A recession-probability model that maps the slope of the Treasury yield curve — typically the 10-year minus 3-month or 10-year minus 2-year spread — onto the likelihood of a downturn over the next twelve months, usually via a probit regression. An inverted spread signals elevated recession risk.
How it works
A probit (or logit) regression estimates recession probability as a function of one variable: the term spread, most commonly the 10-year minus 3-month Treasury yield. The New York Fed's canonical version uses monthly data since 1959. A negative spread (inversion) historically precedes recessions by roughly 12-18 months, which the model converts into a one-year-ahead probability.
Why it matters now
After the deepest and longest curve inversion in decades through 2023-24, term-spread models flagged sharply elevated recession odds that did not materialise on schedule — fuelling 2025-2026 debate over whether term-premium compression and QT-era distortions have degraded the spread's forecasting power.
Example
The New York Fed's term-spread model, built on the 10-year minus 3-month spread, has at points in the cycle put the one-year-ahead recession probability above 50 percent — for instance around 51 percent on a simple specification — even as realised growth held up, illustrating the model's recent calibration tension.
Frequently asked
- What is a term-spread model?
- A term-spread model is a recession-probability model that maps the slope of the Treasury yield curve onto the odds of a downturn over the next twelve months. It typically uses the 10-year minus 3-month spread
Glossary · 10y-3m spread
The 10y-3m spread is the difference between the 10-year Treasury yield and the 3-month Treasury bill rate, a yield-curve gauge the New York Fed favours as a recession predictor. A negative reading (inversion) has preceded every US recession since the late 1960s.
Read more →
- Which yield-curve spread does the model use?
- Most term-spread models use the 10-year minus 3-month Treasury spread, which the New York Fed's canonical specification favours for its forecasting record. Some variants use the 10-year minus 2-year spread, popular in markets, but research generally finds the 10y−3m more reliable because the short end better captures expected near-term policy.
- Why did the term-spread model fail to predict a recession in 2023-24?
- The deep 2022-24 inversion pushed term-spread models to recession probabilities above 50 percent, yet no downturn arrived on schedule. Analysts attribute the miss to compressed term premium
Glossary · term premium
The term premium is the extra yield investors demand to hold a long-dated bond instead of rolling short-dated instruments over the same horizon. It compensates for the risk that future short rates, inflation, or fiscal supply diverge from expectations, and is the residual in nominal yields once the expected-rates path is stripped out.
Read more →
- How does the term-spread model differ from term premium?
- The term-spread model is a forecasting tool; term premium is one of the forces that drives its input. The spread reflects both expected future short rates and the term premium — the extra yield demanded for holding longer maturities. When term premium compresses, the curve can invert without signalling recession, which is precisely the distortion blamed for recent model misfires.
- How far ahead does an inverted yield curve signal a recession?
- An inverted term spread historically precedes recessions by roughly 12 to 18 months, which the model converts into a one-year-ahead probability. Because of this long and variable lead, an inversion does not pinpoint timing — the 2019 inversion preceded the 2020 contraction, while the 2022 inversion produced no on-schedule downturn.