Traders' core economic releases (jobs, CPI, PPI, GDP, retail sales) run 2-8 weeks late in a long shutdown — BLS was untouched in 2018-19 but hit hardest in 2025

Summary

Government shutdowns delay the releases that move markets most: the BLS jobs report (Employment Situation / nonfarm payrolls), CPI, PPI, and — via the Commerce Department — GDP, retail sales, durable goods orders, housing starts, and the trade balance. How long depends on which agencies are funded and how long the lapse lasts. In the 1995-96 shutdown (21 days) the jobs report was about 2 weeks late and CPI 2-3 weeks late. In 2013 (16 days) the jobs report was about 18 days late, plus retail sales, PPI, and trade data. In 2018-19 (35 days) BLS itself was unaffected because it had already been funded for the year — jobs and CPI came out on schedule — but Commerce Department releases (GDP advance estimate, retail sales, durable goods, housing starts, trade balance) were delayed 3-5 weeks. In the record 43-day 2025 shutdown, essentially everything was hit: the September jobs report (due Oct 3) wasn't released until Nov 20 — 48 days late — the October jobs report was never issued as a standalone release at all (folded into the December 16 report), and the Q3 GDP advance estimate (due Oct 30) didn't appear until Dec 23, 54 days late, with one of the normal three GDP estimates skipped entirely. Data collected before the shutdown (like September CPI) could still be processed and released close to schedule; data whose collection window fell inside the shutdown was pushed back the furthest, and in a few cases permanently thinned or merged into a later report.

Which releases traders actually watch — and how a shutdown hits each

Historical shutdowns and measured delays

What the AskAmerica warehouse itself shows

Beyond the news coverage, the corpus's own BLS revision-vintage table (econ.ces_revision_vintages, which stores each monthly Employment Situation release exactly as BLS first published it) directly shows the 2025 blackout: in the release stamped October 2025, the September and October 2025 reference-month values for total nonfarm payrolls (series CES0000000001) are both -1 — BLS's own 'not yet available' sentinel, preserved verbatim rather than estimated. Those same cells only receive real values in the November 2025 release (September = 159,593 thousand; October = 159,488 thousand), consistent with the reporting above that September's report was withheld until Nov. 20 and October's was folded into the December 16 release rather than issued on its own. This is a direct, machine-checkable confirmation of the shutdown's data gap, not just a claim from secondary reporting.

SQL used: SELECT series_id, release_year, release_month, reference_year, reference_month, value FROM econ.ces_revision_vintages WHERE release_year = 2025 AND release_month IN (9,10,11) AND series_id = 'CES0000000001' ORDER BY release_month, reference_year, reference_month

Caveats

• Delay-day counts for 1995-96 and 2013 come from secondary reporting (BLS's own historical-release-dates PDF and news coverage at the time), not a re-derivation from a machine-readable BLS vintage archive going back that far — the AskAmerica ces_revision_vintages table's own reference-month window only reaches back about 6 years per release, so it could directly corroborate 2025 but not 1995-96/2013.
• The exact 2013 CPI delay date was not independently confirmed beyond secondary summaries in this pass; treat that gap in the table as a genuine unconfirmed data point rather than an implied figure.
• Whether October 2025 data (jobs, CPI, PPI) is ever published as its own standalone reference month, versus permanently absorbed into later releases, was still being finalized by BLS/BEA/Census as of the sources checked here — some 2025-26 releases explicitly describe October as a partial, permanent gap in the record.

Every query behind this report

1 warehouse call ran in this session, in order. Each is reproducible against the same snapshot.

SELECT series_id, release_year, release_month, reference_year, reference_month, value
FROM econ.ces_revision_vintages
WHERE release_year = 2025 AND release_month IN (9,10,11)
  AND series_id = 'CES0000000001'
ORDER BY release_month, reference_year, reference_month