The Cyclically Adjusted Price-to-Earnings Ratio, typically referred to as CAPE or sometimes as the Shiller P/E after its inventor, Nobel Laurette Robert Schiller, is the ratio of the current stock price to the average of the past 10 years of earnings, with those earnings adjusted for inflation.
We don’t like it for the same reason we don’t like any simple screens. However, there are other reasons:
- Future accounting earnings (growth) which a P/E indicates are nominal earnings and incorporate the anticipated effect of inflation on those earnings, so comparing price that anticipates those earnings to inflation-adjusted earnings is strange. P/E is also based on the risk to future earnings so comparing price to inflation-adjusted earnings could be justified if inflation-adjusted earnings somehow indicate what the discount for risk in the price should be. However, inflation adjusted earnings are lower than nominal earnings (with positive inflation), so that’s not likely the case.
- The idea behind the 10-year average earnings in the CAPE is that earnings revert to their long-run average. But that is not how earnings behave: Earnings typically grow from the past and it’s that growth that is priced. The observation that the CAPE is typically high (37 in 2024) suggests the denominator in CAPE is too low relative to expectations of future earnings.
- Consider a case with earnings at 100 ten years ago, declining to 50 over the subsequent eight years then recovering to 100 currently in the 10th The 10-year average is 60. At a current price of 2000, the CAPE is 2000/60= 33.3 compared with the trailing P/E of 20.0. This might be an extreme example, but including those down years with prospects now up is misleading. And so for a case where the earnings of 100 ten years ago increases to 150 for eight years but then declines to 100 on bad news: The CAPE = 14.3. While the past can be a guide to the future, more so the recent past, is the future which is being priced, not the past.
- The CAPE came to prominence in 1996 with a claim that the stock market was overvalued due to “irrational exuberance.” That was a bad call, at least for the short run: In the following three years, the market delivered high stock returns, the S&P 500 doubling. It did signal the market downturn in 1999 when the market was very high, but all multiples did that: The training for P/E the S&P 500 reached 33 with the P/B over 5. Had it existed in the 1960s and 1970s, the measure would not have signaled the dismal returns awaiting investors. Since 2014, CAPE has signaled overpricing many times while stock prices steadily increased.
CAPE comes with the same warning about relying on multiples. CAPE does add other information to current earnings in the standard P/E ratio, but that information does not seem to help much. There is little evidence that CAPE predicts returns in the near-turn, although there is some evidence that it predicts longer-term returns. But we have not had many independent long periods in recent history to sample from in order to develop a reasonable level of confidence in that finding.
An argument for the CAPE is that earnings can be affected by business cycles, low in recessions or high during booms. That affects the standard P/E, high in recessions and low in booms. But that ignores the pricing in the P/E: If the P/E is high during low earnings recessions, that can mean that the market sees a recovery…growth in earnings after the recession. If the P/E is low during high earnings booms, that mean that the market sees that boom not continuing. And it can be the case that P/E is low during recessions….as in the 1970s when the P/E for the S&P 500 fell to 7.0. That is the market having doubts about recovery. In boom times of the 1990s, the P/E reached 33.0. That is the market speculating about even more earnings in the future.
Don’t let this discussion of P/E ratios take you eye off the ball: A value investor does not screen on multiples but rather delves into the fundamentals to evaluate future growth prospects and the risk to that growth. That said, when a multiple gets out of line from its normal level, even CAPE, that’s a cue to investigate.