MissingPeriod ending 30 Sep 2026Available 03 Oct 2026 · 00:33 UTC
Sahm-style unemployment warning indicator
History
7 readings.
Showing periods 1–7 of 7 in this date range. Each chart page contains at most 180 original periods; values are not downsampled. All retained snapshot-selected periods are available.
Uses the data available in this saved assessment; later revisions may change historical values. This reconstructed chart does not show what was known on each historical date. Missing source periods break the line; unavailable metric values are omitted. Quality is evaluated at the assessment date.
Exact values · final 12 periods in this chart window
Exact plotted metric readings
Economic period
Value
Unit
Quality
31 Mar 2025
0.3
unit unavailable
Valid
30 Apr 2025
0.3
unit unavailable
Valid
31 May 2025
0.33
unit unavailable
Valid
30 Jun 2025
0.23
unit unavailable
Valid
31 Jul 2025
0.17
unit unavailable
Valid
31 Aug 2025
0.13
unit unavailable
Valid
30 Sep 2025
0.23
unit unavailable
Valid
Method, sources and timing
Economic period & freshness
Calculation uses the latest completed observation available at evaluation time, with its actual economic period and original availability. An older period contributes to assessment only within the configured source-age limit. Newly observed missing or invalid evidence is not replaced with an earlier value. Source status uses economic-age thresholds, not a release calendar; no future release time is inferred.
Source timing diagnostics
Input
Latest source period
Observation age
Available to ECO I
Source status
Unemployment rate
30 Sep 2026
27 days
03 Oct 2026 · 00:33 UTC
Current
THE PUBLISHED TRANSFORMMA3_t - min(MA3_t-1, MA3_t-2, ..., MA3_t-12), where MA3_t = (U_t + U_t-1 + U_t-2) / 3; all fifteen calendar-month observations must be present.
ECO I computes this Sahm-style indicator from the current BLS unemployment vintage; it is not the published real-time Sahm series. Require fifteen consecutive nonmissing calendar-month observations: the current three-month average and twelve preceding three-month averages. Do not skip missing windows, interpolate, or floor negative results at zero. October 2025 CPS data were not collected; any required window containing that month remains unavailable. The published unemployment level and other valid labor evidence remain usable. A rise of 0.50 percentage points is the established Sahm warning threshold, not proof of an NBER recession or a calibrated event probability; false alarms are possible. Population controls and seasonal revisions affect comparability. All required input vintages must be available by the assessment time. This shares the CPS input with unemployment context and cannot count as independent labor evidence or contribute to the legacy ordinal score.
Credit: U.S. Bureau of Labor Statistics. Retrieval dates accompany the observations. BLS.gov cannot vouch for the data or analyses derived from these data after the data have been retrieved from BLS.gov. File hashes identify the exact retrieved bytes; these are not historical point-in-time vintages.
Chart source observations · 33 periods in this window
Showing periods 1–33 of 33 in this date range. Each chart page contains at most 180 original periods; values are not downsampled. All retained snapshot-selected periods are available.
Uses snapshot-selected vintages. The final eight exact values in this window are tabulated above. The pinned JSON query provides all matching observations and provenance; follow its cursor for further pages.
Source: U.S. Bureau of Labor Statistics. BLS.gov cannot vouch for the data or analyses derived from these data after the data have been retrieved from BLS.gov. Data access dates are supplied as retrieved_at in canonical observations. Latest retained BLS access: 03 Oct 2026 · 00:33 UTC. BLS source terms
Source: BIS debt service, credit and credit-gap statistics; debt service is model-based. ECO I's calculations and interpretation are its own; BIS does not endorse them. BIS statistical-use terms