WARN Notices by State: An Acquisition Research Matrix

Executive Summary

WARN notices by state are a research trigger for a specific workforce event, not proof that a company is for sale. The Worker Adjustment and Retraining Notification (WARN) framework concerns advance notice of qualifying closings and layoffs; federal guidance generally specifies 60 calendar days for covered events. This report's recommended acquisition-research response is to recheck the affected facility, investigate an entity match, or leave the signal unresolved. Sale intent and financial condition require separate evidence. ^{[1] [2]}

The report supplies a 50-state official-source matrix, checked for publication on October 8, 2026, with agencies, exposed coverage, formats, notice-document availability, state-rule routes, and explicit access caveats. The original access audit identifies 9 states with specifically documented comma-separated values (CSV), Excel, spreadsheet, or table-export routes. Those include Alabama's CSV route, California's latest workbook, Colorado's spreadsheet, and Virginia's filtered CSV interface. This counts available routes, not successfully downloaded files or comprehensive state coverage. ^{[3] [4] [5] [6]}

History and record type require careful handling. Texas exposes annual Excel files for 2020 through 2026 with an older-record contact. New York directs notices and amendments from January 1 through April 1, 2025 to a retired database. Oregon states a six-year retention policy. Wisconsin warns that received notices are not a complete list of potential or actual layoffs, while its revision guidance says the latest revision reflects the reported affected-worker total. Adding versions together can therefore misstate an event. ^{[7] [8] [9] [10]}

Thresholds also differ. Iowa's agency summary describes employer coverage at 25 employees with 30 calendar days of notice, while Illinois describes 75 full-time employees and Washington describes 50 full-time employees with its stated exceptions. These selected contrasts explain why a federal summary cannot be generalized across jurisdictions. The recommended protocol preserves the employer as filed, exact affected address, source-labelled dates, affected-worker scope, and amendment relationships. A missing notice remains unknown; a matched notice supports the planned event as filed, with current status checked separately. ^{[11] [12] [13]}

Introduction and Background

Worker Adjustment and Retraining Notification (WARN) notices are employer notifications about qualifying workforce events. Federal WARN generally provides 60 calendar days of advance written notice, with recipients including employees or their representatives, the state dislocated worker unit, and the relevant local chief elected official. Its purpose concerns workers and communities facing covered closings or layoffs. ^[2]

For a private equity (PE) origination analyst, the research question is narrower: does this filing justify rechecking a facility, headcount estimate, location, or entity match? A notice is a dated statement of a planned event. Federal regulations specify use of the best information available when the notice is served. It is therefore evidence about what was reported at that time, with later developments requiring separate confirmation. ^[14]

This report's protocol treats sale intent, financial condition, present operating status, and enterprise-wide headcount as separate research fields. None should be populated from a notice alone. A partial reduction at one facility cannot establish that the entire company has closed. Nor should affected-worker counts automatically become a company's remaining headcount.

The source landscape is fragmented. DOL's pandemic-era FAQ states that the federal department does not maintain a WARN notice database, and describes state website publication as voluntary and variable. That historical explanation is consistent with the distinct formats observed here; it is not presented as a newly issued federal statement. ^[15]

The report focuses on 50-state official source discovery and event interpretation, rather than a general guide to acquisition data. It supplies a source matrix, a normalization schema, and decision rules for revisiting a target record. The recommended output of a notice review is a supported update, an assigned research question, or an unresolved signal. Owner willingness to engage requires separate evidence and cannot be inferred from affected employees.

Source Methodology and Coverage

The matrix records a publication check on October 8, 2026, performed for this report. Each state was researched through official agency sources. Landing pages, exposed tables, archive labels, and documented download routes were examined; linked employer letters and workbooks were not exhaustively downloaded. A retrieved interface is not proof that its records are complete.

Exposed coverage means the dates or years visible in the fetched source, or a retention statement on that page. It does not mean every notice within that interval was present. For example, Indiana describes coverage back to 2008, while Oregon states that it retains WARN records for six years. These are different source properties, not competing estimates of layoff activity. ^{[16] [9]}

The table distinguishes original notice availability from a report about notices. Illinois describes its reports as summaries. Michigan says displayed notices are not in their original format, while California documents a public-records request route. Those distinctions affect whether an analyst can inspect exact employer wording. ^{[12] [17] [4]}

Access problems were handled as missing observations. Blocked page fetches were attempted through the supplied permitted fallback. Where a form rendered but its results did not, the matrix confirms the form only. Where the fallback returned navigation, an error, or unreadable document content, the substantive dataset remains unverified.

Format labels refer to Hypertext Markup Language (HTML), Portable Document Format (PDF), and Excel workbook (XLSX) files. FY denotes fiscal year.

The following reading rules apply throughout:

• Unknown means the opened source did not establish that property.
• Linked means a route was visible, without claiming successful file retrieval.
• Tested PDF means a document was opened and readable.
• Archive years describe exposed links or records, not guaranteed continuity.
• Current page describes a route or label, not a complete census.
• Law route identifies verified agency guidance or statutory text, not a full legal comparison.
• Last checked dates the research observation, not the employer filing.
• Amendments remain separate versions unless the source establishes their relationship.

These are research controls. Jurisdiction-specific legal questions require the current agency instructions and applicable statute, including definitions and exceptions.

Official WARN Notices by State

Table 1 is the 50-state official-source matrix. Each source citation opens the checked government page or official document. The state-rule column links only verified official guidance or legal text; an unknown cell does not assert that no state rule exists. A machine-readable version of these rows is included in the accompanying article metadata.

The matrix is a navigation and evidence-quality asset, not a ranking of employers or states. A long archive is useful for historical research, while an original letter can be more useful for resolving a particular address or event. A searchable interface without rendered results still requires a live query or a records request.

The original access audit could not reverify Arkansas and Massachusetts, so their earlier substantive descriptions and citations were removed. An earlier permitted recovery identified Arkansas Workforce Connections. The local board’s PY2024–PY2027 plan describes WARN notices directed through its director or the Rapid Response Team. ^[20] The official employer page could not be independently retrieved in this QA through built-in fetch or the permitted fallback; public-dataset coverage remains unknown. ^[63] For Massachusetts, the matrix records the attempted official route and the failed verification fetch and permitted fallback. The responsible agency, report coverage and format, and letter-request route remain unverified in this report. ^[64] For New Jersey, a readable official business-guidance page was recovered, but the separate labor notice page returned no usable content through the attempted fetch and fallback. Ohio supplied a notice-page heading and submission contact, without notice rows in the recovered representation. Wyoming linked a file whose fallback response did not yield readable record text. These are report-specific access limits.

Historical searches also need separate routes. New York directs notices and amendments from January 1 through April 1, 2025 to its retired database. Texas provides annual Excel files for 2020 through 2026 and an older-record contact; Minnesota provides a request route for records older than three years. ^{[8] [7] [65]}

Normalize Events and Resolve Facility Identity

A normalization process should retain what the source actually says before interpreting it. Federal regulations describe notice information including affected job classifications and worker counts; Wisconsin's guidance identifies the employment site's name and address. These are event and location attributes, not necessarily a company's consolidated identity. ^{[14] [66]}

Table 2 proposes a normalized WARN event schema. These field names are the report's recommendations, not a claim that every state publishes every field.

The schema prevents a source's omissions from disappearing during ingestion. It also makes later correction possible: a changed separation date can replace the current view while the original notice remains in history. Wisconsin specifically says the most recent revision reflects the reported total affected-worker count, which argues against adding original and revised counts together. ^[10]

A practical identity sequence is:

• Locate the exact affected address before matching a company name.
• Compare the filed name with the site's website and official entity record.
• Record a confirmed operating entity separately from a trading name.
• Check whether a location belongs to a subsidiary, contractor, or unrelated namesake.
• Link a parent only when independent evidence establishes that relationship.
• Preserve conflicting names or addresses as open questions.
• Retain the evidence used for each accepted connection.
• Recheck the match when a later notice changes the employer or site.

Facility matching (Hypothetical Example). A fictional notice names Cedar Lantern Packaging at an industrial address. The target list contains Cedar Lantern Holdings at a different headquarters address. The analyst first creates a facility-event record, then looks for evidence connecting the operating business to the holding company. Until that connection is supported, the event stays attached to the facility and an unresolved entity candidate. It does not change the parent's headcount or status.

Analysis of Key Segments

Publication formats and revision behavior

Different source forms support different research tasks. Alabama exposes a CSV route and a year-filtered table; Texas offers annual Excel files. These support repeatable extraction, but a structured row still needs its original field labels and provenance. ^{[3] [7]}

Document-oriented sources support closer reading. South Dakota explicitly identifies linked actual notice PDFs. Mississippi's opened quarterly document was a summary report, while Minnesota exposes employer documents and revision labels. Treating all three as equivalent original letters would lose an important distinction. ^{[53] [36] [34]}

Search forms create another segment. Arizona, Delaware, Kansas, Maine, and Vermont expose query interfaces, but this check did not establish their complete result ranges. An empty or unrendered query result must not become an assertion of no filings. ^{[19] [22] [27] [30] [57]}

Revision management deserves its own workflow. New York publishes modification and retraction instructions, and its retired database includes amended and rescinded labels. Wisconsin defines changes and rescissions; Utah and Nevada expose amended or updated labels. The visible presence of some revisions is not proof of comprehensive revision capture. ^{[44] [67] [10] [56] [40]}

False negatives and mixed record populations

The absence of a retrieved notice is unknown. Federal employer thresholds, event definitions, exceptions, source retention, and publication access all constrain observation. Wisconsin explicitly warns that received notices are not a complete list of potential or actual layoffs and that ultimate affected-worker totals can change without an update reaching the agency. ^[10]

Federal regulations permit reduced notice in specified circumstances, including unforeseeable business circumstances and natural disaster, while requiring as much notice as practicable. A short observed interval therefore needs further assessment; it does not settle whether the requirements were met. ^[14]

A state list may also contain events beyond federal WARN. Maryland includes Economic Stabilization Act (ESA) and other notified dislocation events. Iowa encourages advance notice of layoffs outside the formal requirements. Preserve the source's event classification rather than labelling every row a federally covered filing. ^{[32] [11]}

A research team should therefore separate:

• Not retrieved: access or rendering prevented observation.
• Not located: named searches produced no matching record in the checked scope.
• Outside checked period: the event predates exposed retention.
• Potentially outside coverage: size or event characteristics need legal assessment.
• Matched notice: the filing and entity relationship are supported.
• Revised event: a later source changes the earlier observation.
• Unresolved identity: a notice exists, but the company connection is uncertain.
• Current status unknown: the planned date passed without outcome confirmation.

These are evidence statuses proposed by this report. They support disciplined follow-up without implying a legal conclusion, company distress, or deal availability.

Data Analysis and Evidence

Federal and state scope must remain separate

Federal employer coverage generally includes 100 or more employees, excluding part-time employees, or 100 or more including part-time employees when aggregate work reaches 4,000 hours per week, excluding overtime. The retrieved official regulation edition supplies these definitions. ^[68]

A covered federal plant closing involves employment loss for at least 50 non-part-time employees within 30 days at a single site or relevant operating unit. A mass layoff has a different test: 50 to 499 such employees representing at least 33% of the active workforce, or 500 or more without that percentage condition. Separately subthreshold employment losses can aggregate over 90 days to trigger federal WARN, unless the employer demonstrates that the losses result from separate and distinct actions and causes and are not an attempt to evade WARN. ^[69] These are scope definitions, not an instruction to assess compliance from a public summary. ^{[70] [71]}

Federal employment loss includes a layoff exceeding six months or a reduction exceeding 50% of hours in each month of a six-month period, along with defined terminations. Federal part-time classification includes an average below 20 hours per week or employment for fewer than six of the preceding twelve months. ^{[72] [73]}

The following official state examples demonstrate why federal thresholds cannot be projected nationally:

• California: EDD describes covered establishments with 75 or more full-time and part-time employees in the preceding twelve months, and mass layoffs of 50 or more within 30 days, irrespective of workforce percentage. Its page also describes additional worker-support notice content new for 2026. ^[4]
• Illinois: the agency summarizes coverage at 75 full-time employees and a mass-layoff test involving 25 and one-third of site employees, or 250. ^[12]
• Iowa: the current agency summary identifies employers with 25 or more employees and 30 calendar days of advance notice for covered events. ^[11]
• New Jersey: official business guidance describes an employer threshold of 100, a mass-layoff threshold of 50, part-time inclusion, and 90 days of notice. ^[43]
• Washington: ESD describes coverage at 50 full-time employees and a 60-day notice requirement, subject to stated exceptions. ^[13]
• Vermont: the statute distinguishes 45-day notice to state recipients from 30-day notice to other specified recipients. ^[58]

These are selected contrasts, not a complete compliance table. Definitions and exceptions require jurisdiction-specific review.

Reproducible access audit and date arithmetic

The report's denominator is 50 states checked on October 8, 2026. The original access audit identified 9 states with an explicitly identified CSV, Excel, spreadsheet, or table-export route: Alabama, California, Colorado, Florida, Indiana, Iowa, Rhode Island, Texas, and Virginia. This original calculation counts matrix routes, not delivered files. Oregon and New York document downloads, but their checked text did not establish a specific structured file format. ^{[9] [8]}

Four states expose only PDF summary/annual-report routes among the record formats established here: Idaho, Louisiana, Mississippi, and New Mexico. This classification does not establish that they lack other formats. In the original access audit, four states, Arkansas, Massachusetts, New Jersey, and Ohio, had notice-dataset access unestablished after the required verification and permitted recovery attempts. The remaining rows include mixed HTML, forms, document lists, or formats not established by this check.

Retention and future dates need distinct controls. Oregon states a six-year retention policy. Virginia's selectors extend through FY2027, which does not make that fiscal year complete. Utah's fetched table included a notice-date entry later than the check date; it must be flagged for review rather than counted as an already completed filing. ^{[9] [6] [56]}

Date arithmetic (Hypothetical Example). Suppose a fictional record has a received date of August 3, 2026 and an effective date of October 2, 2026. The observed interval is 60 calendar days, calculated as effective date minus received date. This is an illustrative calculation, not evidence of legal compliance: receipt may differ from service, recipients may differ, and exceptions or later versions may matter.

Implications and Future Directions

A WARN notice should trigger a targeted re-research task. The operating question is which field needs fresh evidence and what source can answer it. The following sequence is recommended:

• Validate the event: retain the exact source record and collection date.
• Resolve the site: compare the affected address with the target's location evidence.
• Review the version: search for an amendment, extension, or rescission.
• Recheck activity: seek dated evidence of whether the facility remains operational.
• Recheck staffing: preserve scope when comparing new headcount evidence.
• Recheck ownership: investigate a separate supported ownership change.
• Assign the gap: give unresolved identity or status questions an owner.
• Log the decision: record the accepted update and what remains unknown.

Federal postponement guidance requires additional notice for shorter delays and a new notice for postponements of 60 days or more. That supports monitoring notice versions rather than scheduling a single irreversible status change on the original effective date. ^[74]

Acquisition due diligence needs an additional distinction. DOL's sale-of-business guidance says that workers who retain their jobs do not suffer employment loss merely because a transaction technically terminates their employment with the seller. Seller and buyer notice responsibilities depend on event timing. A notice is therefore not a dependable proxy for either a completed transaction or an owner seeking one. ^[75]

For Docket, the relevant documented model is sourced target research: identity checks, answer collection, and separate evidence review. The recommendations above fit that research role, while official agencies remain the sources for the underlying notices. Docket does not replace an official filing or resolve a legal question by collecting it.

Table 3 compares the roles of the sources and research outputs used in this workflow. It includes the brand's actual role as a direct target-research provider, without treating it as a state notice publisher.

The comparison separates publishing an event from researching its implication. Further improvements should focus on version links, preserved source labels, and measured retrieval coverage. Increased automation is useful only when a reviewer can still determine which facility, notice version, and date supported a change.

Frequently Asked Questions (FAQs)

How can an analyst find WARN notices by state?

Start with the state row in Table 1, then query the official interface or open the exposed year report. If the event predates the archive or a letter is missing, use the documented request route. Texas and Minnesota illustrate different routes for older data and documents. ^{[7] [65]}

Is there a national official WARN notice database?

DOL's COVID-19 FAQ says that the federal department does not maintain one. This report uses the checked state sources and does not establish the existence of another national official register. Aggregators may help discovery, but their records need official-source confirmation. ^[15]

What does a notice change in acquisition research?

It can justify an event record, a facility-status recheck, or an entity-resolution task. It should not independently change sale intent, companywide headcount, or investment suitability. This is the report's interpretation rule, supported by the notice's prospective, site-specific scope.

Does a missing notice mean no layoff happened?

No. Record the search scope and leave the event unknown. Thresholds, publication limits, inaccessible records, and retention all affect visibility. Wisconsin's warning about incomplete notice coverage is explicit. ^[10]

Conclusion

WARN notices by state are most useful when the authoritative source, affected facility, event version, and observation date remain attached to the finding. The matrix provides an official route for each state and distinguishes verified interfaces from unavailable data, summary reports from letters, and exposed archive years from completeness.

For an origination team, the next action should follow the evidence. A supported site match can create a facility-event record and trigger an operating-status review. A name match without address or entity support should remain an identity question. An inaccessible source should produce a documented retrieval gap. None of those outcomes requires assuming that an owner wants to sell.

The normalized schema makes these decisions reviewable. It preserves employer wording, date labels, affected-worker scope, and amendment relationships. The date calculation provides an observed interval, while legal requirements remain jurisdiction-specific questions governed by current official sources. Revised counts should be treated as versions, and later events should be investigated before updating present status.

The strongest acquisition-research use is disciplined refresh: revisit the relevant location, staffing evidence, ownership record, and active-status question, then record what changed and what remains unresolved. The same protocol works for portfolio monitoring and pre-diligence preparation because it preserves uncertainty rather than forcing every signal into a company-level conclusion.

A maintained source matrix can reduce repeated discovery work. Its value depends on keeping access checks current and allowing unknown findings to remain visible. The resulting target record should express the evidence available, its limitations, and the specific research task justified by the notice.