Sahm Rule vs Initial Claims

The Sahm Rule is a recession-dating heuristic developed by Claudia Sahm in 2019 while she was a section chief at the Board of Governors. It triggers when the three-month moving average of the seasonally-adjusted U-3 unemployment rate (BLS series UNRATE) rises 0.50 percentage points or more above its trailing twelve-month minimum. The Federal Reserve Bank of St. Louis publishes a real-time series, SAHMREALTIME, on FRED that uses the contemporaneous data vintage rather than revised figures.

Initial jobless claims are administrative filings collected weekly by state labor agencies and aggregated by the Department of Labor. The Employment and Training Administration releases the seasonally-adjusted national figure every Thursday at 8:30am ET, covering the prior week (Sunday-Saturday). FRED carries the series as ICSA. The data is not survey-based, which is part of why claims are considered the cleanest weekly read on actual layoff dynamics: every filing is an administrative event, not a respondent's interpretation of their employment status.

The Sahm Rule has fired ten times since 1953. Seven coincided with the start of an NBER-dated recession within one quarter (1957, 1960, 1969, 1973, 1980, 1981, 1990). Two more (2001 and 2008) fired one to three months after NBER's eventual start date and confirmed recessions already underway. The 2020 COVID episode was an obvious special case where the unemployment rate jumped 11 percentage points in one month. The July 2024 cross at 0.53 is the first false positive in the 71-year sample.

Initial claims peaks during recessions: 6.6 million in April 2020 (COVID), 651,000 in March 2009 (GFC peak), 695,000 in October 1982, 580,000 in November 1990, and 517,000 in October 2001. Pre-recession warning thresholds typically sit at 350,000-450,000 sustained on the four-week average. The April 2026 reading of 210,750 on the four-week average is not just below warning levels, it is among the lowest absolute readings since the 1960s, although adjusting for labor-force size it is consistent with full employment rather than unprecedented strength.

The Sahm Rule crossed 0.50 in July 2024 and peaked at 0.57 in August 2024. Claudia Sahm published an October 2024 explanation framing the cross as a labor-supply event rather than a labor-demand event. The Congressional Budget Office estimated that net immigration rose from a five-year average of 110,000 per year through 2020 to 2.4 million in 2023 and 2.4 million in 2024. The CBO's labor-force model implied that the immigration-driven surge raised the U-3 unemployment rate by roughly 10 to 15 basis points by Q3 2024 mechanically, as new entrants typically search before they find work.

The initial claims cross-check would have flagged the false positive in real time. In every prior cycle where Sahm fired and a recession ensued, the insured unemployment rate (continuing claims as a percentage of covered employment) rose by an average 47 basis points year-over-year by the time Sahm crossed 0.50. In July 2024 the insured unemployment rate was unchanged from twelve months earlier and sat at 1.2%. The dual reading, Sahm rising while claims and the IUR were flat, was the institutional signature that eventually convinced the FOMC to ease rather than tighten in late 2024.

Claims and unemployment are derived from different parts of the labor-market measurement architecture. Claims count administrative filings the day a worker is separated and applies for benefits. The unemployment rate aggregates Current Population Survey responses collected during the reference week each month, asking whether the respondent was actively searching for work. Claims therefore precede unemployment-rate moves mechanically by two to six weeks during sharp deteriorations.

The 2008-2009 cycle ran the textbook sequence: claims rose from 320,000 in mid-2007 to 400,000 by April 2008 to 651,000 by March 2009. The Sahm Rule did not cross 0.50 until March 2008, three to four months after claims had decisively broken out of their normal range. The 2001 cycle showed the same pattern: claims rose from 290,000 in late 2000 to 460,000 in October 2001, with Sahm crossing 0.50 in May 2001. In both cases, a four-week claims average breaking 350,000 was the early signal that preceded the Sahm trigger by roughly six months.

The combined April 2026 picture, Sahm at 0.27 and claims four-week average at 210,750, is consistent with a labor market that has cooled substantially from 2022-2023 tightness without breaking. JOLTS job openings have fallen from a 12.2 million peak in March 2022 to roughly 7.6 million in February 2026, well above the 6.5 million pre-pandemic baseline but well below the 2022 peak. Nonfarm payroll growth has averaged 130,000 per month over the trailing six months versus 250,000 in 2023. Continuing claims at 1.78 million in April 2026 are also low.

The recession-indicator dashboard most cross-asset desks now run requires both Sahm rising toward 0.40 and claims four-week average breaking 250,000 sustained for four weeks before the signal is treated as actionable. The April 2026 reading is not close to either threshold. The Iran war that began February 2026 is the most plausible labor-market catalyst over the next 6-12 months given historical oil-shock-to-unemployment-rate lags of 2-6 months. The 1973 oil shock is the cleanest analog: unemployment rose from 4.6% in October 1973 to 9.0% in May 1975. No 2026 data has yet shown that pattern.

The cleanest framework is a layered confirmation rule. Tier 1 (yellow alert): claims four-week average rising above 230,000 sustained four weeks. Tier 2 (orange alert): claims breaking above 270,000 sustained four weeks AND Sahm Rule rising above 0.30. Tier 3 (high-confidence recession signal): claims above 300,000 four-week average AND Sahm Rule crossing 0.50. The 2024 false positive failed Tier 3 because claims never breached the cross-confirmation level, which is the design intent of the dual-indicator framework.

The historical hit rate of Tier 3 since 1967 is essentially perfect for predicting NBER recessions within one quarter, with the single exception of the 2020 COVID shock that compressed the typical six-to-twelve-month deterioration window into five weeks. The 2024 episode increased the institutional confidence in the dual-indicator framework precisely because the single-variable Sahm signal failed in a way the dual-signal framework would not have. April 2026 sits well below the Tier 1 threshold.