The Money Mustache Community
Learning, Sharing, and Teaching => Investor Alley => Topic started by: arebelspy on August 20, 2016, 03:25:02 PM
-
Can someone point me to research around shifting AA based on market valuations (e.g. CAPE/PE10)?
Something like "AA of 50-50 when CAPE is above X, 70/30 between X and Y, 90/10 above Z" or similar?
Obvious market timing, would like to read the research/backtest info around this idea.
Cheers!
-
Other posters may have more direct responses to the OP, but I haven't come across much research on market valuation based AA (in terms of stock-bond split).
There is quite a bit of research regarding starting valuations and subsequent stock returns; the two show a strong negative correlation (over longer time periods such as 5-10 years). However, I am not sure if there is a strong correlation between bond returns and market valuations, for a CAPE based AA to be effective. Although, some positive correlation does exist between starting bond yields and future bond returns.
Arebelspy, this isn't exactly what you asked for, but Mebane Faber posts good links on CAPE - market returns relationships. He does have a few ETFs some of which are based on CAPE, so there could be some conflict of interest there, but I thought they were good reads:
http://mebfaber.com/2014/08/22/everything-you-need-to-know-about-the-cape-ratio/ (http://mebfaber.com/2014/08/22/everything-you-need-to-know-about-the-cape-ratio/)
www.businessinsider.com/charts-that-explain-stock-market-2016-2 (http://www.businessinsider.com/charts-that-explain-stock-market-2016-2) - See the chart titled "Valuation does not do a great job of predicting near-term returns."
-
Meb Faber and Alpha Architect are where I would start. Cullen Roche just put out a white paper on counter-cyclical investing that is similar.
-
What are you planning to do this for? Is it for extra returns or to increase the margin of safety in your investments?
This kind of thing usually loses money because even when the valuation ratios look expensive the long run expected returns of the stock part of your portfolio will be higher than those of the bond part. There are exceptions (the dotcom bubble for example where the yield on bonds was significantly higher then the earnings yield on earnings - in such situations you need some pretty strong assumptions to suggest equities will perform better).
Have you considered shifting your stock allocation into relatively 'cheaper' countries? That way you are likely to capture significant value premia in the long run. You can do that without altering your asset allocation but it does subject you to currency risk.
-
I'm a little surprised you'd be interested in such a scheme.
Would it really be that useful to you anyway? I get the impression that the supermajority of your money is in real estate. If you already have a 90% (say) allocation there and live off of cashflows, what you do with your remaining 10% isn't supremely important. I would imagine that you would want some portion of that remaining bit to be in cash/safe assets (and I'm sure you do) in case we have another crisis and all your tenants stop paying. Other than that buffer fund, just follow your AA with the rest.
I don't think PE-based investing is wise. I also don't think that the past returns with past PEs will be at all the same as future returns with those same PEs. PEs have been increasing for 50-60 years for a variety of technical and societal reasons and I don't see them decreasing to match the 150-year median for the long term (maybe briefly during a crash). I think the new baseline average PE going forward is more like 20.
-
What are you planning to do this for? Is it for extra returns or to increase the margin of safety in your investments?
I'm a little surprised you'd be interested in such a scheme.
I'm not. I'm not sure what gave you the impression I was.
I want to read the research around it.
I don't think PE-based investing is wise.
Okay. But why? Gut feel? Or research/data?
If the latter, that's what I'm interested in reading about.
I don't think it's wise either. But I'd like some solid data around it.
If anyone knows of any research of basically this concept:
Something like "AA of 50-50 when CAPE is above X, 70/30 between X and Y, 90/10 above Z" or similar?
Or knows a good way to run it yourself, I'd be interested in hearing! :)
-
I'm also interested and have been reading about this, or trying to, in the last couple of days.
Kitces mentions it briefly, mainly around adjusting the proportion of stocks to bonds. Eg in this article
https://www.kitces.com/wp-content/uploads/2014/11/Kitces-Report-May-2008.pdf
Notice depending on PE10 there is an ideal ratio. Generally its either 80% or 60% (with presumably a trend in between, linear or otherwise). You could "glide" your stocks/bonds ratio based on PE10, adjusting it say once a year (though how often is hard to tell, but obviously taken the normal measures of avoiding capital gains and sticking to addition/withdrawal based AA adjustment, is better).
This Forbes article, by Wade Pfau, actually shows some modelling of doing so based on a simple rule criteria.
http://www.forbes.com/sites/wadepfau/2015/07/09/is-a-high-cape-cause-for-alarm-part-2-valuation-based-asset-allocation/#6bcaeb804079
Hard to tell in the article (he is brief) but I'm guessing in his model he reassesses AA annually and then makes the adjustment to 25/50/75 ratio based on where PE10 sits (either "high", "average", "low"). This is a pretty broad, 3-level approach.
Kitces table suggests that a finer rule criteria has merit than such a broad one.
Stocks/bonds aside, Shiller Cape/PE10 ratios can be useful long term predictors (10-15 years) of various equity markets. Eg this German report here is fascinating (and often over-my-head!) reading:
http://www.starcapital.de/files/publikationen/Research_2016-01_Predicting_Stock_Market_Returns_Shiller_CAPE_Keimling.pdf
They publish numbers online here:
http://www.starcapital.de/research/stockmarketvaluation
This would suggest that one could use also glide the AA amongst your various equity ETFs. How much to do so is anyone's guess. If there were no such thing as tax implications (often with home country bias), then the AA of your equity ETFs could be designed to reflect the rough proportions of the total stock market. You could then adjust each a few percent either way based on the current PE10 ratios available. Like the stock/bond ratio, perhaps you could make that adjustment annually.
This all gets even more complicated when you factor in rebalancing your AA. Your AA changes potentially once per year after checking the PE10 (etc) ratios. You then have to rebalance. Or do you? There's plenty of good analysis out there suggesting that using a tolerance band rebalancing method is ideal. SO, lets say you are using a 20% method like suggested here (go down the page):
https://www.kitces.com/blog/best-opportunistic-rebalancing-frequency-time-horizons-vs-tolerance-band-thresholds/
If your actual AA is within the tolerance bands of your new glided AA, then you wouldn't have to do any capital stock selling to get to your new target AA. Just keep on going with the addition/withdrawal method but now aiming at your new glided AA.
Hope this makes sense.
I'm still not sure how one would precisely implement all this, particularly with gliding different equity ETFs. Its clearer with the stock/bond ratio as there is data on it.
Also be nice, this is just a thought process :-)
-
Cullen Roche article on Countercyclical Indexing
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2740027
-
This is good reading:
https://www.kitces.com/blog/valuation-based-tactical-asset-allocation-in-retirement-and-the-impact-of-market-valuation-on-declining-and-rising-equity-glidepaths/
And to go along with it:
https://retirementresearcher.com/retirement-risk-rising-equity-glidepaths-and-valuation-based-asset-allocation/
Their paper:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2497053
its interesting that the focus generally seems to be to make sure that the portfolio doesn't fail after retirement starts and you are in draw down. However, this kind of thinking is also useful in the accumulation phase.
-
Is the Kelly criterion applicable here?
https://en.wikipedia.org/wiki/Kelly_criterion
EDIT: here's a calculator:
https://dqydj.com/optimal-asset-allocation-with-the-kelly-criterion/
-
Kelly Criterion only works if you know your edge (and best for single, repeated contests).
This book is an awesome read on the Kelly Criterion... quite entertaining:
Fortune's Formula: The Untold Story of the Scientific Betting System That Beat the Casinos and Wall Street (https://smile.amazon.com/Fortunes-Formula-Scientific-Betting-Casinos/dp/0809045990)
-
What are you planning to do this for? Is it for extra returns or to increase the margin of safety in your investments?
I'm not. I'm not sure what gave you the impression I was.
I want to read the research around it.
Ah ok. People are usually thinking about doing strategies they ask about so I just assumed.
If you're looking to do backtests then Robert Shiller's website has great data on the S&P 500 includings CAPEs. Even better is Ken French's data library as he has data on around 15 economies (both of these sources are free), but for French you need to import the data into excel and calculate the PEs yourself. I don't think he does CAPE but he does provide other value measures such as P/FCF, P/B and dividend yield.
As I said before, the big problem with these strategies (which I deduced from my own backtesting), is that the equity premia is just so darn high - being out of the market can suck even when CAPE is high (see the tech bubble, and the recent bull market in the US). Furthermore, the relationship between CAPE and returns is non-stationary - the regression line moves up and down depending on the time period you're looking at. If you were doing this kind of research 50 years go the relationship would look a lot stronger than it does now (you can see this by using a rolling regression window and seeing how the the regression coefficient on CAPE in this equation changes: E(R) = a + b*CAPE + e, I can send you my own backtests if you're interested)
-
Podcast of Wes Gray from Alpha Architect, he mentions in here how they are trying really hard to make a value-based market timing strategy but it's been difficult so far. http://mebfaber.com/2016/06/29/podcast-episode-4-guest-wes-gray-even-god-get-fired-active-investor/
Some more stuff from Wes:
http://blog.alphaarchitect.com/2015/07/21/eureka-a-valuation-based-asset-allocation-strategy-that-might-work/
http://blog.alphaarchitect.com/2015/07/22/market-timing-with-value-and-momentum/
http://blog.alphaarchitect.com/2015/03/04/do-valuation-spreads-matter-for-market-timing/
-
Kelly Criterion only works if you know your edge (and best for single, repeated contests).
This book is an awesome read on the Kelly Criterion... quite entertaining:
Fortune's Formula: The Untold Story of the Scientific Betting System That Beat the Casinos and Wall Street (https://smile.amazon.com/Fortunes-Formula-Scientific-Betting-Casinos/dp/0809045990)
On my shelf...unread. Just moved up to top of the queue after I finish Bernstein.
-
Thread on Bogleheads:
Damodaran: CAPE is no-good horrible & a market timing spreadsheet
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=198262
-
Thread on Bogleheads:
Damodaran: CAPE is no-good horrible & a market timing spreadsheet
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=198262
Great thread, article from the OP of that, and spreadsheet. Thanks!
-
Meb Faber wrote a bit about it in this paper:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2801856 (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2801856)
-
The Bogleheads thread and associated blogpost was a good find, thanks for sharing, I'll need time to digest it.
Rebs, if you get time, please do summarize what you are finding in your research.
Initially, I can see how it might be tempting, but it opens up a whole new space: do you then also adjust your asset allocation based on bond yields, and what if they are pointing in different directions? What other variables might you look at? It starts to become a complex multi variable optimization process.
Naively speaking, I prefer to just slice and dice with a bit of extra small value allocation, and hope that pays off in the wash.
-
Thread on Bogleheads:
Damodaran: CAPE is no-good horrible & a market timing spreadsheet
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=198262
Great thread, article from the OP of that, and spreadsheet. Thanks!
I liked the post by Larry Swedroe (emphasis mine):
"Also there is no evidence one can use CAPE 10 for timing even though IMO it does provide important information about expected returns, just as current earning does, or CAPE 5 or CAPE 8, nothing magically about CAPE 10, And it says nothing about market being under or overvalued."
-
Thread on Bogleheads:
Damodaran: CAPE is no-good horrible & a market timing spreadsheet
https://www.bogleheads.org/forum/viewtopic.php?f=10&t=198262
That original article raises some good points but I find his evidence pretty lacking. As was mentioned in the thread he's only looking at one and five year time horizons - and you have little business being in stocks if that's the maximum you can invest for
If anyone is looking to market time based on valuations there are measures which work much better than CAPE over the short term (although still not great):
http://gestaltu.com/2012/03/estimating-future-returns.html/ (http://gestaltu.com/2012/03/estimating-future-returns.html/)