Here are another set of charts comparing the P/E10 Ratios for the S&P 500 and subsequent 10-year annualized real returns, courtesy of Mebane Faber based on Professor Shiller’s data. Can we really decide if the market is “overvalued” or “undervalued” by looking at one single number?
As noted in my earlier post about P/E ratios as a long-term predictive tool, the “P/E10 ratio” is the market’s current share price divided by average earnings over the last 10 years. By taking a long-term average as opposed to the more common single past year’s earnings, the idea is to smooth out the noise and bumps. The initial use of this ratio has been credited to famous value investor Benjamin Graham.
In the first chart above, you can see what appears to be a very strong relationship between lower P/E10 ratios and future 10-year returns. Everything is neat and tidy. High P/E10 = Bad. Low P/E10 = Good. The approximate current price of the S&P 500 is noted by the highlighted grouping. This would suggest that the median expected annualized real return for the market over the next decade would be about 5%.
In this next chart, Faber splits the data up into deciles instead. He notes a more precise trend of “great returns up to about 13, then decent returns up to about 20, then crappy returns over 20.” I personally just see a less convincing relationship. If there is such a strong correlation between lower ratios and higher returns, why should the fifth decile with P/E10 of 13 to 15 perform worse than P/E10s of 15 to 19? Hmm.
Finally, we have the actual data points. We see that although there is a nice trendline that can be created from such scattered data points, for any given P/E10 ratio there is a very wide variety of returns. Accordingly, in my humble opinion, I would be careful not to make P/E10 your main basis for setting asset allocations. It’s a nice idea that scores well in the common sense category, but in reality has been far from perfect.