How the rule is calculated
Take the unemployment rate's three-month moving average. Subtract the minimum of that same average over the prior twelve months. When the difference reaches 0.50 percentage points, the rule triggers.
The three-month averaging filters out single-month noise in a survey-based statistic; the twelve-month lookback anchors the comparison to the recent cycle rather than a distant baseline. The live value of this calculation is charted on the Sahm rule page linked below.
The historical record
Applied retroactively, the rule triggered during every recession since 1970, typically two to four months after the recession's official start — far earlier than the NBER's dating committee, which often declares recessions a year later. False positives have been rare; the 1959 and 1969 borderline cases predate its design window.
That record is why the rule became a fixture of recession dashboards. It is worth remembering what it is for: Sahm designed it to trigger stimulus payments quickly, prioritizing few false alarms over the earliest possible warning.
Where the rule can mislead
The rule reads rising unemployment as collapsing labor demand. But the unemployment rate can also rise because more people enter the labor force to look for work — the denominator grows, and the rate ticks up without layoffs accelerating. Sahm herself flagged this ambiguity during 2024's immigration-driven labor-force surge, when the rule briefly triggered outside a recession.
It is also a coincident-to-lagging signal by design: by the time it fires, the recession has typically already begun. It answers 'has a recession likely started?' — not 'is one coming?'. Forward-looking indicators like the yield curve and jobless claims, linked below, address the second question.