Valuation and the CAPE Ratio
Siegel's own contributions to valuation measurement, and how starting valuation affects subsequent long-run returns.
Beyond the historical-returns case, Siegel devotes significant attention to valuation — specifically the cyclically-adjusted price-to-earnings ratio (CAPE, also associated with economist Robert Shiller), which smooths out short-term earnings volatility by averaging inflation-adjusted earnings over the trailing ten years rather than using a single year's earnings, producing a steadier valuation gauge less distorted by any one unusually strong or weak earnings year.
The book's empirical finding on valuation is that starting CAPE level has historically had meaningful predictive power for subsequent long-run (10-year-plus) real returns — buying at a historically low CAPE has tended to precede stronger subsequent long-run returns, and buying at a historically elevated CAPE has tended to precede weaker ones — while being far less useful for predicting shorter-term (one-year) returns, where the relationship is much noisier and less reliable.
Averaging ten years of earnings smooths out the distortion a single unusually strong or weak year would otherwise cause in a standard, single-year P/E ratio.
Siegel's explanation for why CAPE's predictive power shows up mainly at long horizons is that valuation extremes take time to correct — an expensive market can stay expensive, or get more expensive, for a year or even several years before eventually mean-reverting, so a high starting CAPE says very little about next year specifically, but says considerably more about the average return over the subsequent decade, once that mean-reversion has had time to play out. He is careful to caveat that this is a statistical tendency across history, not a precise, reliable timing tool for any single specific period — some genuinely high-CAPE periods went on to deliver perfectly reasonable subsequent returns, and the relationship has itself shifted somewhat across different eras.
- CAPE smooths earnings over a trailing 10 years to produce a steadier valuation gauge than a standard single-year P/E ratio.
- Starting CAPE has historically had meaningful predictive power for subsequent 10-year-plus real returns, but is much less reliable for predicting one-year returns.
- Valuation extremes take time to correct, which is why CAPE's predictive power is concentrated at long horizons rather than short ones — and even then it is a statistical tendency, not a precise timing tool.