NY Fed recession probability
The New York Fed publishes a monthly recession-probability estimate derived entirely from the 10Y–3M Treasury spread, using a probit model estimated on decades of cycle history. When the curve inverts deeply, the probability soars — it peaked near 71% in 2023–24, the highest reading since the deep inversions of the early 1980s.
One signal, done well
The model's strength is its simplicity: one input, long history, fully transparent methodology. Its weakness is the same: it carries the yield curve's blind spots — the term-premium distortions, the 2023–24 false alarm — with no diversification across the credit, housing, and labor channels that also precede recessions.
In the panel
We include it as a reference signal rather than building on it — it anchors the 'what does a single-signal model say' answer and keeps the composite honest: when our panel agrees with the curve probability, confidence is higher; when they diverge, that's information too.