Posted by on Jan 9, 2013 in Academic, Economics, Investing, Statistics

According to the efficient market hypothesis no publicly available information can be used to produce excess returns consistently. However, a persistent pattern exists between current stock prices, earnings, and future returns, all of which are readily available to the public, which could be used to generate excess returns over the long run. An analysis of the S&P composite from 1871 to 2012 showed that a strong negative correlation exists between the 10 year average P/E ratio of the S&P composite (PE10) and the succeeding 20 year return of the the composite. Presumably, one could buy stocks when the ten year average P/E is relatively low (below 15 or so) and sell stocks when the P/E becomes too high (above 25 or so), and beat the market in the long run.

PE10 is very strongly correlated with returns in the succeeding 20 years.

PE10 is very strongly correlated with returns in the succeeding 20 years.


This profit opportunity should not exist if the U.S. stock market were truly efficient. According to the efficient market hypothesis, the fact that this correlation exists would mean investors would begin to buy more stocks when P/E ratios begin to fall below their long run average in anticipation of increased returns and sell when P/E ratios rise too far above their long run average, keeping prices relatively stable in relation to earnings and thus eliminating any future excess profit opportunities from this data. This is obviously not the case.

This pattern has persisted for various periods of time in the past and persists to this day, years after it was first observed and discussed publicly. In fact, the pattern has shown striking accuracy in predicting stock returns in recent years. In March of 2009 the PE10 stood at 13.32, the lowest it had been in over 20 years, and below the historical average. That month turned out to be market bottom of one of the largest stock market declines in history. An investor who would have bought at that time would have seen an over 50% return within a year. Conversely, the PE10 hovered around 27 for much of 2007, an obvious sell signal according to the pattern, which if heeded would have prevented an investor from experiencing the decline that began near the end of that year.

PE10 is strongly correlated with returns in the succeeding 1 year.

PE10 is strongly correlated with returns in the succeeding 1 year.

potent