Source: U.S. Department of Labor, Office of Foreign Labor Certification (OFLC) · Bulk disclosure files downloaded quarterly from DOL · Refreshed Sep 2, 2026
$2.7B
denied
$144.3B
certified
220.6K
workers requested
Stated another way: 98.2¢ of every modeled-gap dollar sits on a filing DOL certified.
The record
The three outcomes
Every LCA ends in one of three outcomes. The modeled gap splits across all three.
Certified
$144.3B
15M workers · 7.9M filings
DOL approved the application. These filings are evidence the employer sought workers at the stated wage; the record does not show a hire or a paycheck.
Denied
$2.7B
220.6K workers · 119.4K filings
DOL refused the application, so no placement could follow from this filing.
Withdrawn
$3.3B
276.3K workers · 189.9K filings
The filing was pulled before DOL ruled on it.
Modeled wage gap by DOL outcome
- Certified$144.3B · 96.0%
- Denied$2.7B · 1.8%
- Withdrawn$3.3B · 2.2%
Median days from submission to decision, by outcome
Withdrawals are fast: the median withdrawn filing was pulled 1 day after submission (90th percentile: 20 days). For comparison, DOL takes a median of 6 days to certify a filing and 4 days to deny one. The record does not say who pulled each filing. Fast withdrawals are consistent with employer-initiated pulls (duplicate filings, changed plans, procedural errors); that is a hypothesis the timing supports, not a fact it proves.
Enforcement over time
Recorded denial rates range up to 1.36% in 2018. The latest recorded rate is 0.53%. Partial years are separate from the historical line. Rates describe decisions already in the archive; they are not projected.
Projected annual filings
2026 projection: 110,574 recorded January–March submissions ×4 = 442,296 annual filings; 585 recorded denials ×4 = 2,340 annual denials. This uses decisions through March 31, 2026 in LCA_Dislclosure_Data_FY2026_Q2.csv. The denial rate stays observed. This simple annualized estimate excludes pending and later decisions and does not adjust for seasonality. It is not a forecast of the final submission cohort.
DOL disclosure files and reporting periods2018
1.36%
8,543 denied
627,149 recorded filings
2019
0.78%
5,168 denied
659,947 recorded filings
2020
0.69%
3,725 denied
542,421 recorded filings
2021
0.47%
2,650 denied
558,556 recorded filings
2022
0.52%
3,029 denied
578,564 recorded filings
2023
0.60%
3,139 denied
521,485 recorded filings
2024
0.73%
3,989 denied
548,651 recorded filings
2025
0.70%
3,868 denied
552,202 recorded filings
2026 Partial / reporting lag
0.53%
593 denied
112,765 recorded filings
Projected: 442,296 filings / 2,340 denials
Denial rate vs filing volume: the enforcement trajectory
Each point is a submission calendar year (2018–2025), connected in chronological order. Incomplete years are excluded from this comparison. These are recorded decisions, which can change as later decisions enter the archive.
Filings with unusually large worker counts
These denied filings each name more than 10,000 workers (the largest names 2,031,308 workers on a single form). Counts remain exactly as filed.
All 5 denied filings naming more than 10,000 workers are listed.
Who is responsible
Who is responsible
Four federal offices decide what happens to every H-1B filing. None of them is villainous; the record above is what their combined capacity produces.
OFLC · Office of Foreign Labor Certification (DOL)
Adjudicates every LCA before a single worker arrives. The certified-vs-denied split above is this office's record.
The archive records 552,202 LCAs submitted in 2025, the latest closed calendar year shown here. Later decisions can revise this cohort. These are recorded filings, not the partial 2026 count or its annualized estimate.
WHD · Wage and Hour Division (DOL)
Investigates employers after the fact: the only unit that audits H-1B wage practice on the ground.
The archive holds 2,887 H-1B cases in total. The chart groups concluded investigations by findings start year. Open-year counts are partial; reporting lag can change earlier years.
USCIS (DHS)
Approves the visa petition itself, but only sees what OFLC already certified.
A certified LCA arrives at USCIS as a clean input, whatever its wage says. The upstream decision is never re-litigated.
DOJ · Civil Division & criminal sections
Prosecutes visa fraud when someone refers a case.
Referrals need evidence assembled first. No office above is funded to build that evidence at program scale.
Give the people inside these offices the benefit of the doubt: the statute asks them to review the 552,202 recorded filings submitted in 2025 against a fraud pattern that only shows up across filings. That is not a diligence gap. It is missing infrastructure. XH1B is that infrastructure: employer-family identity resolution, per-filing risk scoring against a decade of history, and an evidentiary chain of custody on every number above. What gets caught, who profits, and the off-ramp name what the data shows and where the tool plugs in.
What gets caught
How this compares
DOL's Wage and Hour Division, the unit that enforces wage law across all industries, has adjudicated 2,887 H-1B cases assessing $123.4M in back wages against 2,780 employers. The LCA figures on this page are a different measure: a modeled gap on employer-filed requests, $20.6B per year averaged over the certified filings' seven most recent fiscal years. WHD amounts are agency findings on wages actually paid; the LCA gap compares filed offers with an independent BLS benchmark and records no payroll. Different legal bases, different periods, different units: the 167× figure is a size comparison, not a recovery ratio.
What DOL's Wage & Hour Division actually caught
The numbers above are modeled from LCA wage data. These are different: hard, government-adjudicated findings from WHD investigations: the floor of confirmed H-1B harm that DOL itself documented.
2,887
H-1B cases adjudicated
$123.4M
back wages assessed
$11.7M
civil money penalties
13,640
workers affected
Recorded WHD H-1B cases by findings start year
The archive records 112 cases with findings starting in 2014 and 13 in 2025. These comparisons exclude 2026 and later incomplete cohorts. WHD publishes concluded investigations, so reporting lag can change earlier cohorts too.
WHD projection unavailable: no verified reporting cutoff exists for this findings-start-year cohort. 2026 and later counts are partial and are not comparable to a full year.
Top employers by H-1B back wages assessed (top 50 of 2,780)
By state (H-1B cases)
- NJ559 · 26.8%
- CA389 · 18.6%
- TX298 · 14.3%
- NY214 · 10.3%
- IL180 · 8.6%
- MI171 · 8.2%
- FL150 · 7.2%
- VA126 · 6.0%
By industry (H-1B cases)
- Custom Computer Programming Services599 · 41.0%
- Computer Systems Design and Related Services267 · 18.3%
- Computer Systems Design Services207 · 14.2%
- Other Computer Related Services205 · 14.0%
- Other Scientific and Technical Consulting Services131 · 9.0%
- Other Management Consulting Services53 · 3.6%
H-1B Willful Violators (20 CFR §655.810)
Employers DOL formally designated as willful H-1B violators: the most serious determination, requiring intentional disregard or plain indifference.
Source: DOL WHISARD enforcement data (data.dol.gov), FY2005–present. H-1B figures are the subset of WHD cases with H-1B statute violations.
HHS-OIG excluded providers
The federal List of Excluded Individuals/Entities (LEIE) bars providers from federal healthcare programs. An employer appearing here and still filing H-1B petitions is a fraud signal, surfaced per-employer on each org page. This is the national pulse, 2018–present; the type breakdown below starts at 2020.
Exclusion type (2020–present)
- Program-related convictions6,173 · 38.9%
- Patient abuse/neglect (b)4,662 · 29.4%
- Patient abuse or neglect2,048 · 12.9%
- Healthcare-related convictions1,559 · 9.8%
- 1128A41,019 · 6.4%
- Mental health conviction191 · 1.2%
- Loss of license161 · 1.0%
- Felony healthcare fraud47 · 0.3%
Who profits
Top employers
Toggle between the employers whose below-BLS-median wages DOL certified and the ones whose filings it denied.
Employers whose below-market H-1B filings DOL certified: the largest modeled gaps on certified filings.
The same record, ranked and consolidated: the leaderboard ranks every employer, signer, and law firm with per-worker figures and week-over-week movement.
Attorneys whose certified filings carry the largest modeled wage gaps
The attorneys whose own certified H-1B filings carry the largest modeled wage-gap totals: BLS median − offered wage, floored at zero, worker-weighted, summed per attorney. The signature ties each filing to that attorney. Client and willful-client counts are relationship evidence from G28 and PERM links; they never add a client's other filings to the total. These aren't rogue actors; the aggregate shows whose signed filings sit next to the largest totals.
- Karine Faure Wenger - 200 clients (4 willful)worst: Qualcomm Technologies, Inc. ($1.1B)
- Rebecca Michaels - 7 clients (1 willful)worst: Deloitte Consulting LLP ($1.6B)
- Francesca Moschetta - 23 clients (1 willful)worst: GOLDMAN SACHS & CO. LLC ($556.8M)
- Melanie Bradshaw - 3 clientsworst: Ernst & Young U.S. LLP ($1.2B)
- Isha Atassi - 14 clients (1 willful)worst: Deloitte Consulting LLP ($831.5M)
- Navneet Chugh - 419 clients (23 willful)worst: Cyient, Inc. ($453.5M)
- Sapna Annicelli - 3 clients (2 willful)worst: HCL AMERICA INC ($583.7M)
- Thomas Allen - 1,300 clients (106 willful)worst: 3i INFOTECH, INC. ($15.3M)
The distribution beneath the outliers
22,702 attorneys with nonzero modeled wage gaps on their certified filings, bucketed by total. Most cluster in the $100K–$10M range; 5 carry over $1 billion each.
Where
By state
Two views: where the modeled gap on certified filings concentrates, and which DOL regions deny more filings (min 5,000 filings for a stable rate).
Modeled gap share on certified filings, top 8 states
- TX$13.3B · 25.1%
- CA$8.7B · 16.4%
- NY$7.2B · 13.6%
- IL$6.7B · 12.7%
- PA$4.6B · 8.7%
- FL$4.6B · 8.7%
- NJ$4B · 7.6%
- GA$3.8B · 7.2%
Largest modeled gap on certified filings
Highest denial rate
WARN layoffs by state & year
Worker Adjustment and Retraining Notification (WARN) layoff counts for the top 12 states, where the displacements that overlap H-1B hiring concentrate. Brighter cells = more workers laid off that year.
The off-ramp
For enforcement agencies
Every number on this page was computed from public DOL disclosures, the same data OFLC publishes quarterly. The gap between what DOL certifies and what it denies is not a staffing problem or a budget problem. It is an infrastructure problem: the agency adjudicates filings one at a time, with no cross-filing pattern detection, no employer-level wage-theft accumulation, and no signal that flags a filing against the employer's seven-year history before the certified-worker clock starts.
01
Pattern detection across filings
XH1B resolves employer identity across FEIN, name variants, registered agents, and parent–subsidiary links. A filing by 'Amazon.com Services Holdings LLC' is scored against the 91K filings DOL has already certified for the Amazon corporate family. Not in isolation.
02
Pre-certification risk scoring
Every LCA gets a displacement score (wage depression × workers × WARN overlap) before the certified-worker clock starts. WHD investigators can sort the queue by harm, not by submission date.
03
Evidentiary chain of custody
Every number on this site traces back to its original government file, with a verifiable integrity trail from ingest to publication. Findings are courtroom-ready by construction. No screenshots, no spreadsheets, no 'we lost the PDF.'
XH1B is the audit layer the H-1B program has never had. It is available to federal and state enforcement agencies as a hosted deployment, an on-prem install, or a sublicensed API. We are sandbox-ready for a 30-day pilot against any agency's active caseload.
FAQ
Questions about the data & methods
Enforcement FAQ
How does the modeled wage gap split between certified and denied filings?
1.8% of the modeled gap sits on denied filings: $2.7B there, and $144.3B (98.2%) on certified filings. The ratio measures the split between filing outcomes; it is not the share of theft DOL stopped, because the gap is a modeled comparison, not an adjudicated amount.
How is 'wage theft' measured here?
Per filing: gap = max(0, BLS median wage for the occupation − offered annual wage), multiplied by workers requested. Only filings where the offer sits below the market median contribute.
What does the modeled gap on denied filings represent?
A denied LCA is a request DOL refused, so no placement could follow from that filing. Its modeled gap estimates the wage pressure in the refused request against the BLS benchmark. These dollars stay separate from the certified-filing gap and are never summed with it.
Where do these numbers come from?
From the Department of Labor's own quarterly disclosure files (every LCA OFLC publishes), cross-checked against BLS occupational medians and WHD's adjudicated enforcement record. Every figure on this page traces to a public government file.
Method & calculation notes
The modeled gap per filing = (BLS OES national median wage for the SOC − the employer's offered annual wage) × workers requested. Only filings where the offered wage is below the BLS median contribute. The BLS median is an independent market benchmark: it can expose a low offer even where the employer chose a low wage level, which a comparison of the employer's own two filed values would miss. The result is a modeled comparison, not an observation of wages paid.
Certified = DOL approved the application. The filing is evidence the employer sought workers at the stated wage; it does not record a hire or payroll. Denied = DOL refused the application. Withdrawn = the filing was pulled before a decision.
The enforcement ratio is denied ÷ (certified + denied) modeled-gap dollars: the share of the modeled gap that sits on denied filings. It is a ratio between two modeled filing groups, not the share of theft DOL stopped.
Per-filing total_workers is not trusted blindly: a single LCA claiming more than 5,000 workers carries a data-quality flag, because real filings stay far below that. Flagged rows are excluded from the aggregates but stay queryable, and the source values are never altered or deleted. The largest flagged denials appear in the large-worker-count section above.
Source: DOL OFLC LCA disclosures + BLS OES, over the LCA years now loaded (known year gaps are listed on the data sources page). Computed in the wage_gap_status_breakdown materialized view.
Public source records and site analysis. We do not attest to accuracy. Read the full disclaimer.
Dataset guides and site methods
- [1] Dataset documentation · DOL OFLC. LCA public disclosure record layout (FY2026 Q3). Pages 1–4: application, DOL outcome and dates, requested positions, employer, representatives, worksites and pay unitsThis layout covers determinations issued October 1, 2025 through June 30, 2026. It documents employer-provided application fields and DOL decisions, not USCIS decisions, payroll verification, legal bills or worker-linked moves. Older layouts can differ.Checked 2026-10-17.
- [2] Dataset documentation · BLS. OEWS technical notes (May 2025). Scope of the survey; occupational coding; concepts and estimation methodologyDocuments occupation and geographic wage estimates, not an individual worker’s pay. Use the estimate year and geography stated by the site; older releases can differ.Checked 2026-10-17.
- [3] Dataset documentation · DOL Wage and Hour Division. Wage and Hour Division Compliance Action Data. Dataset description: concluded actions since FY2005, back wages, employees and civil money penaltiesThe published dataset covers concluded compliance actions, including whether violations were found. It is not a list of pending complaints or proof that every reported amount was collected.Checked 2026-10-18.
- [4] Primary record · DOL Wage and Hour Division. H-1B Willful Violator List. List effective October 1, 2026; named employers and willful-violation determination datesThis list gives determination dates, not disqualification periods. A historical listing does not prove a current bar or wrongdoing by a similarly named entity.Checked 2026-10-17.
- [5] Dataset documentation · HHS OIG. LEIE Database & Supplement Downloads. LEIE Database Download; monthly file strategy; filename datesThe full file lists exclusions in effect. OIG removes reinstated entries and replaces the file monthly; the filename refers to actions through the previous month.Checked 2026-10-18.
- [6] Guidance · HHS OIG. Background Information & Exclusion Authorities. Types of exclusions; effects; statutory authority tableMandatory and permissive authorities differ. The type code must be read against its statutory ground; an exclusion is not automatically a criminal conviction.Checked 2026-10-18.
- [7] Guidance · HHS OIG. LEIE Database Record Layout. One-page field list: names, EXCLTYPE, EXCLDATE, REINDATE, WAIVERDATE, WAIVERSTATEThe layout records exclusion, reinstatement and waiver fields. It does not define every field’s legal meaning or verify an identity match.Checked 2026-10-18.
- [8] XH1B method or registry · XH1B. WARN Notices (mass layoffs): source registry; Reading Public Records and Derived Indicators. Source registry entry: warn; Reading records: units, source coverage, wage gaps, entity links, and inference limitsXH1B dataset description: One row per source_notice_id. The definition of a mass layoff and the layoff threshold vary by state. The threshold is typically 50 or more employees.; These are the site's review methods, not validated predictions of wrongdoing. Inspect underlying records and alternative explanations.
- [9] XH1B method or registry · XH1B. Displacement Model and Assumptions. Eligibility, requested-position weighting, scopes, WARN matching, and counting limitsModel outputs are estimates. A matched layoff notice does not prove that an H-1B worker replaced a laid-off worker.