From the March 2010 newsletter:
Last time we discussed Q, which seemed to predict future 10yr returns decently well for the US stock market. This time, I introduce a valuation metric that many of you are probably familiar with, the P/E ratio. It is typically defined as price divided by earnings over the prior 12 months. The biggest problem with this definition is that earnings over one year are highly volatile, making this version of the P/E ratio not very useful.
One way to make the P/E ratio more meaningful is to use earnings over a much longer period of time. Using 10 years or more of earnings is ideal. The chart below shows the P/E ratio (using 40 years of earnings!) vs Q and the following 10-year return of the S&P 500 index. It looks like this P/E ratio does a good job of predicting future 10-year returns. In fact, it’s even slightly better than Q (r-squared of 0.83 vs 0.69).
Both Q and the P/E ratio are useful metrics to estimate future stocks returns. However, like Q, the P/E ratio does not predict short term results that well. So when I recommend investing less in US stocks than your target allocation due to very high levels in Q and P/E, the intended holding period is many years.
Next time someone mentions the P/E ratio, know that it’s only useful if earnings over many years are included and you are estimating returns for at least 5-10 years. It will not help anyone guess where the S&P will end this year.