APProjects/us-mass-layoff-and-plant-closing-statistics-monthly

US Mass Layoff & Plant Closing Statistics — monthly series, rebuilt every day

The federal government stopped publishing this. The Bureau of Labor Statistics ran a Mass Layoff Statistics program until sequestration cut it: the final release was 2013-06-21, covering May 2013 and the program was eliminated on 2013-09-30. Since then there has been no monthly, national, machine-readable count of US mass-layoff and plant-closing events.

This series is one answer to that gap, built from state WARN Act filings and rebuilt from the source portals every single day.

2,186 events	mass layoffs & plant closings, 2025-08 → 2026-07
356,674 workers	affected in those same 12 months
455 months	every month covered, 1988-11 → 2026-09
36,587 events	in the full history
2026-09-16	last rebuilt from live state portals

Events per month, last 24 complete months (2024-08 → 2026-07):

▁▂▃▃▁▄▅▆▅▄▄▄▄▅█▂▁▇▆▅▇▄▃▂
2024-08          2026-07
low 112  high 254 events/month

What one row is

An event is one WARN Act notice affecting 50 or more workers at one employer site. That threshold is the WARN Act's own mass-layoff floor, so the series counts the layoffs large enough that federal law required warning.

data/monthly_national.csv — one row per month:

month	events	employees_affected	closure_events	layoff_events
2026-07	142	22,737	40	46
2026-06	160	20,583	45	49
2026-05	177	47,263	51	53
2026-04	221	33,510	69	52
2026-03	186	23,786	67	47
2026-02	210	28,850	69	50
2026-01	221	35,448	86	57
2025-12	112	14,558	39	31
2025-11	138	25,432	38	42
2025-10	254	47,330	71	97
2025-09	198	32,746	57	59
2025-08	167	24,431	52	64
2025-07	181	27,641	59	52

data/monthly_by_state.csv — the same, split by state (48 states currently scraped).

Both files also carry unclassified_events and is_partial. Read the next section before you use either.

Read this before you cite it

1. This is not the BLS series and cannot be spliced onto it. BLS MLS counted unemployment-insurance claim events (50+ initial claims against one establishment within 5 weeks). This counts WARN notices with 50+ affected workers. Different universe, different trigger, different lag. Use this as a successor in purpose, never as a continuation of the BLS numbers.

2. The last 2 months are incomplete and flagged is_partial=true. State portals publish in batches, so recent months fill in for weeks after the fact. Drop them or label them — do not read a trend off the final bar.

3. The closure/layoff split covers part of the file, not all of it. Of 36,587 events, 10,176 are clearly plant closings and 11,342 are clearly layoffs; 15,069 carry a notice type we refuse to guess at. The upstream field is free text from 48 separate agencies — values in the raw data include state, warn, cl, 2 and blank. We classify only unambiguous text and report the rest as unclassified, because a confidently wrong split is worse than an admitted gap. events and employees_affected are unaffected by this and are the columns to trust.

4. WARN is not all layoffs. It binds employers over a size threshold, exempts several situations, and enforcement varies by state. This measures notified mass layoffs — a floor, not a total. Coverage is 48 states, not 50.

5. The COVID shock will dominate any model you fit. 2020-04 alone is 2,836 events and 529,360 workers.

6. The early years are thin, and that is an archive artefact, not history. Every one of the 455 months carries at least one event, but only 1,606 of the 36,587 total fall before 2000 — most state portals simply do not publish that far back, and the ones that do vary in depth. Treat pre-2000 as a partial archive; the series is dense from roughly the mid-2000s on.

Quickstart

from datasets import load_dataset

nat = load_dataset("APProjects/us-mass-layoff-and-plant-closing-statistics-monthly", "national", split="train")
st  = load_dataset("APProjects/us-mass-layoff-and-plant-closing-statistics-monthly", "by_state", split="train")

import pandas as pd
url = "https://huggingface.co/datasets/APProjects/us-mass-layoff-and-plant-closing-statistics-monthly/resolve/main/data/monthly_national.csv"
df = pd.read_csv(url, parse_dates=["month"])
df = df[~df.is_partial]                 # drop the months states are still filing
df.set_index("month").events.plot()

How it is built

1. A scheduled pipeline re-scrapes 48 state labor-department WARN portals daily — not once, daily, because states amend headcounts, re-issue notices and silently drop rows.
2. Rows are normalized to one schema, deduplicated, and employer names are canonicalized. That output is published free as APProjects/us-warn-act-layoffs-notices-daily — every notice, no threshold, no delay.
3. This series is a deterministic aggregation of that file: filter to employees_affected >= 50, key on the WARN filing date, group by month.

Because step 1 runs every day, the numbers here move. A static copy of this table starts rotting the week it is downloaded — that is the whole point of publishing it as a live dataset rather than a paper.

Licence, provenance, contact

CC BY 4.0. Underlying WARN notices are US state public records. Attribution: US Mass Layoff & Plant Closing Statistics, WARN Feed, 2026-09-16.

Corrections and questions: open a discussion on this dataset — they are read and answered.

This dataset is produced by an automated pipeline. Methodology, per-state coverage, source portal links and the raw notice-level data are at https://approjects-warn-act-notices.static.hf.space.

---

Related

• APProjects/us-warn-act-layoffs-notices-daily — the notice-level source data, free, 61,330 rows, no threshold.
• Per-employer and per-state alerts — if you track specific employers or states rather than the aggregate, WARN Watch runs your list against each daily refresh and gives you a private alert page and RSS feed. Free for 30 days, no card.
• Coverage and methodology — which portals, scraped when, and what is missing.

A layoff record you can audit, not just download

This dataset is one cut of a single daily rebuild: 61,330 US WARN Act layoff notices from 48 state agencies, 1988 to today, one schema, no login, no delay, CC BY 4.0. Snapshot as of 2026-09-16; the files above are rebuilt every day, so the live count is the truth.

Several projects publish a current WARN scrape and two of them carry more rows than we do. None of them publish what the records used to say:

• 617 observed changes to already-published notices, logged daily since 2026-08-31. data/revisions.csv records every field that differed between two consecutive daily builds — employee counts, effective dates, notice types, company names — with the old value, the new value and the date we saw it. We publish the observation and not the cause: a change is equally explained by the agency amending the notice or by our own parser improving, and we do not guess which (see data/revisions.README.txt). A scrape that starts tomorrow cannot backfill any of it; it only exists if someone was watching.
• 6,799 notices whose state agency page no longer lists them. Agencies take notices down. We keep them, flagged as archive-only, so a count you ran last year still reconciles.
• Point-in-time employer identity. The ticker crosswalk resolves a filer to the company as it existed at the time of the notice — Kmart, Sears Holdings, Symantec — not to whatever is on today's ticker file.

If you have to defend a number to an editor, a referee or a compliance reviewer, that provenance layer is the part you cannot rebuild yourself. How to cite this dataset →

Look something up right now — free, no signup, nothing to install. Check any employer or state against the last 180 days → It runs in your browser against these same files.

Building something with it? The same files are a free HTTP API — JSON and CSV, no key, no signup, access-control-allow-origin: * so fetch() works from a browser: endpoints, schema and curl examples →

Need one industry only? The same filings, cut by an auditable employer-name rule (each row keeps the rule that fired): tech companies · hospitals & healthcare · retail store closings · restaurants & hotels · factory & plant closings · banks, insurance & finance · warehouses, trucking & logistics · all 20 sectors.

Or have it watch a list for you. Coming back to look is the part a CSV cannot do. WARN Watch — $49 for a year, one payment, nothing auto-renews, 14-day refund, no login: up to 500 employer names plus whole states, matched on every daily refresh, delivered to a private alert page + calendar (.ics) + RSS + an optional Slack / Discord / Teams webhook. Every alert carries that employer's whole filing history from the archive, which a keyword rule on an RSS feed cannot see. There is no built-in email — we do not claim one.

• See a real alert page before paying · what you get
• Try it free for 30 days, no card · Buy — $49/year

Not deciding today? Join the update list → — one email when a new dataset or tier is published; nothing promotional. A state added or a column renamed ships in the daily release instead, no address needed. The list is shared across APProjects datasets, holds an email address only, is run by Gumroad, and any message unsubscribes you. Rather give no address at all? Watch the repo's releases — GitHub notifies you on every daily republish, and a new state or changed field is in those notes the day it lands.

Reaching a human. WARN Feed is published by APProjects, an automated data publisher — that is stated plainly rather than dressed up. Corrections, coverage gaps, schema questions and refund requests all go here and are read: open an issue. Payments are handled by Gumroad as merchant of record, so an invoice can carry your company name.

Source, scrapers and methodology · the 48-state site

_{Card views are counted anonymously: one 1×1 image on a public CDN, no cookies, no script, no personal data. The count is public: jsDelivr stats.}