Because it has been bearish for several years now, many people have stopped paying attention to the CAPE indicator. It is unfair. You should give it a second look.
I’m referring to the cyclically-adjusted price-to-earnings ratio made famous by Yale professor and Nobel laureate Robert Shiller. When we judge it over a longer horizon – say, since the top of the internet bubble— the CAPE is doing quite well. Although stocks since that peak have not performed as badly as the CAPE ratio projected at the time, they still posted one of their worst 22-year returns in US history.
The CAPE ratio is similar to the traditional P/E ratio but for the denominator. Instead of using one-year earnings, CAPE uses 10-year inflation-adjusted average earnings. Shiller argues that this substitution greatly increases the predictive power of the ratio.
Like the P/E ratio, higher CAPE readings suggest a more overvalued market. Since 1881, according to data from Shiller, the highest CAPE level was 44.2, recorded in December 1999, at the height of the dotcom bubble. Prior to the go-go years of the 1990s, the highest CAPE reading occurred at the top of the stock market before the crash of 1929, when it reached 32.6.
At the stock market’s recent all-time high in January of this year, the CAPE stood at 38.6. Although not as high as December 1999, it was still higher than pre-crash 1929 and above 98.5% of all other monthly readings in Shiller’s database.
CAPE and the Internet bubble
To judge the very bearish message of CAPE at the top of the dotcom bubble, I built a simple econometric model based on the correlation so far between its values dating back to 1881 and the actual total return (adjusted for inflation) of the S&P 500 over the next 22 years. years. I chose this period because it is the time since the peak of the dotcom bubble.
In early 2000, this model predicted that the real total return of the S&P 500 over the next 22 years would be minus 0.6% annualized, a weaker return than ever before in US history. As fate would have it, the market’s actual real total return was 3.7% annualized, more than four percentage points annualized better.
Although the actual market return is significantly better than the model predicted, it remains one of the worst 22-year returns in US history. According to Shiller’s data, in fact, it’s worse than more than 85% of all sliding periods of this length since 1871.
In other words, contrary to a projection that the stock market over the next 22 years would be at the zero percentile of the historical distribution, stocks have actually arrived at the 15th percentile. In the messy world of stock market predictions, this close deserves at least an honorable mention.
You could say that the projection deserves even more than an honorable mention. Given that the actual 22-year stock market return is two standard deviations below what the model projected, it is possible to conclude, at a 95% confidence level, that the model projection was on target.
The next 22 years
We reject the current bearish message of the pattern at our peril. Based on the current CAPE situation, my econometric model predicts that the S&P 500 will produce an inflation-adjusted total return by 2044 of 3.1% annualized, less than half of the historical average of 6, 6%.
Certainly, if the actual performance of the stock market over the next 22 years is above this projection as much as it has been for the past 22 years, then the stock market as a whole will be doing well by 2044. But it seems dangerous to hope for such a happy outcome, as it is entirely possible that the stock market over the next two decades will underperform rather than exceed its expected return. After all, the stock market can’t stay above average forever.
Another reason not to dismiss the bearish message of CAPE is that the same story is told by other valuation indicators with excellent long-term records and which have entirely different theoretical underpinnings. Consider buffett indicator, for example, which is the ratio of total stock market capitalization to gross domestic product. It bears the name of
CEO Warren Buffett because of a comment he made two decades ago that the indicator is “probably the best single measure of where [stock market] valuations hold at some point.
Higher values of the Buffett indicator indicate a more overvalued market. The ratio is currently at the 97th percentile of the historical distribution.
Or consider the so-called Q ratio, based on research by the late James Tobin, winner of the Nobel Prize in Economics in 1981. It is calculated by dividing the total market value by the companies’ replacement costs. As with the CAPE and the Buffett indicator, higher Q ratios indicate a more overvalued market. The Q ratio is currently above 98% of historical readings.
The bottom line? We could get lucky and outperform the lackluster returns projected by CAPE and other valuation indicators. But luck is not a strategy.
Write to [email protected]