Market Timing with Shiller CAPE

[melveyr]

In case you have never heard of the Shiller CAPE, first have a look at the description here:
http://www.multpl.com/shiller-pe/

For my model, I converted the PE 10 into an earnings yield by inverting it. Next, I weighted the amount of equities held by looking at the earnings yield of equities versus the interest rate on a 1 year note. My model weights based off of the respective rates, so if the market was yielding 15% and the 1 year was yielding 5%, my respective allocation for that year would be 75% equities and 25% 1 year notes.

Here is how the weightings of equities would have changed over time...




Here is how it would have performed versus a static mix. The static mix was arrived at by looking at the average weighting of the model over the test period, a little over 60% equities and 40% in one year notes...




Finally here is the excess return of the model versus the mix compounded over time...



The outperformance periods are most notable when the strategy is very heavy into equities. The model is great at getting you in at market bottoms, but you don't fully participate during frothy periods like the 90s...

[D1984]

A couple of questions:

1. Are you using regular (nominal) CAPE or real (inflation-adjusted) CAPE?

2. How does this market timing model deal with periods of heavy inflation? I noticed it would have had you only about 45 or 50% in stocks right before the inflationary/oil shock crash of 1973-74 but it would have had you at about 70-75% stocks during the inflationary WWI years and immediate aftermath (1917-1920) when nominal equity returns were zero or a few percent but real returns came in at over negative 12% per year compounded for four years...ouch. The same happened in 1946-47 when equities declined a few points in nominal terms but lost about 30% in real terms; the timing model still had the investor weighted at about 85-90% equities at this point. This is kind of a "damned if you do, damned if you don't" situation as TIPS weren't available back then and short-term yields didn't even begin to compensate for inflation either but some kinf of model (postive vs negative real rates perhaps) to tell you to switch to a commodity index (since gold wasn't free-market priced at the time) might have helped.

3. It seems the model has the investor in nearly 100% equities at the present time but I am worried that this is more of an artifact of ZIRPed and QE'd lowering of yields than any real indication of equities' relative value vs 1 year Treasuries. Perhaps other stock markets/stock indices (international developed, emerging, value, small cap, etc) should also be included and weighted more heavily when they are a better value than the broad US market.

[melveyr]

A couple of questions:

1. Are you using regular (nominal) CAPE or real (inflation-adjusted) CAPE?

2. How does this market timing model deal with periods of heavy inflation? I noticed it would have had you only about 45 or 50% in stocks right before the inflationary/oil shock crash of 1973-74 but it would have had you at about 70-75% stocks during the inflationary WWI years and immediate aftermath (1917-1920) when nominal equity returns were zero or a few percent but real returns came in at over negative 12% per year compounded for four years...ouch. The same happened in 1946-47 when equities declined a few points in nominal terms but lost about 30% in real terms; the timing model still had the investor weighted at about 85-90% equities at this point. This is kind of a "damned if you do, damned if you don't" situation as TIPS weren't available back then and short-term yields didn't even begin to compensate for inflation either but some kinf of model (postive vs negative real rates perhaps) to tell you to switch to a commodity index (since gold wasn't free-market priced at the time) might have helped.

3. It seems the model has the investor in nearly 100% equities at the present time but I am worried that this is more of an artifact of ZIRPed and QE'd lowering of yields than any real indication of equities' relative value vs 1 year Treasuries. Perhaps other stock markets/stock indices (international developed, emerging, value, small cap, etc) should also be included and weighted more heavily when they are a better value than the broad US market.

1. Real CAPE

2. Yes inflation appears to be another weakness of this model. You are so right about TIPS. I wanted to use them but couldn't! I agree about the international stocks, they could be weighted within the equity component. I have to go hunting for some other CAPEs first though 

BTW I am not using this, just looking for suggestions so thanks for your help!

[MachineGhost]

My thoughts are the equity allocation is too Fed model based, so it won't work in those periods D1984 outlined.  There is no correlation between bond yields and reciprocal P/E's.

[Mark Leavy]

My model weights based off of the respective rates, so if the market was yielding 15% and the 1 year was yielding 5%, my respective allocation for that year would be 75% equities and 25% 1 year notes.

I am genuinely curious as to your reasoning in weighting the allocations proportional to their historic real yields.
This seems to be fundamentally in opposition with the risk parity approach that you have spoken about in other threads.

I can see the intuitive logic in the risk parity model, but even as a momentum play, I'm not grasping the underlying reason for weighting an allocation proportional to it's yield.  I assume that I'm missing something very straightforward.

Thanks!
Mark

[Reub]

This is all very complicated for a retired air controller!

Can't someone just tell me when to buy and when to sell?

[melveyr]

My model weights based off of the respective rates, so if the market was yielding 15% and the 1 year was yielding 5%, my respective allocation for that year would be 75% equities and 25% 1 year notes.

I am genuinely curious as to your reasoning in weighting the allocations proportional to their historic real yields.
This seems to be fundamentally in opposition with the risk parity approach that you have spoken about in other threads.

I can see the intuitive logic in the risk parity model, but even as a momentum play, I'm not grasping the underlying reason for weighting an allocation proportional to it's yield.  I assume that I'm missing something very straightforward.

Thanks!
Mark

Hey Mark. This isn't really a risk parity strategy at all. This was just a simple test of how one could use the Shiller P/E to attempt to time the market. It appears that it worked reasonably well, but there are some serious weaknesses in the strategy of course. I am not using this with real money, I just did the test and I found it interesting so I thought I would share the results. 

[melveyr]

Here is another chart showcasing how the PE10 inverted (making an earnings yield) has been a decent indicator of subsequent stock returns.