Skip to main content
← All News

Markets · November 16, 2026 · 2 min read

SF Fed research links stock valuations to labor-market fundamentals, without proving sustainability

By Econ Data Tools Editorial Team

Published November 16, 2026 • Updated November 16, 2026

A November 16 Economic Letter fits 89% of historical CAPE variation with four variables. Its model value is 11.6% below the actual ratio, not evidence that a stock-market decline is certain.

A November 16, 2026 Economic Letter from the Federal Reserve Bank of San Francisco examines how longer-term economic variables relate to U.S. stock valuations during the AI boom. Kevin J. Lansing and Federico C. Nucera find that four variables account for much of the historical movement in the cyclically adjusted price-earnings ratio, or CAPE. They caution that fitting past data does not establish that current valuations are sustainable. CAPE divides the inflation-adjusted S&P 500 price index by inflation-adjusted earnings averaged over the preceding ten years. The letter puts the ratio at 40.5 in the third quarter of 2026, below the 43.1 peak in the first quarter of 2000 but above the other historical technology-boom peaks it compares. This is a quarterly valuation measure, not the S&P 500's November 16 closing price. The regression uses the neutral real interest rate, potential output growth, the noncyclical unemployment rate and an index of macroeconomic uncertainty. The sample runs from 1961 through the third quarter of 2026, with each explanatory variable lagged one quarter. The full model accounts for 89% of CAPE variance within that sample; excluding the unemployment measure reduces the fit to 49%. Those are in-sample fit statistics, not probabilities that a forecast will be correct. For the third quarter, the full model produces a fitted CAPE of 35.8, which the authors describe as 11.6% below the actual 40.5. Without the noncyclical unemployment measure, the fitted ratio is 21.4, or 47.2% below the actual ratio. The percentages use the actual ratio as their reference. They should not be restated as the actual ratio being the same percentages above the fitted values, or as a predicted market decline. The authors suggest that hiring and stock prices may be linked because employers value future benefits from workers much as investors value future cash flows. Replacing the unemployment measure with a trend in labor's share of income explains 71% of the historical variance. That alternative fit points to model sensitivity, not proof that AI alone caused changes in wages, hiring or stock prices. The letter explicitly warns that historical relationships can change and that earlier technology booms also came with narratives used to justify high valuations. It does not issue an investment recommendation or a monetary-policy decision. The views are the authors' and do not necessarily reflect the Bank of Italy, the San Francisco Fed or the Federal Reserve System.

Sources

Related News