He illustrates the phenomenon with a few charts:
The first one shows earnings per share [EPS] rising faster than the S&P 500.
The next shows the effect that has on the P/E ratio.
Finally, he shows that the earnings yield [E/P] on the S&P 500 is now much higher than it was throughout the 1990’s.
On this last point, Eddy says:
Today, the market’s p/e ratio is about 15.6 (based on trailing operating earnings). That works out to an earnings yield of about 6.4% (1 divided by 15.6). The market’s earnings yield has most often been about 1% to 2% lower than the yield on the 30-year Treasury bond. Now it’s 1.4% higher. (Note: This is slightly different from the Fed model which uses estimated earnings).
This seems to be where Eddy is simply looking at a historical data set that just doesn’t go back far enough. While 15 years may appear a long time horizon, comparing the bear market of the early 2000’s to the bull market of the 1990’s simply is inadequate. One must go back at least as far as the previous bear market of the 1970’s to even begin to make comparisons.
Which is why we went to an obscure (sarcasm) text called The Intelligent Investor by Benjamin Graham to source our last image, the table appearing at the end of this post. It appears on page 71 of the revised edition.
To start with, we note that Graham did not consider a single year’s earnings when comparing stocks to bonds, but rather a trailing three year average earnings yield. This alone would make Eddy’s comparisons look more expensive. Simply by eyeballing the earnings chart we can ballpark trailing three-year earnings at around $70 - giving us an earnings yield for comparison purposes of roughly 5.4%.
Secondly, we can evaluate whether earnings yields higher than bond yields are particularly unusual in the historical context. Is a 1% to 2% discount to bond yields “normal?” Since Graham’s table uses the AAA corporate yield as its benchmark, we gather the same data from the Federal Reserve to get a bond rate of 5.64%, which was the average rate on AAA bonds for the week of August 4, 2006.
Now let’s compare the current difference between the earnings yield and the bond yield to Graham’s historic data. Today the earnings yield/bond yield is approximately 0.96, which is closest to the level in 1958. In two of the sample years it was lower, in three it was higher - in two cases significantly so.
Call it behavioral patterns (as bear markets continue we increasingly distrust stocks) or long-wave cycles (Elliott wave theory) or whatever you want. But increasingly research shows that P/E ratios tend to run in long (15-20 year) cycles of expansion and contraction. So knowing the current P/E is not enough to tell you whether the time is right to buy stocks. You also need to know whether the P/E next year is more likely to be higher or lower.