Sahm rule: the recession indicator built from unemployment
The Sahm rule compares the three-month average of the unemployment rate with its low of the last twelve months. Whenever the gap exceeded 0.5 percentage points since 1970, the US economy was in recession. Claudia Sahm proposed it in 2019 as a trigger for automatic fiscal relief.
History since 1959
What it says today
Unemployment is not rising. The labour market holds the economy up. 24 % of the history sits at or below today's value.
Unemployment is not rising. The labour market holds the economy up.
Unemployment creeps up. It means nothing yet.
The reading approaches half a point. Summer 2024, when the signal did not hold.
Threshold crossed. Since 1970 a recession followed every time except 2024.
Unemployment rises fast. The economy is in recession.
Where it stood at key moments
| Moment | Value | Band |
|---|---|---|
| Dot-com bubble peak (March 2000) | 0.00 pp | Calm |
| Post-bubble low (October 2002) | 0.57 pp | Recession signal |
| Pre-crisis peak (October 2007) | 0.20 pp | Calm |
| Financial crisis low (March 2009) | 3.13 pp | Recession under way |
| Covid crash low (March 2020) | 0.30 pp | Mild rise |
| Bear market low (October 2022) | 0.07 pp | Calm |
| Today | -0.07 pp | Calm |
What it measures
At the start of a recession unemployment does not jump, it accelerates. The Sahm rule captures that acceleration: the three-month average filters monthly noise and the comparison with the one-year low removes the level, so the rule works the same at 4 % unemployment and at 8 %.
The real-time version on FRED uses the data as first published, without later revisions. The 0.5-point threshold was crossed at the start of every recession since 1970, usually two to four months after its official start but years before the NBER committee dated it.
How to read it
The rule confirms, it does not lead. When it lights up, the recession is usually already running and stocks have mostly fallen. Its value is speed: official recession dating arrives a year late, the Sahm rule a month late.
Readings between 0.3 and 0.5 deserve attention, because historically the indicator rose above the threshold from there more often than it fell back to zero. In 2024 it did not, and the reading receded.
When it failed
In August 2024 the July reading reached 0.53 and set off recession headlines. No recession came. Sahm herself explained that unemployment rose because of an inflow of new workers, mostly immigrants, not because of layoffs. The rule measures unemployment, not its cause.
Before 1970 the rule gave false signals more often, which is why its record is counted from that year. The data are monthly and revised, so the real-time reading can differ from the later recomputed one by tenths of a point.
How to read market indicators
The seven indicators in the Indicators section say how expensive, frightened or tired the market as a whole is. What the bands mean, why none of them times the market, and how to combine them with the Fair Price Index.
Learn moreTerms in the glossary
Data is everywhere. Context is on Bulios.
In-depth company analyses, the fair value of thousands of stocks and StockBot AI. Bulios Black is the membership for investors who want to beat the market.
About Bulios BlackMethod and source
- Source: Bureau of Labor Statistics via FRED, series SAHMREALTIME (the real-time data version).
- Value = three-month average of the unemployment rate minus the low of that average over the last 12 months, in percentage points.
- Updated monthly, on the first Friday after the jobs report.
- Bands: under 0.2 calm, 0.2 to 0.35 mild rise, 0.35 to 0.5 near the threshold, 0.5 to 1 recession signal, above 1 recession under way.
Source: Claudia Sahm / BLS via FRED (Federal Reserve Bank of St. Louis) · Updated September 16, 2026
Frequently asked questions
Other indicators
The indicators describe the market as a whole from public data. They say where the market stands against its history, not what it does next month, and each of them has failed before. They are not investment advice. Learn more