Everyone knows about the P/E ratio, but nobody knows how to use it. Price / last year’s earnings and Price / next year’s earnings are pretty much worthless, yet these are the formulas we hear referenced the most. Why? Because they make the current situation look reasonable. Normie P/E is bunk and the current situation is insane.
What’s also insane is using a highly volatile variable like 12-month earnings as a benchmark of value. Earnings swing wildly from year-to-year for a given company, and but they are also cyclical for the S&P500 as a whole. They can collapse to near zero in recessions, resulting in normie P/E going through the roof – thus signalling “sell!” just when any reasonable metric would say “buy.” Likewise, when earnings are nice and fat, at the top of an expansion, the P/E will of course look lower, suggesting we hold.
Professor Robert Shiller gets credit for simply taking a 10-year running average of earnings for the S&P500 and using that as a more solid denominator (though Ben Graham mentioned this for individual stocks). A decade is usually enough to account for a full business cycle, and when you normalize earnings like this you see a nice steady growth curve of something like 6% after inflation. The S&P itself swings up and down around this line – when beneath the line, it’s a buy, when above, it’s a sell. It works great as a slow timing indicator, the kind of metric you might use to adjust your equity and bond allocations every year.
There’s one more little glitch in earnings data that keeps Shiller’s CAPE from joining the likes of Price/Sales or Market Cap / GVA, and that’s the margin cycle. Average corporate profit margins have their own swings, ranging from roughly 3%-10%, averaging 5.4% over the last century. John Hussman, the dean of S&P valuation, has often mentioned this deficiency, since complacent bulls like to point out that CAPE, while high at 30, is still far beneath the all-time high of 47 achieved in early 2000 at the top of the dot-com bubble. The trouble is that it is hard enough to get normies to normalize profits, let alone profit margins, since the margin data is expensive and only goes back to the 1970s.
While scribbling another warning on the situation this summer, it occurred to me that one might divide GDP by corporate earnings data (90 years of which are freely available on the Fed’s website), and use the result as a proxy for margins. It works – the result correlates nicely with true margins – and CAPE, using margin-normalized profits, has a much stronger correlation with actual subsequent market returns (roughly 0.90 with 12-year total returns). I pinged Hussman about this, and he verified the methods and results. So there you have it, a margin-normalized, cyclically adjusted price-to-earnings ratio. Should we call it the Ritger PE? Nah, just margin-adjusted CAPE (MADCAPE?).
Shiller PE (CAPE):
From Hussman, MADCAPE in blue, actual subsequent S&P 500 returns in red (returns stop in 2005 because that was 12 years ago):
In conclusion, sell, sell, sell! This market is more expensive than either the summer of 1929 or the winter of 2000, the two worst times ever to be invested in US stocks.