Can CAPE Forecast Returns Without Timing a Market Crash?
A valuation measure can contain information about average returns over several years without telling anyone when the market will fall. That distinction is easy to lose when a long-term ratio is presented as a warning light.
Return predictability is about a horizon
Suppose a study tests whether today's valuation is related to the cumulative return over the next five years. That is a statement about the distribution of multi-year outcomes in a sample. It does not say the return will be smooth, negative in every year, or lower next month.
John Cochrane's “The Dog That Did Not Bark” discusses the relationship between dividend-growth forecasts and return predictability. Palazzo's paper extends the discussion by separating price appreciation from dividend growth and testing the proposed CAPE-H measure.
What the CAPE-H manuscript reports
The manuscript reports improved out-of-sample forecasts of cumulative price appreciation at three- and five-year horizons relative to traditional CAPE. It finds little robust out-of-sample predictability for dividend growth under either measure. Its interpretation is that the valuation signal, when present, operates mainly through price changes rather than predictable dividend growth.
The paper also studies lower-tail outcomes. It reports that high CAPE-H is associated with greater odds of a market correction at several horizons, while crash prediction is less reliable at the longest horizon. The author notes that bottom-decile outcomes are rare, leaving relatively few effectively independent long-horizon observations.
Why a forecast is not a crash timer
Long-horizon return observations often overlap. A five-year return measured each month shares most of its months with the prior observation, so the number of data points can overstate how much independent information the sample contains. Rare crashes create an additional statistical challenge.
Even a valid relationship between valuation and future returns leaves the path unknown. Markets can remain at elevated valuations, rise further, or decline gradually. A study may estimate that high starting valuations are associated with weaker average outcomes, but that is different from identifying a date, trigger or likely size for a selloff.
Use research findings to understand the limits of a valuation metric, not as a standalone instruction to buy or sell. Start with what CAPE measures, then visit CAPE, Accounting and Long-Term Stock Returns: A Plain-English Pillar Guide for the related accounting and measurement guides.