Improving CAPE, One Stock at a Time

Key Takeaways

• Traditional CAPE weights earnings, not market cap
• New method computes CAPE per stock, then aggregates
• Correlation rises from 0.68 to 0.76, R² to 0.58
• Forecast error drops to 3.1%, improving portfolio decisions

Pulse Analysis

The CAPE ratio, popularized by Robert Shiller, traditionally divides an index’s current price by the average earnings over the previous ten years. While this approach smooths cyclical earnings volatility, it implicitly assumes that each constituent’s earnings contribution mirrors its market‑cap weight. In practice, high‑margin firms can dominate earnings figures, and index turnover introduces historical earnings that never reflected the index’s composition, distorting the signal for future returns, especially in the U.S. market where the metric has lagged.

Ma et al. address these flaws by constructing a stock‑level CAPE: each S&P 500 component’s price is divided by its own ten‑year earnings average, then the resulting ratios are weighted by the firm’s market cap within the index. This granular method preserves the true earnings‑price relationship for every company and eliminates the bias from index rebalancing. Empirical testing from 1974 onward shows the revised CAPE’s correlation with ten‑year real returns climbs to 0.76, with an out‑of‑sample R² of 0.575, and the mean absolute error contracts to 3.1 %. Such statistical gains are not merely academic; they translate into tighter confidence intervals for long‑term return forecasts.

For practitioners, the enhanced CAPE offers a more reliable barometer for strategic asset allocation, risk budgeting, and valuation benchmarking. Portfolio managers can integrate the refined metric into factor models or as a standalone overlay to adjust equity exposure during periods of elevated valuation risk. However, implementing the approach demands comprehensive historical data on index constituents and their monthly weights—a non‑trivial data engineering task. As data providers streamline such datasets, the method could become a new standard for valuation analysts seeking a robust, forward‑looking gauge of market expectations.