unless most of one's funds are in tax protected accounts.

Ha

 

[Totoro]

I made a note to see if adding any such effect would make a difference.

Probably October before I get to it.

Big downside I can see is that you'll end up force-fitting - torture the data until it confesses ..

 

[clifp]

Intuitively I suspect that R^2 of .31 sounds about right for CAPE. I'm also inclined to think the universe and the market are too random and irrational to get a much higher correlation.

The problem I have is I don't see any alternative investment class that is in less over valued. I bet the "PE10" for bonds (e.g. 100/average yield of the 10 year treasury over the last 10 years) is probably near 40 which I bet puts them in even more nosebleed territory.

 

[HFWR]

It might be that in our modern era of politically controlled markets you do better by being less than completely rational.

I'm good...

 

[haha]

Intuitively I suspect that R^2 of .31 sounds about right for CAPE. I'm also inclined to think the universe and the market are too random and irrational to get a much higher correlation.

The problem I have is I don't see any alternative investment class that is in less over valued. I bet the "PE10" for bonds (e.g. 100/average yield of the 10 year treasury over the last 10 years) is probably near 40 which I bet puts them in even more nosebleed territory.

One difference in favor of bonds is that bonds with 2-4 year durations can be chosen. Yet the implied duration of an equity yielding 2.5% may be close to 50 years. There is also cash.

Ha

 

[samclem]

But for portfolio allocation, should we not also consider the prospect of alternative investments, such as bonds, precious metals, cash, etc...? Right now, the other asset classes do not look appealing either.

That's true. But if Pfau's data and analysis is right, when equities are in the present range, they typically return about 1.5% over the next 10 years. If we believe that says something about the future, maybe present returns in ST bonds, etc are more attractive--relatively.

One difference in favor of bonds is that bonds with 2-4 year durations can be chosen. Yet the implied duration of an equity yielding 2.5% may be close to 50 years. There is also cash.

Other work done by Pfau and Kitces suggested that, in reaction to PE10, investors should not only vary their % in equities, but change the "safehaven" from corporate bonds to ST govt bills when PE10s get high. I think the explanation for the historical outperformance of this strategy is that when the markets take a big drop (from a high PE10) that the subsequent "flight to quality" favors govt bills, but that the higher return from corp bonds during periods of more moderate equity valuations make them the better choice.

What Pfau is writing about in the links at the OP is ground he has covered before recently. We discussed it a bit here.

 

[NW-Bound]

That's true. But if Pfau's data and analysis is right, when equities are in the present range, they typically return about 1.5% over the next 10 years. If we believe that says something about the future, maybe present returns in ST bonds, etc are more attractive--relatively.

But to play devil's advocate and to borrow the argument from the EMH camp, perhaps the market has also discounted the future returns of other assets, hence is willing to pay more for equities now.

 

[Packman]

It makes sense to be cautious when P/E is high, whether the current value or a sliding average over past years like PE10.

But for portfolio allocation, should we not also consider the prospect of alternative investments, such as bonds, precious metals, cash, etc...? Right now, the other asset classes do not look appealing either.

+2. It would be nice to see someone recommend other investment options and how they forecast other options, especially in today's economy.

 

[clifp]

Other work done by Pfau and Kitces suggested that, in reaction to PE10, investors should not only vary their % in equities, but change the "safehaven" from corporate bonds to ST govt bills when PE10s get high. I think the explanation for the historical outperformance of this strategy is that when the markets take a big drop (from a high PE10) that the subsequent "flight to quality" favors govt bills, but that the higher return from corp bonds during periods of more moderate equity valuations make them the better choice.

What Pfau is writing about in the links at the OP is ground he has covered before recently. We discussed it a bit here.

But S/T government bonds offer near 0 nominal returns and negative real returns. Even if the the yields of stocks is only 1.5% over the next 10 years isn't that still better than 90 day T-bill at .03%

 

[samclem]

But S/T government bonds offer near 0 nominal returns and negative real returns. Even if the the yields of stocks is only 1.5% over the next 10 years isn't that still better than 90 day T-bill at .03%

If stocks go down 30% and people rush to buy bills/ST govt bonds, the ones we already own will probably go up in value considerably more than .03%. So, today's yields don't necessarily tell us what we'll earn on those assets.

 

[Fermion]

If stocks go down 30% and people rush to buy bills/ST govt bonds, the ones we already own will probably go up in value considerably more than .03%. So, today's yields don't necessarily tell us what we'll earn on those assets.

Exactly how much would you expect a bond paying 0.03% to go up to be considered considerable? I guess a $100,000 bond might go up to $100,010 and you could use that extra $10 to buy into the lower stock market.

Or just hold some cash.

 

[samclem]

Exactly how much would you expect a bond paying 0.03% to go up to be considered considerable? I guess a $100,000 bond might go up to $100,010 and you could use that extra $10 to buy into the lower stock market.

Or just hold some cash.

The broad US equity indexes decreased substantially in value between Jun of 2007 and Jun of 2009. During this period, Vanguard Short Term Govt (VSGDX) shares appreciated about 9%.

 

[photoguy]

Do we call "noise" a myriad of other factors, things that are too complicated to consider in a simple number like PE?

What about lack of other investment opportunities, low interest rates, low bond yields, few foreign markets with geopolitical stability, currency exchange rate, etc...?

These all seem to be reasonable factors to include in a model but I think due to feedback inherent in the market (especially positive feedback which is unstable) including these would at best result in a minor improvement (e.g. a few extra percent variation explained).

I'm not aware of any research where PE10 has been substantially bettered by a more complex model. This could be because I don't follow the literature closely however I think more likely is that people/researchers have tried but just obtained negative results.

I wonder if PE10 modified by some sort of momentum factor to prevent a very early exit from a blow-off might not help? It might be that in our modern era of politically controlled markets you do better by being less than completely rational.

In my original statement, I meant that the PE based prediction of returns in 10 years won't get much better. However i do believe that trading strategies based on PE10, could potentially yield benefits and momentum might be useful.

But personally I haven't seen a lot of rigorous analysis on PE based trading and I don't have the chutzpah to do it myself. Also to some extent a fixed AA approach will also shuffle more money into slices with low valuations.

That's true. But if Pfau's data and analysis is right, when equities are in the present range, they typically return about 1.5% over the next 10 years.

I think "typically" is not quite the right word here. It suggests that a return of 1.5% will be common in the current circumstance. I think this understates the variance -- 1.5% is the center of the distribution of returns but the distribution is going to be pretty wide.

 

[Fermion]

The broad US equity indexes decreased substantially in value between Jun of 2007 and Jun of 2009. During this period, Vanguard Short Term Govt (VSGDX) shares appreciated about 9%.

This isn't the 2007 bond market.

 

[samclem]

This isn't the 2007 bond market.

But it is close to the 2007 equities market. S&P PE10 is now 27.19, on Jan 1, 2007 it was 27.21. (Link to PE10 chart). So, (IMO) the more important question is whether to decrease equity holdings at times like this and seek a refuge somewhere else for an increased amount of the portfolio (ST corp bonds, ST govt bonds, or even cash, as you suggested). Which "refuge" is chosen may be less important. Given that this PE10 timing thing is far from precise and can be "wrong" for a long time, an investor should probably choose something with yields and interest rate sensitivity he can live with.

 

[DEC-1982]

I don't recall whether profits were terrible in 2007/2008/2009. If they were, will PE10 fall dramatically once those bad profit years stop being counted in a few years?

 

[Totoro]

I don't recall whether profits were terrible in 2007/2008/2009. If they were, will PE10 fall dramatically once those bad profit years stop being counted in a few years?

There's that .. earnings index

Earnings 2007: 954
Earnings 2008: 569
Earnings 2009: 198
Earnings 2010: 805
Earnings 2011: 1008

If you choose to discount 2009 you can roughly subtract 10% from the PE-10. Still ends up at 25x (4%). Lofty but not 1999 crazy.