The Past Predicts the Future
Price Earnings Ratios as Forecasters of Returns
Robert Shiller and John Campbell initially published on P/E Ratio predicting later real returns in this 1988 paper. In 1996 Shiller followed up the research with this easily digestible paper on the topic. In summary he found that the random walk theory does not look right and that current index valuations have a strong predictive ability for future long term returns. Over the short term noise prevents prediction, whereas over the long term, ten years, current valuation levels are predictive of future returns.
As I have said before the voting machine should be used for the short term, the weighing machine for the long term.
In this 2004 paper, Proxying for Expected Returns with Price Earnings Ratios, Charlotte Strunk Hansen and Bjorn Tuypens expanded on the work of Campbell and Shiller.
Long-run regression models using trailing earnings over price ratio to predict future returns suggested by Campbell and Shiller (1988, 2001) work quite well. However, in this note we show that this variable might result in a downward biased proxy for expected future returns. Instead we suggest using a moving average of the log of 1 plus the earnings price ratio when forecasting long-run returns. The empirical results for the S&P 500 show the superiority of our approach to existing ones.
So rather than only looking ahead and wringing our hands about the economy, the recovery, PIIGS and all the other noise, let’s try ignoring that for a moment and practice driving using our rear view mirrors.
The average ratio shown over Shiller’s data was 18.28. From 1811 to now the average cyclically adjusted P/E10, CAPE, is 16.36 and from 1950 on the average is 18.59. The current ratio is 20.31 indicating a slightly overvalued market which will return slightly less than average over the next ten years. Too many numbers? OK then, time for a graph.
Anyone paying attention to Shiller ten years ago would have been saved a lot of pain. As he and Campbell suggested the high CAPE ratio correctly predicting poor returns during the naughties. Shiller’s detractors point out that he said the high CAPE in 1996 would lead to poor returns over succeeding decade. Whereas the actual real returns over the next decade were a respectable 47%. I’d suggest that is simply a matter of trends going further and for longer than anyone can ever predict. Plus Shiller never claimed a perfect correlation. Like any stock market tool, it should not be used in isolation but in context.
Let’s look at some averages.
This chart shows the 10 year returns based on ranges of P/E10 CAPE (blue squares). The red diamonds are the number of occurrences. Keep in mind these are real returns adjusted for inflation, hence they appear considerably lower than the usually referenced nominal returns.
Now let’s drill down into a more relevant subset for the current ratio. This chart shows the average ten year returns for CAPE ratios around the current P/E10 CAPE of 20.31.
Naturally the more detailed I drill down into the data the less reliable any assumptions are. However, in general the trend is pretty obvious. The higher the P/E10 the lower the expected returns. P/E10 ratios around the current 20.31 deliver considerable less than the average return of 57% across the period used, 1950-2010.
Let me leave you with a similar warning to Shiller’s warning in his 1996 paper with a small adjunct. Driving by only looking in your rear view measure is hazardous and great caution must be used. Driving without rear view mirrors is also hazardous and great caution should be used. So look forward, but be mindful of the past.
Related posts:
Hi Dean,
another interesting post. Unfortunately it seems I need to register/pay to see the Hansen and Tuypens paper, but I assume that the S&P composite data was share capital only, not with dividends reinvested (the accumulation index).
Another question I’d have is that the market liberalisation during the 80’s (and likewise automation in the 20’s-30’s) would have improved both earnings and returns and therefore I’d have liked to see the comparison long term averages broken down into 10 or 15 year blocks. Maybe this liberalisation freed companies and capital to make better returns for the same earnings reinvested?
It’d also be interesting also to see the same analysis applied to the ASX200, although so many structural changes have been made to our markets here in the last 30-50 years I am not sure what it would tell us. IMO Australia is a relatively immature market with plenty of growth potential.
Hi Darren
Shiller’s data for the S&P comp includes dividends. The real returns are total returns including dividends all adjusted for inflation using the CPI. You can get his data set from here http://www.econ.yale.edu/~shiller/data.htm.
Good ideas for future work for me. That’s exactly the sort of feedback I was hoping for Darren, so thanks a lot.
In his warnings Shiller said, “There could be fundamental structural changes occurring now that mean that the past of the stock market is no longer a guide to the future.” My predisposition is things never change and the consistency of long term returns certainly support that; however, it will be interesting to see what the data says.
Up to now I haven’t had access to good long term data for the ASX. Though I should be able to source some through Bloomberg or the like now. If anyone can point me in the right direction I’d appreciate it.
http://www.bwts.com.au/text.cfm?14
I’m a firm believer that in the long term markets and particularly indexes are mean reverting. If the PE is elevated in that period that will increase the long term mean. And of course most of the time the structural change turns out to be a fad/bubble/just leverage. How long it takes to get mean is the question.
The single most important idea I ever got was the idea that markets are ketokudic ie. fat tailed and not normally distributed. This is probably due to human behaviour/ herd behaviour/ constant psycholgoical factor. The chances of a 1:100 year event occurring happen about once every 5-10 years.
Minski’s stages of a bubble are interesting in identifying bubbles but give you no insight into how to cope with a bear market. For that I suppose you need to live through one and they only happen every decade or 2. In that sense investors who have been around for a long time like Buffet have an edge.
Some hedge and other funds use talebs barbell strategy. Market weight but insure for the known risks and use the rest of the portfolio for risks with a perceived asymetric return profile. The problem is that the premium on the out of the money puts is going to take up 2-5% of returns p.a. I wonder whether you would be better off keeping 20% in cash instead and just investing in the perceived asymetric return investments. In the long term it depends on what you’re paying for that implied volatility I suppose.
Just my random thoughts. Hope the course is going well!
Sean, speaking of mean reversion are your familiar with the BMW Method? I like to use these charts for my US stalwart investing http://invest.kleinnet.com/bmw1/xref.html
You can find out more about the BMW Method here http://bmwmethod.com/ and there is a discussion board at The Motley Fool on it, but the quality of that has slipped in the last year or two since the founder was bullied off.
I’m not familiar with the ketokudic, but certainly am aware of skewness and fat tails. I’ve already got the feeling at Uni that challenging the wisdom of normal distribution is not the way to win friends amongst lecturers. I think it takes a very special person to be able to invest like Taleb. His style certainly does not suit my personality, yet being aware of what he has to say is valuable. Though boy oh boy is he an arrogant tosser! I was left wondering if he has some sort of cocaine psychosis. “I’m so frigging marvelous and anyone else who is successful is just lucky” the world according to Taleb. It’s like he wanted to be insulting so that only people who could see past his arrogance and insults could see his message. Saying all that I really enjoyed his books.
I think you have to regurgitate the stuff. That’s academia for you.
I spelt it wrong, it’s actually ketokurdic. I think ketokurdic just means fat tailed on both ends. It’s a nice long word though. Black-scholes assumes normal distribution. Everyone now knows this is not the case. But how to account for it ?
Taleb’s investment strategy just sounds silly. I think he was a good quantitative analyst but not a very good investor. I haven’t read his book, just ahve noted funds saying they used his barbell strategy. Personally, I’d rather invest in Lotto than out of the money puts in the long term.
I like the screens. The BMW method sounds interesting. I’ll have to find out more. CAGR sounds like a good way to screen as long as you can assume the company is sound and going back to average. I just have no confidence with doing this in scale with an individual stock but I’m comfortable doing this with an index as a) firm specific risk diversified out and b) unlikely the economy of a country will go belly up so you can bank on mean reversion back to fair value based on mean reversion in %profit share companies & GDP growth.
I found it interesting that Buffet calculates returns on his investments based on intrinsic value increases rather than price and benchmarks this with changes in intrinsic value in S&P500.
I wish there were better tools available on the Australian market. I don’t invest in US markets at the moment as all my expenses are in AUD and I don’t think the AUD has peaked and I couldn’t cope with the currency loss.
This is a post I found interesting this week from a guy trying to implement portfolio optimisation:
http://forumserver.twoplustwo.com/30/business-finance-investing/passive-portfolio-construction-q-working-perfect-weights-concepts-723193/
Hi Sean
I don’t understand why a retail investor would bother would portfolio optimisation. Go with what you know I say. Stick with you circle of competence. I don’t know bonds or precious metals, I’ve only got average knowledge on REITs. There is no way to be an expert across all those areas and is there really a need to optimise? For example, tradies should stick to property as that is what they know, is there really any need for them to diversify into areas where they have no edge and then have to TRUST others? Do I have any need to invest in Bonds? Concentrate on your strengths and forget about your weaknesses.
I agree it would be much easier to buy into a mutual fund like vanguard in a balanced or growth portfolio. What I found interesting was that he was actually trying to implement CAPM and the Markowitz optimisation. I thought it was interesting to see someone go from first principles, which is what you do in a course (CAPM models) to discovering the optimisation from recent actual data, trying to figure a point on the efficient frontier and calculating drawdowns on the portfolio.
One powerful idea is that that you gain no extra reward from firm specific risk but a free kick from diversification. How you reconcile that with Buffet or concentrated investing is a challenge.
What I do is buy a mutual fund in super for market risk & diversification and tilt my portfolio allocation with a discretionary account in ETF’s, cash and individual stocks.
Sean, is there a need to reconcile theory with practical proof? Buffett (two t’s), Berkowitz and others who invest for a living show how it should be done. Reward is gained from firm specific risk. Diversification can at best result in market performance. Yes, I will have to bite my tongue and regurgitate CAPM and EPH, but I’m hoping I learn something of value from it. Few theories stand the test of time, that’s why I say do what works for you. The most important thing I want anyone reading this site to come away with is that the key to investing is knowing yourself and finding what works for you. There are many paths and many theories, but knowing what works for you is the key.
So I guess I found that link frustrating, as it was taking a theory and trying to apply it without regard to the individual. I give that approach to investing minuscule chance of outperforming.
By the way, thanks for you interesting discussion. Over three hundred unique people visit this site everyday and you’re one of the few who takes the time to comment. Thank You.
Hi Dean, your site is an excellent read, so keep it up. Thank you for putting in the effort for what is arguably one of the best sites in Australia on investing.
I guess we have differences in our investing so it is interesting to see the way they both work. I guess your way is by concentrating on firm specific risk with an edge. For me I tend to concentrate on market timing the index. I use the index to diversify firm specific risk so I can take a large market position. The academic research would indicate that neither of our strategies could work over the long term. Which is interesting but academic (unless it turns out to be true!).
With the course, I think you will learn some useful things. I think you may enjoy the corporate finance subject more than the others.
Dean and Sean thanks very much for link to the PER data site. Great effort by Colin to maintain it.
Its scary how much different PER data sources vary.
BTW do you mean markowitz? – berkowitz was the psychopath wasn’t he?
re MVA optimisation, it is useful as a maths proof of the concept of diversification. it is also useful for investment consultants to imply a level of precision that does not exist. as nobody can accurately estimate any of the inputs and the output is highly sensitive to the inputs it shouldn’t be considered a useful practical tool.
you might also like to read this -
http://online.wsj.com/article/SB123093692433550093.html
Hi Jeff, I did mean Berkowitz as in Bruce Berkowitz from Fairholme.
Leave your response!
SUBSCRIBE by RSS
or Subscribe by EmailTags
Categories
Archives
Free Spreadsheets
Blog of the Day
Blogroll
Aus/NZ Blogroll
Recent Posts
Most Commented
Quotes