sergyinfo/h1b-lca-open-data

H-1B / LCA Open Data — aggregate tables (FY2010–FY2026)

Aggregate statistics on 9,331,644 US Labor Condition Applications (LCAs) — the wage-and-worksite filing every employer must submit to the Department of Labor before sponsoring an H-1B, H-1B1, or E-3 worker. Covers 126,466 employers and 782 occupations across fiscal years 2010–2026.

This is a mirror. The canonical home of this dataset — always the newest release, with the data dictionary and license — is https://h1b.report/open-data. Mirrors are refreshed at least one quarterly release behind the canonical page. The same data is also served interactively (charts, rankings, per-employer analytics) at https://h1b.report.

Source

US Department of Labor, Office of Foreign Labor Certification (OFLC) — quarterly LCA disclosure files. Public domain at source.

What's here

sample_occupation_by_state.csv is a 100-row preview. Column-level definitions are in `data_dictionary.md`.

How it's cleaned

Raw OFLC files are inconsistent across years — the schema changed in FY2020 (the FLAG release), employer names are free text, and wages are quoted hourly, weekly, bi-weekly, monthly, or yearly. This release is derived from a normalised corpus:

• —Canonical employers. Name variants ("IBM Corp.", "I.B.M. Corporation", "International Business Machines") are resolved to a single canonical employer, so an employer's filings aren't scattered across spellings. Aggregates key on the canonical employer; the underlying name-variant mapping is not published.
• —SOC occupation codes are normalised to a single taxonomy, with job titles matched against the BLS Direct Match Title File and a classifier for the rest.
• —Annualised wages. WAGE_RATE_OF_PAY_FROM is converted to annual USD (hourly × 2080, weekly × 52, bi-weekly × 26, monthly × 12).

Caveats — please read before citing

• —An LCA is not a visa. It is a prerequisite filing. It is not an H-1B petition, not an approval, and not a hire. One filing can cover multiple positions, and many filings never become petitions. Filing counts are not visa counts.
• —Wages are offered wages, not take-home pay. WAGE_RATE_OF_PAY_FROM is the bottom of the offered range on the filing. It excludes bonus and equity, and it is not verified against what was actually paid.
• —FY2026 is partially reported. It currently holds 210,387 filings against 596,552 in FY2025 (65% lower), which reflects reporting lag, not a real drop. DOL publishes quarterly, so the most recent fiscal year is the least complete and its totals will rise in later releases. It is left uncorrected here — do not read it as a real decline.
• —A small share of filings carry no occupation code. 4.7% of filings (443,000) carry no SOC code. This is a flat background rate, not a gap in particular years: per-year coverage runs 94.5%-95.9%, so years remain comparable with each other. The three occupation_* files only include filings that carry a SOC code. The national, state, and sector files are unaffected — they count every filing. Per-year coverage is published in soc_coverage_by_year.csv.
• —Small cells are suppressed. Any row backed by fewer than 25 filings is omitted, so columns do not sum exactly to national totals.
• —Wage outliers are excluded. Only wages between $20,000 and $1,000,000 annualised are counted, dropping data-entry errors.
• —Percentages are of all filings for that entity, including still-pending ones.

License

CC BY 4.0 — free to share and adapt, including commercially, with attribution. See `LICENSE`.

Citation

h1b.report (2026). H-1B / LCA Open Data: aggregate tables, FY2010–FY2026. Derived from US DOL OFLC LCA disclosure data. https://h1b.report