Long term equity indicator turns negative

The primary long term asset allocation indicator that I use for equities has recently gone negative for the first time since the crisis began. The cyclically adjusted p/e ratio averages the last ten years earnings in real terms and compares them to the current price. It is far superior to one year p/e ratios as it effectively smoothes out for economic cycles. Indeed, one year p/e ratios have been shown to have no predictive power at all with regard to future equity market returns.

I take the publicly available Robert Shiller data on the US market which shows real earnings and index levels going back to 1871. I then look at the real return for every individual one, two and ten year period and compare this with the cyclically adjusted p/e ratio at the start of the period. Obviously, the idea is that subsequent returns should be lower when the starting valuation is higher and vice versa.

What I have found by playing around with the data series is that looking at quartiles provides a fairly good indicator. For example, if I take every reading where the cyclically adjusted p/e ratio is in the bottom quartile, I find that the average subsequent real compound annual return (CAGR) is 10% for the following year, 9% for the following 2 year period and 6.6% for the following 10 year period. This is from a total of 387 observations (the data set is monthly). If, on the other hand, I take every reading where the cyclically adjusted p/e ratio is in the top quartile, I find that the average subsequent real CAGR is barely 0% for each of the one year, two year and ten year periods.

This tells me that, statistically speaking, I should be reducing equity holdings when the cyclically adjusted p/e moves into the top quartile. Unfortunately, according to the Shiller data, the ratio for US market has just in the past month moved back into the top quartile for the first time since the crisis began. So prospects for long term equity returns don’t look that rosy any more.