Is a High CAPE Cause for Alarm? Part 1: CAPE’s Relationship to Stock Returns
Market predictors have taken many forms over the years, but no formula or person has ever gotten it completely right. The market is unpredictable. That’s just the way it is. But that hasn’t stopped people from trying, and it doesn’t necessarily mean we should ignore them all.
In the mid-1990s, Yale professor and Nobel laureate Robert Shiller popularized the concept of the cyclically-adjusted price-earnings ratio (commonly abbreviated either as CAPE or PE10) as a useful predictor of subsequent stock market returns. The PE10 measure is found using the following formula:
A 1998 research article published by John Campbell and Shiller justifies this measure as a way to remove cyclical factors from earnings, though there is no particular reason to pick ten years. The concept stems from Benjamin Graham and David Dodd’s classic Security Analysis text, first published in the 1930s, which suggests using a period of seven to ten years to average out business cycles from the earnings data. Today, Robert Shiller provides updated data on the key variables used to calculate PE10 at his website.
Using this data source, I plotted the historical values of PE10 at the start of every year going back to 1881 (Figure 1). (Shiller’s data extends to 1871, but 10 years of earnings are needed to calculate PE10.) While the historical average is 16.6, PE10 today is above 27. With respect to the historical data, this is the nosebleed section, as PE10 previously only reached these levels in 1929 (the year which witnessed the stock market crash that sparked the Great Depression), and then again with an even more extreme run-up in PE10 values in the late 1990s.
Figure 2 updates the research results from Campbell and Shiller’s article. It shows the PE10 value at the start of each year plotted against the compounded real total returns on the S&P 500 in the subsequent 10 years.
You might see why the high PE10 value today creates concern among investors. There is a negative relationship between these variables, as the higher PE10 suggests lower subsequent real returns for stocks. The relationship is far from exact as there is still a lot of randomness to be found in the numbers, but statistical analysis suggests that PE10 can explain about 31% of the fluctuations in real stock returns over ten-year periods.
Historically, the compounded real return since 1871 on the S&P 500 was 6.7%. However, what this model predicts is that the subsequent real compounded returns on stocks over the next 10 years will only be 1.5%. This projection is well below the historical average because PE10 is well above its historical average.
In part two, I will consider how much weight the PE10 should hold in making investment decisions.