The 3-month/10-year spread (3m10y) is the yield difference between the 10-year Treasury note and the 3-month Treasury bill. The Federal Reserve Bank of New York uses this specific spread as its primary recession probability indicator, and among all curve slope measures, it has the most rigorous academic backing as a leading macro signal.
Portfolio managers sometimes ask why the Fed anchors its recession model to the 3-month bill rather than the more widely quoted 2-year note. The answer lies in information content.
The 3-month T-bill yield prices almost nothing beyond the next FOMC meeting or two. It sits within 10-20 bps of the effective Fed funds rate in normal conditions, giving it a clean read on current monetary policy stance. The 2-year note, by contrast, already embeds roughly 24 months of expected rate moves. In an easing cycle, the 2-year falls ahead of the funds rate, compressing the 2s10s even before policy has actually shifted. That forward-loading makes the 2s10s a noisier real-time policy gauge.
The 3m10y therefore isolates a precise gap: where the Fed has set rates today versus where the long-run equilibrium is priced. When that gap goes deeply negative, the curve is signaling that current policy is too restrictive relative to long-term fair value. That is the compression that precedes credit tightening and recession.
Arturo Estrella and Frederic Mishkin published their model in 1996 using the 3m10y spread as the single predictor of U.S. recession with a 12-month forecast horizon. The model fits a probit regression of NBER recession incidence on the lagged spread level.
Key calibration points from their work:
The New York Fed publishes updated estimates monthly at its public website. At the May 2023 extreme of -189 bps, the model's implied probability was above 70%, the highest reading since the early 1980s tightening cycle.
The 3m10y has inverted before every U.S. recession since the 1960s, with no false signals in the post-war data on a sustained basis. The five most recent pre-recession inversions:
The typical lead time from initial inversion to recession onset runs 6 to 18 months, which is why the NY Fed frames its model on a 12-month horizon.
The 3m10y is available as a live chart at /spreads/3m10y on this site, with daily updates and historical context going back to the 1990s. Regime classification and slope comparisons across multiple tenor pairs are available in the slopes charting tool.
What does the current 3m10y spread level imply for recession probability?
Map the current spread directly to the Estrella-Mishkin calibration above. A spread at or below -100 bps puts the 12-month recession probability above 60% in the historical model. You can check the live spread at /spreads/3m10y and cross-reference the NY Fed's monthly recession probability series, which uses the same methodology. Note that the model is a statistical estimate from historical data, not a deterministic rule. The 2022-2023 inversion has tested that historical relationship, given that no NBER recession was declared through mid-2024 despite the extreme spread level.
How do I find the 3m10y spread on this site?
Go directly to /spreads/3m10y for the dedicated spread page, which shows the current level, the historical series, and regime context. The slopes tool lets you compare the 3m10y against other curve slope measures on the same chart. For deeper historical slope analysis and percentile rankings across multiple tenor pairs, use the slope charting tools from the main navigation.