Wage and Hour Due Diligence for Private Equity: DOL Hits

Executive Summary

Wage and hour due diligence for private equity starts with employer identity, then the published outcome. A company-name search hit in the United States Department of Labor (DOL) Wage and Hour Division (WHD) database is a candidate record to resolve. DOL describes coverage as concluded compliance actions since fiscal year 2005, including whether violations were found. Inclusion alone therefore does not establish an adverse finding. The Government Accountability Office (GAO) also explicitly includes cases with and without violations in its explanation of WHD case statistics. ^{[1] [2]}

The recommended protocol separates matched, not matched, and unresolved identities from the outcome reported in the source. Keep the employer, worksite, legal aliases, action identifier, relevant laws, findings dates, outcome fields, amounts, source and freshness evidence together. An Employer Identification Number (EIN) is a federal tax identifier; it is corroborating evidence when legitimately available, rather than an identifier to assume appears in every public WHD record. ^[3] At agency level, WHD reports 16,924 concluded actions and $259,294,764 in back wages for fiscal year 2025; those totals provide context rather than a target-risk benchmark. ^[4]

Currency requires separate evidence. The Data.gov catalog displayed a dataset-last-updated timestamp of March 6, 2026 and a catalog-last-checked timestamp of September 10, 2026 when consulted for this report. Neither establishes a verified current downloadable-file date or a maximum delay after conclusion. Record those unknowns rather than treating a successful search as current clearance. ^[5]

State records require their own interpretation. New Jersey's Wage Violation Watchlist uses final determinations and retains employer names after payment; Massachusetts warns that posting complaint information does not establish its merits. Those sources cannot share an undifferentiated adverse-hit flag. ^{[6] [7]} Nor can national employment statistics supply a defensible violation-rate denominator: the Bureau of Labor Statistics (BLS) Quarterly Census of Employment and Wages (QCEW) describes employment and wage coverage, not a population of WHD compliance examinations. ^[8] This report provides a decision table and fictional worksheet for research handoff. A matched finding prompts a defined records request and counsel review, while an empty result remains a documented search outcome rather than evidence of compliance.

Introduction and Background

Labor-intensive service businesses can present a research problem before a deal team reaches substantive diligence: which employer does an official record actually describe? A target may be marketed through a brand, operate through several legal entities, and employ people at multiple locations. The Small Business Administration (SBA) cautions that different businesses can use the same doing-business-as (DBA) name within a state. A trading-name match therefore deserves verification before it enters an acquisition screen. ^[9]

The unit being researched matters. Census defines an establishment as a physical business location and cautions that an establishment is not necessarily a company or enterprise. A location-level record should not automatically become a conclusion about an entire acquisition perimeter. ^{[10] [11]} The practical question is narrower: does the published action refer to an employer or establishment included in the proposed target, and what does that specific record actually state?

For Docket's private-equity origination audience, the recommended deliverable is an evidence-backed research handoff. It should enable a deal team to inspect the identity match, the published outcome and the unresolved questions. It should also distinguish an analyst's screening recommendation from a transaction-specific legal conclusion.

The search mandate should define operating states, legal entities, aliases, relevant periods and the decision the screen supports. An initial sourcing exercise may identify a need for further diligence; a later acquisition workstream may require records and explanations unavailable through a public search. California's wage-claim guidance, for example, points to paystubs as a source of the employer's actual name. That supports asking for permitted employer documentation when public identity evidence remains ambiguous. ^[12]

This guide addresses federal database discovery, record interpretation and escalation. It proposes an operating method, not a government-prescribed matching test or a certification of compliance. The following distinctions preserve the evidence needed for a useful handoff.

What a WHD Record Can Answer

Separate identity, outcome and present status

A case-hit review should answer distinct questions:

• Identity: Is this the employer or worksite within the target perimeter?
• Outcome: What does the published action report for the relevant statute or program?
• Timing: Which dates describe the findings, publication and search?
• Follow-up: What evidence is still needed before the deal team can assess significance?

These questions should have separate fields. DOL's description says that the dataset includes whether violations were found; it does not say that every included employer had a violation. GAO's case-count explanation makes the inclusion of no-violation outcomes explicit. ^{[1] [2]} A research system that assigns an adverse label simply because a row exists loses this distinction at ingestion.

A matched record is also different from a matched parent. Census defines an enterprise through establishments under common ownership or control. That statistical grouping helps explain why location and enterprise are different units, but it does not determine which entity bears a transaction-specific obligation. ^[10] Record the actual named employer first; add any documented family relationship as a separate, dated assertion.

Avoid translating unlike state records into one outcome

New Jersey's Watchlist is based on final determinations, whereas Massachusetts expressly warns against treating published complaint information as a merits determination. ^{[6] [7]} New York's process likewise distinguishes assignment and handling of a claim from the later determination whether a violation occurred. ^[13] The word “case” therefore cannot be used as a universal synonym for a substantiated finding.

Recommended outcome labels include published finding, published no-violation outcome, complaint without a merits determination, and outcome unresolved. These are analytical labels whose mapping should retain the original source value. Do not fill a missing outcome with a negative value simply because monetary fields are blank.

An empty search answers an even narrower question: the specified queries did not return a usable record in the source consulted. GAO describes complaints that do not become compliance actions, including matters outside WHD jurisdiction. That historical account illustrates why complaint intake and concluded-action publication are different populations. ^[14] A no-hit result should preserve scope and uncertainty.

Locate the Case-Level Dataset and Record Its Currency

Begin with Wage and Hour Division Compliance Action Data in the official catalog. Data.gov identifies it as the concluded-action dataset; DOL distinguishes case-level enforcement data from its fiscal-year aggregate tables. An aggregate chart is useful context, but cannot establish an individual employer's history. ^{[5] [15]}

Use the following source-navigation sequence:

• Find the dataset: Search the official DOL catalog by dataset name and agency.
• Read metadata: Confirm the population, variable definitions and access route.
• Open case data: Use the portal's download or documented query mechanism.
• Preserve the extract: Retain the exact returned rows and the query conditions.
• Record limitations: Keep access failures, missing fields and update uncertainty visible.

DOL's user guide supports sorting or filtering its catalog and recommends understanding variable properties before using a dataset. Its dataset page provides a download route and API Query Builder. ^{[16] [17]} An application programming interface (API) can support repeatable collection, but DOL requires an API key for its metadata and data requests. The documentation's access requirements should be checked before designing automated collection. ^[16]

Keep distinct dates distinct

The catalog displayed March 6, 2026 at 05:04 PM as “Dataset Last Updated” and September 10, 2026 at 05:54 PM as “Catalog Last Checked.” These are different labels and must remain different fields. ^[5] Its quarterly accrual metadata describes cataloged frequency, not a verified deadline for publication after an action concludes. ^[5]

For this report, the latest underlying file-modification date and a quantified closure-to-publication lag were not established. The source was accessible through the official portal, but accessibility does not answer either freshness question. Do not substitute the research date or catalog-check date for file currency.

The recommended worksheet records:

• Findings dates: Values describing the findings period in the native record.
• Conclusion date: A separately sourced conclusion date, or “not established.”
• Source update: The publisher's applicable file or dataset date, if verified.
• Catalog date: Metadata timestamps, retained with their original labels.
• Access date: When the research team obtained the material.

If a native field's meaning is uncertain, pause its interpretation rather than its preservation. Retaining the raw value allows later correction without repeating the identity search. For any comparison across extracts, retain the version and the rule used to reconcile changed records.

Resolve Employer and Establishment Identity

Build a dated crosswalk

Start from the proposed acquisition perimeter, then map each published employer to it. Recommended crosswalk fields include:

• Searched company: The target label used by the investment team.
• Legal entities: Exact names and jurisdictions supported by documents.
• Aliases: Trade names, former names and documented spelling variants.
• Locations: Operating and historical addresses, with relevant dates.
• Relationships: Worksite, employer, subsidiary and parent, kept separately.
• Evidence: Source, supporting passage, access date and unresolved conflict.

A DBA is not a unique identifier, and a domain name need not match the legal name, trademark or DBA. Both SBA cautions support using names and websites as leads rather than automatic merge keys. ^[9] California's instruction to identify the employer from paystubs supplies a useful route when a target can provide documentation. ^[12]

The IRS says EIN applications should use the business name shown on registration or formation documents. It also says a business-name change does not itself require a new EIN. A name difference may therefore warrant historical reconciliation rather than immediate rejection. ^[3] Conversely, Census explains that an enterprise may have several EINs and an EIN may cover multiple establishments. Neither an EIN nor an address should be treated as a universal parent-worksite crosswalk. ^[11]

Use external identifiers within their scope

For entities with securities filings, the Securities and Exchange Commission (SEC) assigns a Central Index Key (CIK). SEC guidance says searching by it narrows the search to the exact company, and company-search results can provide address and incorporation details. These are useful corroborating fields, not evidence that the WHD row contains a CIK. ^{[18] [19]}

Regulation S-K's subsidiary-exhibit requirement includes jurisdictions and names under which subsidiaries do business. It also permits omission of particular subsidiary names under its significance condition. An absent name in a subsidiary exhibit therefore does not alone disprove a relationship. ^[20] For a government contractor, Defense Logistics Agency guidance identifies SAM.gov as a route to entity registration records (Source: www.dla.mil); use any registration evidence with its own date and scope.

Recommended match decisions are deliberately reversible. Assign matched when the employer and relevant location or historical identity have adequate documentary support. Assign not matched when evidence affirmatively establishes a different employer. Assign unresolved when similar names, shared addresses, old locations or unsupported parent links leave material ambiguity. Preserve rejected candidates so a future run can explain why they were excluded.

Read Outcomes, Dates and Amounts Literally

The native export and the analyst worksheet should remain distinguishable. The official dictionary defines case_id as the case identifier, trade_nm as employer name, and legal_name as employer legal name. case_violtn_cnt is total case violations. The monetary and employee fields include bw_atp_amt, total back wages agreed to pay; ee_atp_cnt, total employees agreed to pay; and cmp_assd, total civil monetary penalty assessments. Preserve address fields and the statute-specific blocks alongside those totals. ^[17]

In particular, violation found is an analyst interpretation, not a native Boolean field. Preserve the raw case and relevant statute counts: a positive count reports violations, a zero reports none in that field, and a blank remains unknown. Zero dollars alone cannot establish a no-violation outcome. findings_start_date and findings_end_date describe when findings first and last occurred; findings can include a violation or no violation found. They are not case-opening or case-closing dates. The latter dates are absent from this dataset. load_dt is labeled for internal purposes, so it should not substitute for verified file currency. Leave conclusion date unestablished unless another source supplies it. ^[17]

DOL describes employees due back wages and penalties assessed; the dictionary also uses agreed-to-pay back-wage language. These labels do not, by themselves, establish cash recovery, current unpaid balances or an acquisition adjustment. ^[17] Ask counsel and the deal team which underlying documents are needed to interpret a matched amount.

Table 1 sets out the recommended case-hit decision table. Its statuses are proposed research controls, not official WHD classifications.

The table keeps identity confidence separate from outcome interpretation. A different employer should be excluded even if its record is detailed. A no-violation outcome should be retained even when a screening workflow expected an adverse result. A missing field remains unknown.

Use case totals for case-level reporting and keep statute amounts as components; do not add a total to its components. If a working file expands statutes into rows, do not count those rows as distinct actions. DOL warns that multiple laws can apply in a single compliance action. ^[15] Its dictionary further warns that violation counts after the procedural change on October 1, 2025 are not directly comparable with earlier counts. Preserve that break when analyzing history. ^[17]

Compare Federal and Chosen-State Sources

Choose state checks from the worksite footprint and written mandate. Do not represent the following selection as national coverage. Its purpose is to show how scope and source type change the meaning of a hit.

New Jersey illustrates a material difference from the federal dataset. Its Watchlist publishes employer and DBA information, addresses, claim categories, affected employees, assessed wages, fees and penalties, status, and a final-order date. ^[6] The page displayed October 15, 2026 as its latest update ^[6] and says it updates monthly. ^[6]

Table 2 compares the federal source with selected state routes. State complaint forms and records-request pages are navigation aids, not substitutes for case-level findings.

These sources support different statements. A federal concluded action, a state final determination and a complaint entry should retain their original categories. A records-request route establishes where to ask, not that a target has a published adverse outcome.

Coverage exclusions deserve a separate field. New Jersey excludes its Prevailing Wage Act and Public Works Contractor Registration Act findings from this Watchlist and points toward a separate source. ^[6] Massachusetts says complaint data older than six years is unavailable. ^[7] A researcher who searches only one list, or only the visible retention window, must disclose that boundary. The search result cannot establish that excluded programs or older periods are clear.

Implement the Match-and-Escalate Handoff

Preserve the search and assign the next action

Use a search log alongside the case worksheet:

• Scope: Legal entities, jurisdictions, programs and periods included.
• Queries: Exact company names, aliases and location filters.
• Result: Candidates, exclusions, matched actions and no-hit searches.
• Provenance: Case source URL, retained extract and collection date.
• Currency: Publisher dates, catalog dates and unresolved update questions.
• Owner: Researcher, reviewer and person responsible for pending work.
• Decision: Match status, outcome label and rationale.
• Escalation: Specific records needed, recipient and review condition.

These are recommended operating fields. They allow a team to distinguish a completed negative search from an unfinished review, and a rejected candidate from a matched action whose implications remain open.

Check existing public records before requesting more material, as FOIA.gov recommends. A federal Freedom of Information Act (FOIA) request should be written and reasonably describe the records sought. It should ask for an identifiable record, not ask the agency to conduct acquisition analysis. FOIA does not require creation of new records, research or answers to questions. ^[25]

A useful counsel handoff identifies the specific employer, action identifier, available findings period, source fields and factual uncertainty. Ask counsel to determine what the available material means for the transaction and which additional documents are needed. Do not convert a published amount into a liability estimate or a current-status label without that work.

State requests have their own route. New York explicitly asks for relevant names, dates and descriptions; California directs wage-claim records toward the Labor Commissioner's Office. ^{[22] [23]} Record-request timing should remain an open dependency: FOIA.gov says processing depends on complexity and backlog. ^[25]

Fictional filled worksheet (Hypothetical Example)

The following employer, identifiers, dates, addresses and amounts are entirely invented. They demonstrate record structure and do not describe an actual company or DOL case. The example intentionally leaves a date and a source link unresolved.

Table 3 shows a filled analyst worksheet for an invented service business. Values are illustrative, not claims about a real employer.

The worksheet demonstrates the intended handoff: an employer-level match can coexist with uncertainty about the parent, the conclusion date and current status. The fictional source URL is deliberately not replaced with a real link, which would create false evidence. A production record with the same gap would remain incomplete.

Data Analysis and Evidence

Quantify published activity without estimating target risk

WHD's aggregate All Acts table reports 16,924 concluded compliance actions, $259,294,764 in back wages, 176,957 employees receiving back wages, and $58,699,936 in civil money penalties assessed for fiscal year 2025. ^[4] For fiscal year 2024, the corresponding figures are 17,300, $202,676,115, 151,989, and $35,920,310. ^[4] These are agency-wide published totals, not measurements for a target company.

Calculated from those figures, concluded actions decreased by 376, approximately 2.2%, while reported back wages increased by $56,618,649, approximately 27.9%. ^[4] These calculations compare the displayed annual totals only. They do not explain the causes of the change or establish a comparable-employer benchmark.

Selection matters. GAO's historical report says most Fair Labor Standards Act (FLSA) actions in its studied period targeted priority industries. Its methodology also distinguishes agency-directed matters from complaint-initiated cases. This is a reason to avoid treating observed actions as a random sample, not a claim about the current distribution of cases. ^{[14] [2]}

BLS says QCEW covers more than 95% of U.S. jobs and tabulates establishments, employment and wages. ^[8] That breadth does not turn QCEW establishments into a denominator of employers examined by WHD. Census's distinction between establishments and companies creates an additional unit mismatch. ^[11] BLS also does not publicly release a particular company's employment count through QCEW under its statistical-confidentiality pledge. ^[26] The dataset cannot resolve an individual action's affected-employee denominator.

Count matched actions under a written rule

For a target screen, the recommended quantitative output is a set of counts whose definitions can be audited:

• Distinct matched actions: Deduplicate by the action identifier within the retained source version.
• Found-violation actions: A separate count using the supported outcome mapping.
• No-violation actions: Preserve these as their own published outcomes.
• Unresolved outcomes: Do not include them in either confirmed-outcome count.
• Unresolved identities: Keep candidate records outside the matched totals.
• Coverage: Report queried entities, periods and jurisdictions separately.

Any date grouping should state the native date used. If conclusion dates are unavailable, label the analysis by findings dates rather than describing it as a closure-year series. If statute-specific records are used, reconcile their relationship to distinct action identifiers before summing.

Violation-count comparability requires special care: DOL's dictionary flags a procedural change affecting records loaded after October 1, 2025. A trend spanning that boundary should not equate counts produced under different procedures. ^[17] Report matched actions and literal outcomes with explicit coverage instead of a headline violation rate.

No peer benchmark or target compliance probability is calculated here. A rate would require a defensible population, time window, observation process and comparable unit of analysis. Public search visibility alone supplies none of those conditions.

Implications and Future Directions

The useful output of origination research is a reproducible decision about follow-up. It should make clear which employer was matched, which action was interpreted, and which questions remain for the deal team. That design allows a later reviewer to correct identity or outcome mapping without erasing the original evidence.

Freshness should be managed by field. A name may change while an EIN persists, according to IRS guidance. ^[3] A state list may retain a paid historical entry, as New Jersey explains. ^[6] A complaint dataset may exclude older material under its retention policy, as Massachusetts states. ^[7] These are different changes and should prompt different research actions: update an alias, examine status documentation, or disclose a historical-coverage gap.

For future refreshes, the proposed controls are:

• Retain versions: Keep the source extract used for the earlier decision.
• Compare identities: Recheck changed names, locations and perimeter membership.
• Compare outcomes: Preserve the old value and the reason for any revision.
• Recheck definitions: Review dictionary changes before extending a trend.
• Reopen dependencies: Assign unresolved records and document-request results.
• Review significance: Route new substantive questions to transaction counsel.

Automation can collect and reconcile material under these rules. It should preserve uncertainty rather than fill empty fields with assumed clearance or adverse status. Where the public record does not answer the question, request the relevant existing documents and retain the dependency. FOIA's limits on creating records or conducting research make precise requests particularly useful. ^[25]

The same discipline should govern acquisition screening thresholds. A rule that stops research because an employer name appears in a source should be reconsidered: identity and published outcome are necessary inputs. Conversely, a search that returns nothing should remain bounded by its documented sources and queries. Neither result substitutes for the transaction's broader diligence mandate.

Conclusion

Wage and hour due diligence for private equity is most useful when it produces a traceable employer-to-action match and a precise description of the published outcome. The recommended protocol begins with the target perimeter, searches documented names and aliases, resolves the employer and location, and retains the source record before deciding what to escalate.

The case-hit decision table keeps identity separate from outcome. The source comparison keeps federal concluded actions separate from state complaints, final determinations and records-request routes. The fictional worksheet shows how a matched employer can still have unresolved parentage, dates, disposition and currency.

For practical handoff, preserve the legal aliases, employer address, worksite relationship, action identifier, statute, native findings dates, separately sourced conclusion date, outcome mapping, amount labels, case URL, update evidence, match status and counsel-review trigger. Keep excluded candidates and no-hit searches, too, so the next reviewer can reconstruct the scope.

The resulting research statement should be narrow enough to verify and specific enough to support the next action. A matched finding calls for defined follow-up. A published no-violation outcome remains that action's reported outcome. A blank field, unresolved identity or empty search remains an explicit limit. Transaction counsel and the deal team can then assess the documents and current facts needed for the acquisition.