I’ve been writing about the CAPE ratio since 2018. A reader once pushed back and asked me to defend it. My answer then: it’s not a market timing tool, but it’s one of the few signals with a track record of predicting long-term returns. I still believe that. A new academic paper just made the case stronger, while also exposing a flaw in how CAPE has always been calculated.
I’ve covered the basics before: what CAPE is and why it works, and why it belongs in your investing toolkit alongside a healthy respect for what it can’t do. This research builds on both.
The Problem With Traditional CAPE
Researchers at La Trobe University, Auckland University of Technology, and Massey University identified a structural inconsistency at the core of the Shiller CAPE calculation. The traditional CAPE divides the S&P 500 price level by aggregate earnings, which implicitly weights each company by its earnings. But the S&P 500 itself is market-cap weighted. Two different weighting schemes can equal two different answers.
Apple (~5.9% of the index by market cap) and Nvidia (~7.2%) are two of the largest positions in the S&P 500. Both trade at P/E ratios well above the index average, around 33 and 41 respectively, versus roughly 25 for the broader market. That means their earnings will represent a smaller slice of total S&P 500 earnings than their market cap represents of the index. In the traditional earnings-weighted CAPE, they get underweighted. The traditional CAPE ends up understating how expensive the index is, because higher-valuation companies are currently dominating the largest positions.
The Fix: Component CAPE
The researchers built what they call a “Component CAPE” and calculate CAPE for each company individually, then weight by market cap to match how the S&P 500 index works. The Component CAPE averages around 29.7, versus 21.7 for the traditional version. The gap widens during periods of high valuation dispersion, like today, when mega-cap tech trades at multiples that are more expensive than the rest of the market.
The Component CAPE also predicted long-term returns more accurately, improving out-of-sample predictive accuracy by more than 10% over the traditional version across both 5- and 10-year horizons, holding up across different time periods, different earnings adjustments, and different methodologies.
What This Means for Stackers
This doesn’t change much for us because CAPE is not a market timing mechanism. It doesn’t tell us when the market will fall. It tells us that when valuations are elevated, expected returns over the following decade are lower. No version of CAPE, improved or otherwise, is going to help us call the next correction.
The Component CAPE is a more accurate version of the same signal. It reinforces the message rather than changing it.
If anything, today it suggests the market is even more expensive than the traditional CAPE implies today. The practical improvement in day-to-day investment decisions is minimal but the case for CAPE as a long-run return signal got stronger.
In my personal portfolio and advisory work, I use valuation indicators like CAPE to inform future expected returns, shape retirement income projections, and consider allocation tilts across asset classes and geographies. The Component CAPE didn’t change any of that because we were already tilted away from the S&P 500 and the more expensive stocks. The inputs got slightly more precise; but my conclusions stayed the same.
A more precise ruler doesn’t change what it’s measuring.
