US Recession Risk Composite
Five signals, each classified green / red against a trigger threshold. Composite score is the count of triggered signals. NBER recession periods are shaded on the historical timeline and component sparklines.
Composite History
Count of triggered signals over time. Orange = 2 signals (elevated), red = 3+ (high). Grey bars mark NBER-dated recessions for calibration. The composite is a leading indicator — it can trip before a recession and stay tripped during one.
Methodology
Each signal is binary: triggered or not. Thresholds are drawn from professional consensus (cited per signal), not tuned to fit history. The composite score is the simple count — no weighting, no smoothing. This is deliberately conservative: a noisy single signal doesn't move the score, but multiple signals agreeing does.
- Sahm Rule (Sahm, 2019, Brookings)
- 3-month moving average of the unemployment rate minus its 12-month prior minimum. Triggers at +0.50 pp. FRED publishes this as
SAHMCURRENT, so we use that directly rather than recomputing. - Yield Curve (10Y − 3M) (Estrella & Mishkin, Fed NY)
- Every US recession since 1960 was preceded by an inversion of this spread. Triggers when the spread goes negative. Typical lead time is 12–18 months.
- High-Yield OAS (ICE BofA BAMLH0A0HYM2)
- Option-adjusted spread on US high-yield corporate debt. Captures credit-market stress. Triggers at ≥6.00%. This threshold approximates the long-run median plus ~1σ; historically, sustained readings above this level coincide with or precede recessions.
- Unemployment 6-Month Change
- Current UNRATE minus UNRATE 6 months ago. Triggers at ≥ +0.5 pp. Used as a confirmation signal for Sahm — if the labor market is actually cooling (not just a one-month print), this should move too.
- Payrolls 3-Month Average
- Average monthly change in nonfarm payrolls over the last 3 months. Triggers below 50,000/month. Below-trend hiring (trend is ~150k/mo in a healthy economy) suggests demand for labor is fading.
What this is not: a recession forecast. It's a scoreboard of widely-watched signals. When multiple trip simultaneously, the historical record says recession probability rises — but the composite is not calibrated as a probability model. Read it as "how many canaries are coughing," not "P(recession) = X%."