Do WARN notices really give 60 days' warning?

The claim: a WARN notice gives you 60 days' warning of a layoff. The federal rule requires 60 days' notice before a covered plant closing or mass layoff, so a notice filed on day zero should carry an effective date at least 60 days out. We measured that on the notices we serve at `/v1/alt/warn/notices`, and then checked whether, for listed companies, the notice arrives before the company's own 8-K restructuring disclosure (Item 2.05).

The first part mostly holds. The median lead is 60 days, and 63.0% of notices since 2021 are filed at least 60 days ahead. The second part does not: against 80 notices that overlap an Item 2.05 8-K, the WARN notice came first 32 times, on the same day 35 times and after the 8-K 13 times. WARN is early on the separations. It is not early on the company's disclosure.

Which states count

The feed holds 7,949 state WARN notices with a notice date from 2020-01-02 to 2026-11-02. 7,280 carry an effective date. Lead time is `effective_date - notice_date`, so it is only meaningful where those two fields are what they say they are. We read the stored `raw` payload for every state to check, and split the states in two.

• Used for the headline (4,747 notices): CA, FL, IL, OR, WA and WI. Each one's payload carries a separate field for when the state received or was notified of the notice and a separate field for the first layoff date.
• Excluded (2,533 notices with an effective date): TX, PA and NJ. New York has 568 notices and no effective date at all, so it cannot be measured.

Why the three are out. Texas has 2,203 rows with an effective date, and the median lead is 2 days. In 36.0% of them the layoff date is before the notice date, and only 21.7% are 60 days or more. The layoff dates look like real separation dates; what fails is the notice date, which appears to be a date on the state's side (the payload has a received date, and it matches the notice date in 995 of the 2,203 rows), not a date the employer gave notice. We cannot tell from the payload whether Texas employers file late or the state publishes late, so Texas is not evidence either way. Pennsylvania's payload has an effective date in free text and no separate notice field, so its notice date does not come from a labelled notice field, and 38.2% of its leads are negative. In New Jersey, the effective date equals the notice date on all 102 rows; the payload has an effective date and a posting month, and the two columns hold the same value. We have not changed the ingest. These are data-quality notes, not corrections to the numbers below.

An all-states figure would be 43.9% at 60 days or more, with a median of 56. That number mostly measures which states we parse well, and we do not use it.

The headline

Across the six verified states, 4,747 notices from 2020-01-02 onward:

• 56.1% give 60 days or more (2,662 notices). The median lead is 60 days.
• 17.0% give 30 to 59 days, 15.0% give 0 to 29 days, and 12.0% give a negative lead: the layoff date is already past on the day the notice is dated.

2020 distorts that. COVID closures were sudden, and many fell under the law's unforeseeable-business-circumstances exception, which allows short notice. In 2020 the verified states' median lead was 1 day and 43.4% of leads were negative, with 23.9% at 60 days or more. From 2021-01-01, 3,902 notices show 63.0% at 60 days or more (2,460), 17.7% at 30 to 59, 14.1% at 0 to 29 and 5.2% negative. Restricting to the four states with the cleanest fields (CA, FL, OR and WA, 3,090 notices) gives 63.0%. Dropping negative leads altogether leaves 63.7% (4,178 notices).

So the plain reading is that about two notices in three arrive with a full 60 days, one in six with 30 to 59, and about one in five with less than a month (14.1% at 0 to 29 days plus 5.2% negative). The law says 60 and the data is close to it, but a fifth of notices give under 30 days. A median of 60 reflects the rule being followed on the typical filing; it says little about the tail you care about.

By state, closure type, size and mapping

Since 2021-01-01, share of notices with a lead of 60 days or more:

• OR 68.4% (158 notices), CA 66.6% (2,005), WA 66.7% (111), IL 59.8% (463), WI 59.4% (426), FL 55.8% (739). Medians are 60 or 61 days in every one of them.
• Closures 71.1% unmapped and 86.2% mapped to a listed company; layoffs 62.9% unmapped and 63.0% mapped. Employers shutting a site give more warning than employers cutting staff, and for listed closures the median lead is 63 days (145 notices).
• Notices with no closure type (a field several states do not report) sit lowest: 52.6% unmapped, 60.5% mapped.

Size over the whole verified window (all years): 54.8% for notices under 50 workers (2,414), 58.1% for 50 to 99 (1,089), 58.7% for 100 to 249 (819) and 53.5% for 250 or more (282). Larger notices are not earlier. The 250-plus group has the highest share of negative leads, 24.1%, much of it 2020. Many of the smallest notices are below the federal coverage threshold, filed under state rules, so the 60-day expectation applies less cleanly to them.

Mapped companies, meaning notices we could tie to a `symbol_id`, are 62.0% at 60 days or more against 54.4% for unmapped ones (1,035 and 3,712 notices, 170 symbols), all years. Since 2021 the gap sits in closures (86.2% against 71.1%) and in notices with no closure type; for layoffs mapped and unmapped are level at 63.0% and 62.9%.

Does WARN lead the 8-K?

Item 2.05 of Form 8-K is the disclosure of costs from exit or disposal activities, which is where a restructuring usually shows up. We matched WARN notices to 2.05 filings on `symbol_id`, with the 8-K filed within 90 days of the notice either side. That produces 95 notice-filing pairs: 80 notices, 21 symbols and 26 distinct 2.05 filings. Of the 80 notices, 73 are from the verified states (59 from California), and the notice dates run from 2020-06-30 to 2026-10-23. This is a case set. It is not a rate to extrapolate from.

For each notice we took the nearest 2.05 filing, and counted `filed_date - notice_date`:

• The WARN notice came first in 32 cases, on the same day in 35, and after the 8-K in 13.
• 38 of 80 fall within seven days either way. The median gap is 0 days, the quartiles are 0 and 15 days, and the range is -62 to 85.
• Counting each filing once, against the earliest notice near it: 26 filings, with the notice first for 11, the same day for 10 and the 8-K first for 5.
• By size, notices of 100 or more employees preceded the 8-K 10 times in 22; smaller or unknown ones did so 22 times in 58.

A WARN notice cannot be said to lead the 8-K from this. The more common pattern is that company and state paperwork land together, which is what you would expect of a company announcing a restructuring. A strategy that waits for a notice to front-run Item 2.05 would be early in a minority of cases; the 75th percentile gap is 15 days. What the notice adds is the site, headcount and separation date: the operational detail that an 8-K usually leaves out.

Querying it

Notices are public. Filter by state, ticker, headcount or notice date, and read the lead from the two date fields:

# Large notices since the start of 2026, California
curl -H "Authorization: Bearer $HEDGEFRIEND_KEY" \
  "https://api.hedgefriend.dev/v1/alt/warn/notices?state=CA&min_employees=100&from=2026-01-01&limit=2"

# One ticker's notices
curl -H "Authorization: Bearer $HEDGEFRIEND_KEY" \
  "https://api.hedgefriend.dev/v1/alt/warn/notices/META?limit=10"

{
  "data": [
    {
      "state": "CA",
      "company": "Example Corp",
      "ticker": "EXMP",
      "notice_date": "2026-10-26",
      "effective_date": "2026-12-31",
      "employees_affected": 181,
      "closure_type": "closure"
    }
  ]
}

Lead days for a row are `effective_date` minus `notice_date`. In the example that is 66. Filter to the states above before you use the number. The queries behind every figure here are in the research folder for this post.

Caveats

• Notice date is when the state recorded the filing, not when workers were told. Employees can learn earlier, and a state posting date can lag, so a measured lead can be shorter than the true one, and a negative lead can come from either late filing or late publishing.
• Effective date is the first layoff date. For rolling layoffs it understates how long the process runs, and a notice can be amended after filing; we store the version the state published.
• We verified the six states by reading their payload fields. We did not verify that every employer filed on the date written, and exceptions to the 60 days (unforeseeable business circumstances, faltering company, and state-specific rules) are not flagged in the data.
• The states differ in coverage and start dates: California begins in 2025-01-29, Washington in 2026-03-09 and Florida in 2024-01-03. Recent periods are dominated by California and Florida, so the headline leans on them.
• Excluded states are excluded for a field problem, and some may be good data on a different footing; Texas in particular is the largest state by count. The headline is for six states, not for the United States.
• Only 1,035 of the verified-state notices map to a listed company. The overlap with 8-K 2.05 is 80 notices and 21 symbols, so any single direction here could move with a handful of cases.
• We did not look at price or returns. The daily price table covers only four days today, so a market-reaction test is not possible yet.