What are the Fed model and its limitations for valuing stocks?

The short answer

The Fed model arose from a comparison some economists noticed in the 1980s and 1990s: when the S&P 500 earnings yield (E/P, the inverse of the P/E ratio) was high relative to the 10-year Treasury yield, equity returns subsequently looked good; when it was low, equities struggled. The framework appeared in a 1997 Federal Reserve Humphrey-Hawkins report — hence the popular name — although the Fed itself never officially endorsed it.

Its appeal is simplicity. By a single comparison of two numbers, investors could allegedly tell whether stocks were fairly valued relative to bonds. Pre-2000, the framework appeared to work reasonably well: rising bond yields suggested stocks needed higher earnings yields, and vice versa.

The problem is theoretical. Earnings yields are real (inflation-protected over the long run because corporate earnings grow with nominal GDP), while nominal bond yields incorporate inflation expectations. Comparing them mixes real and nominal quantities — a basic financial error. Asness (2003) documented that the model implicitly assumes the inflation illusion: investors mistakenly use nominal yields to discount real cash flows.

→ New to valuation models? Investing for beginners hub

What the data shows

FRED data on S&P 500 earnings yield and 10-year Treasury yields, combined with Asness (2003) and subsequent academic studies, document the model’s empirical limitations:

• The Fed model “spread” (E/P minus 10-year yield) ranged from -3% in 2000 to +5% in 2009 to roughly 0% in late 2024
• Asness (2003) showed that pre-1965 data — when inflation was low — produced very different model conclusions than post-1965 data
• Long-run regressions of subsequent equity returns on the Fed-model spread show R² values below 0.10, well below alternative valuation metrics like CAPE
• The 2000 reading suggested equities were not extremely overvalued, despite CAPE of 44 and subsequent 50% drawdown
• The 2009 reading flagged extreme cheapness, which proved correct, but the model also flagged similar conditions in 2011-2012 with much weaker realized returns

The exception worth noting: the Fed model can have explanatory value within a single inflation regime. From 1980 to 2000, when inflation expectations were stable and falling, the model’s correlation with subsequent returns was higher. Across regime changes — disinflation to inflation, and back — its predictive power collapses.

→ Dataset: 10-year Treasury yield

Why it happens — the macro mechanism

Three theoretical issues undermine the Fed model’s reliability.

Inflation illusion problem. Modigliani and Cohn (1979) first identified the inflation illusion: investors irrationally use nominal yields to discount real cash flows. The Fed model embodies this error directly. In high-inflation regimes, the model says equities are cheap when bond yields are high — but corporate earnings have already adjusted upward with inflation, so the comparison is misleading. Real vs nominal yields details this distinction.

Mismatched risk premia. The Fed model implicitly assumes equity risk premium is constant or absent — that the only thing distinguishing stocks from bonds is the earnings yield versus bond yield gap. But ERP is a major variable that compresses or expands with regime. Damodaran (2024) and others document ERP shifts of 200-400 bp across cycles, dwarfing the Fed model’s signal magnitude. Equity valuation places ERP in context.

Forward versus current earnings. The model uses trailing or current-year earnings to compute E/P. But equity values reflect expected future cash flows over decades. Using a single year’s E/P captures only a slice of the valuation reality. CAPE (Shiller’s 10-year averaged earnings yield) addresses this partially. Forward implied ERP models do so more comprehensively.

Synthesis by regime: the Fed model can appear to “work” within stable inflation regimes but fails across regime transitions. As a long-run valuation framework, it is dominated by alternatives like CAPE and implied ERP that account for inflation, real rates and forward earnings.

The Fed model is the right idea — comparing equity to bond yields — built on the wrong foundation: nominal versus real arithmetic.

→ Framework: Equity markets pillar

What it means for different economic actors

Savers rarely use the Fed model directly but encounter its conclusions in financial media that frequently cites the framework. Understanding its limitations helps avoid making allocation decisions on a misleading signal.

Investors who use the Fed model should be aware of its theoretical issues and complement it with inflation-aware metrics — CAPE, equity yield gap based on real yields, or implied ERP. Asness, Friedman, Krail and Liew (2000) propose alternative formulations that adjust for inflation explicitly.

Pension funds and institutional managers rarely rely on the Fed model for strategic asset allocation. Long-horizon valuation frameworks used by major institutions typically combine multiple metrics — CAPE, normalized earnings, and forward implied ERP — rather than the simple yield-spread comparison.

A common error is interpreting the Fed model spread as a definitive timing signal. The 2000 reading was nowhere near as extreme as alternative metrics suggested, and following the model would have produced inadequate caution before the 50%+ drawdown that followed. The model’s apparent simplicity hides a meaningful theoretical flaw.

Practical observation

Go deeper

Related questions

• What does the CAPE ratio tell us?
• What is the equity risk premium?
• Why real yields matter more than nominal
• How the 10-year Treasury yield affects the economy

Frequently asked questions

Why is the Fed model so popular if it’s flawed?

The model’s popularity reflects its simplicity rather than its rigor. A single chart of E/P versus 10-year yields is easy to produce, intuitive to interpret, and convenient for media commentary. Sophisticated alternatives like CAPE require longer time series, normalized earnings, and explanation of why current earnings differ from cyclical averages. The Fed model fits a 30-second TV segment in ways that academic frameworks cannot. Its persistence is a useful reminder that simple metrics often outweigh better-grounded ones in market commentary.

Did the Fed model ever officially come from the Fed?

No. The framework is named after a 1997 Federal Reserve Humphrey-Hawkins report that contained a chart of E/P versus 10-year yields, but the Fed never officially endorsed the model as a valuation tool. Strategists at investment banks subsequently popularized the name, and it stuck. The Fed itself has not used this framework in any policy-making capacity, and several Fed economists have explicitly criticized it in research papers.

What should replace the Fed model?

Several alternatives address the inflation illusion problem. The Asness-corrected version uses real yields (TIPS) instead of nominal Treasury yields, restoring the like-for-like comparison. Shiller’s CAPE provides a 10-year normalized earnings yield that captures cycle smoothing. Damodaran’s implied ERP solves current valuations for the discount rate that justifies them, providing a direct readout of the equity premium. Combining these inputs produces more robust signals than any single metric, including the Fed model.

Last updated — 5 May 2026