BLS Attribution, Extraction Dates, and Production Data Pipeline Specification

A Transparency and Credibility Reference for the Wage Floor Calculator and Related Platform Documents

v1.0 · Created May 13, 2026 for v3.7.51 (transparency reference responding to user request: "is a real BLS catalog list and/or real BLS wage data available to use in the platform for reference and calculated estimates? document the source and date extracted in the platform. make sure any reference in the platform describes the source data used to provide transparency and creditability.")

Jason Robertson · Ohio · 2026

1. Why This Document Exists

The wage floor architecture described across the platform's Pillar 2 documents (Wage Floors) is the platform's most empirically-dependent design element. The proposed mechanism sets regional wage floors at the 25th percentile of each occupation's wage distribution within each metropolitan statistical area, with annual recalibration via smoothed three-year moving averages. The credibility of this proposal depends on the existence of a reliable, ongoing, authoritative data source for the underlying wages.

This document answers three questions: (1) does such a data source exist; (2) what specifically does it publish and on what schedule; (3) what is the production data pipeline for the platform's wage floor calculations to consume that data. The answers are documented here so that every reference to wage data elsewhere in the platform can point to this central reference rather than restating the source attribution.

2. Yes — The Data Source Exists and Is Authoritative

The relevant data source is the U.S. Bureau of Labor Statistics (BLS), specifically the Occupational Employment and Wage Statistics (OEWS) program. OEWS is the federal government's authoritative source for occupational employment and wage statistics. It is a continuous semiannual survey of approximately 1.1 million establishments collected over a three-year period for each annual release. The May 2025 estimates, for example, are based on responses from six semiannual panels (May 2024, November 2023, May 2023, November 2022, May 2022, November 2021), representing approximately 55 percent of total national employment with a response rate of approximately 65.7 percent of establishments.

OEWS publishes employment and wage estimates for approximately 830 detailed occupations across 22 of the 23 major SOC groups (major group 55 Military Specific Occupations is not included), at the following geographic levels: national; states (50 states plus DC plus Guam plus Puerto Rico plus the Virgin Islands); approximately 388 metropolitan statistical areas; approximately 31 metropolitan divisions; and approximately 175 non-metropolitan area definitions. National data are also published by NAICS industry classification.

OEWS uses the 2018 Standard Occupational Classification system maintained by the Office of Management and Budget. This is the same classification system used by every other federal agency and is the legally-defined occupational taxonomy referenced in federal statute and regulation. Production-grade wage floor calculations under the platform's proposal would use OEWS data and the 2018 SOC classification directly.

3. Publication Schedule and Currency

OEWS publishes annually. The most recent release as of this document's authoring (May 13, 2026) is the May 2025 estimates, which were released April 2025. The the May 2025 estimates were released May 15, 2026. This document and the calculator integrate the May 2025 data. After May 15, 2026 the platform's wage floor data would be refreshed to incorporate the May 2025 release.

The annual publication cadence aligns well with the platform's proposed annual recalibration of wage floors. The platform's wage floor methodology calls for smoothed three-year moving averages, which works directly with the rolling six-panel OEWS estimation methodology BLS already uses.

4. Authoritative URLs and Data Formats

The primary URLs for OEWS data are listed below. The full XLSX downloads contain percentile breakdowns (10th, 25th, 50th median, 75th, 90th) that are not visible in the HTML news release tables (which show only median and mean). For the platform's 25th-percentile-based wage floor architecture, the XLSX downloads are the production data source.

The XLSX downloads are ZIP archives containing Excel files. Each row in the national file corresponds to a SOC occupation and includes employment count, mean wage (hourly and annual), median (50th percentile), 25th percentile, 75th percentile, and 90th percentile. The MSA file extends this with one row per occupation × area combination. The all-data file combines national, state, MSA, non-metro, and industry breakdowns.

5. What v3.7.51 Currently Uses

The Wage Floor Comparison Calculator integrates real BLS OEWS May 2025 national wage data for 792 of the 866 occupations in its catalog. For these occupations the 25th-percentile wage floor is now read DIRECTLY from the BLS national XLSX (the A_PCT25 column), retiring the earlier median x 0.74 approximation. These BLS-sourced occupations are tagged with a BLS indicator in the calculator's National card. The verbatim BLS source file is archived as a dated snapshot with SHA-256 checksums (see Section 10).

The remaining 81 occupations fall into three groups: 71 display illustrative figures (SOC codes with no matching BLS May 2025 detailed occupation, tagged ILLUS), 7 are held at their prior value and flagged for a SOC-restructure cleanup (occupations whose 2018-SOC code renumbering could not be resolved 1:1 without ambiguity, e.g. the Database Administrator/Architect split and the railroad family renumbering; see the mapping-corrections record in the data archive), and 3 are BLS-suppressed (BLS does not publish a 25th-percentile for that occupation). The illustrative figures are approximate BLS-typical magnitudes and are not BLS-published values; they should not be cited as authoritative.

Regional wage data is now real. Each platform region that maps to a BLS area (430 of 452) carries real occupation-by-area 25th-percentile wages from the BLS metropolitan and nonmetropolitan XLSX files, served as per-region data shards that the calculator fetches on demand. Each region also has an empirical single-factor summary (the median of its real occupation-level regional/national p25 ratios) used as an offline fallback. 29 regions could not be cleanly matched to a single BLS area (nested in a larger metro, renamed, or with no BLS nonmetropolitan territory) and retain an estimated factor, transparently flagged. A key correction from real data: the Direction F wage-floor gap is concentrated in Puerto Rico (empirical factors 0.60-0.63); no mainland region falls below the 0.70 threshold on its regional factor, though specific low-wage occupations within a region still can.

6. Production Data Pipeline (Future Implementation)

The production data pipeline for the wage floor calculator is now implemented for both national and regional data via re-runnable ingestion scripts: tools/build_wage_floor_data.py (national p25) and tools/build_wage_regional.py (occupation-by-area regional p25 shards). Both consume immutable inputs (the raw BLS XLSX files plus curated crosswalk/correction specs) and produce deterministic output.

Stage 1: Automated download. Schedule annual ingestion to fire on the BLS OEWS release date (typically late March or April for the previous year's May data). Download the national XLSX, the MSA / non-metro XLSX, and the All Data XLSX from the BLS special-requests endpoints. Verify file integrity via checksums or file-size sanity checks. Archive raw downloads with a date-stamped folder structure for audit trail.

Stage 2: Parse and normalize. Parse the Excel files into structured tables: one row per SOC occupation × geographic area × statistic. Extract employment count, mean wage, and all five published percentiles (p10, p25, p50, p75, p90). Normalize SOC codes against the authoritative 2018 SOC catalog. Flag any occupation × area pairs where BLS suppressed the data (typically because the underlying sample was too small for reliable estimation).

Stage 3: Compute operative wage floors. For each occupation × area, the operative wage floor is the BLS-published 25th percentile directly. For occupation × area pairs where BLS suppressed the p25 (low-employment occupations in small areas), apply the platform's defined fallback methodology (typically: aggregate to the state level; if still suppressed, aggregate to the BLS-defined non-metropolitan area; if still suppressed, fall back to the national p25 for that occupation).

Stage 4: Apply Direction F threshold check (if Direction F is implemented in production). For each occupation × area, compute regional p25 as a percentage of national p25 for that occupation. If the percentage is below the chosen Direction F threshold (60, 65, or 70 percent, calibrated by credentialed external review), set the operative floor to the threshold × national p25 rather than the regional p25. Otherwise, the operative floor remains the regional p25.

Stage 5: Publish and integrate. Publish the operative wage floors as an open dataset, downloadable by occupation, by region, or as bulk download. Make the wage floors accessible via the calculator UI (replacing the current illustrative-plus-partial-BLS data). Integrate with the IRS / DOL enforcement infrastructure described in Direction G. Smooth annual updates with a three-year rolling average to prevent year-to-year volatility from disrupting compliance plans.

7. Documentation Standards Going Forward

Every place in the platform that displays or references a wage figure should follow these documentation standards. (a) Cite BLS OEWS as the authoritative source with a specific data vintage (e.g., "May 2025 estimates"). (b) Note the extraction date when the platform's stored data was last refreshed. (c) Distinguish BLS-sourced figures from illustrative approximations using a consistent visual indicator. (d) Link to the BLS source URLs so readers can verify or download original data directly. (e) Note any derivation steps applied to BLS data (e.g., 25th percentile derived as median × 0.74 when only median is available from the HTML release).

The Wage Floor Comparison Calculator implements these standards in its caveat box and per-occupation source tags. The Wage Floor Concept Analysis Section 8 exhibit and the Regional Variation Examples companion document have been updated to cite the same BLS source with the same extraction date. The Pay Gap and Indirect Mechanisms Direction F discussion has been updated to point to this Data Sources document as the centralized reference. Any future addition that references wage data should cite this document or restate the standards above.

8. Limitations of the v3.7.51 Implementation

Limitations of the current implementation. First, 792 of the 866 occupations (91 percent) carry real BLS national 25th-percentile wage data read directly from the XLSX; the remaining 81 are illustrative (71), flagged for SOC-restructure cleanup (7), or BLS-suppressed (3). Second, regional data is real occupation-by-area 25th-percentile for 430 of 452 regions; the 22 unmatched regions use an estimated factor and are flagged. Third, the empirical region factor is a median-of-ratios summary; the authoritative regional values are the per-occupation shards, which the calculator uses directly. Credentialed external review of the Direction F threshold calibration remains an open work item.

None of these limitations affect the underlying wage floor architecture or the empirical evidence base cited in the Wage Floor Concept Analysis (Dube and Zipperer 2024 NBER meta-analysis). The limitations are about the current calculator's display values, not about the policy proposal's empirical foundation. The policy proposal stands on its own and would be implemented against authoritative BLS data in production.

9. Cross-References

This document is the central transparency reference. The following platform documents reference this document or restate its standards: Wage Floor Concept Analysis (02_Wage_Floor_Concept_Analysis_v02.docx), Wage Floor Regional Variation Examples (02_Wage_Floor_Regional_Variation_Examples.docx), Pay Gap and Indirect Mechanisms (05_Pay_Gap_And_Indirect_Mechanisms.docx) for Direction F and Direction G discussion, Wage Floors As Tax Architecture (05_Wage_Floors_As_Tax_Architecture.docx), and the Wage Floor Empirical Analysis Excel model (04_Wage_Floor_Empirical_Analysis.xlsx). The interactive Wage Floor Comparison Calculator (06_Wage_Floor_Comparison_Calculator.html) implements the documentation standards directly in its UI.

10. Audit-Trail Snapshots in _data_archive/

The platform's _data_archive/ directory implements an audit-trail archival pattern: each external source is captured as a verbatim snapshot with date, SHA-256 checksums, and machine-readable provenance metadata. As of the national BLS ingestion, this pattern is POPULATED: the directory _data_archive/2026-07-10_bls_oews_may2025/ contains the verbatim BLS national OEWS May 2025 XLSX with a real SHA-256 checksum in MANIFEST.json, plus the mapping-corrections record. The pattern serves four purposes: reproducibility (a reader can verify cited national wage values against the archived source file), audit trail (if BLS revises its data, the platform's basis is preserved), credibility defense (disputed national values can be defended with snapshot evidence), and a refresh pattern (each annual data refresh adds a new dated snapshot). Note: the national wage figures are now a real verbatim BLS pull; regional figures remain illustrative single-factor approximations pending Phase 2.

The snapshot pattern is realized as follows for the national data pull: the dated directory _data_archive/2026-07-10_bls_oews_may2025/ contains the verbatim BLS source file (national_M2025_dl.xlsx), a MANIFEST.json recording each file's real SHA-256 checksum and source provenance (including the note that the May 2025 panel overlapped the Oct-Nov 2025 federal shutdown), the mapping-corrections record documenting every SOC rename/remap/flag decision, and a corrected-BLS dataset (CSV and XLSX) emitted by the ingestion script. This directory is populated; the national displayed wage figures are backed by this verified snapshot.

The license on archived snapshots is public domain (BLS data is U.S. Federal Government work under 17 U.S.C. section 105). Snapshots may be freely copied, distributed, and used. The refresh policy is to add new dated snapshot directories when new authoritative source data is released, never to overwrite or delete previous snapshots. The next BLS OEWS release is scheduled for May 15, 2026 at 10:00 AM ET (May 2025 estimates), at which point a new snapshot directory _data_archive/YYYY-MM-DD_bls_oews_may2025/ would be added.

Verification procedure: the snapshot directory's MANIFEST.json includes SHA-256 checksums for every file, so a reader can run sha256sum on each file and compare against the manifest to confirm the snapshot has not been altered. This now applies to the real national BLS pull; the regional figures remain illustrative pending Phase 2.

Large-file handling: the metropolitan-area source, served split by state.

One archived source, the BLS OEWS May 2025 metropolitan-area file (MSA_M2025_dl.xlsx, roughly 29 MiB), exceeds the 25 MiB per-file limit of the platform's web host and therefore cannot be served whole on the live site. To keep the real data available and easy to use, it is provided on the site split by primary state, one native Excel file per state (plus DC and Puerto Rico), under _data_archive/2026-07-10_bls_oews_may2025/MSA_by_state/. The split is a deterministic partition by the state column: the header is replicated in each file, empty cells are preserved, and no value is cut, changed, aggregated, or rounded. Reassembling the shards reproduces the original dataset exactly (all 150,023 rows), and this was verified at creation. The complete, byte-for-byte original file is preserved in every downloadable platform package, and its SHA-256 checksum in the snapshot manifest lets anyone confirm it against the figures on the site. In short, this is a hosting accommodation only: the source of truth is unchanged, preserved in full, and cryptographically verifiable.

11. Closing

Real BLS data is freely available, has an annual publication schedule that aligns with the platform's proposed recalibration cadence, and is now the production data source for the national wage floor figures. The current release integrates real BLS national 25th-percentile data (read directly from the XLSX, via a re-runnable ingestion script) alongside single-factor regional approximations, with explicit per-row source attribution so readers can distinguish real from approximate. Phase 2 (real occupation x MSA regional data) is the remaining step to a fully BLS-sourced calculator.

Robertson, J. (2026). Data Sources and Methodology (BLS Attribution Reference). We The People Platform (Version 3.7.746). https://wethepeopleplatform.com/_web_html/06_Presentation_Materials/06_Data_Sources_And_Methodology.html

Robertson, Jason. 2026. "Data Sources and Methodology (BLS Attribution Reference)." We The People Platform v3.7.746. https://wethepeopleplatform.com/_web_html/06_Presentation_Materials/06_Data_Sources_And_Methodology.html.

@misc{wtpp_2026_119_data_sources_and_methodology_bls_attribu,
  author    = {Robertson, Jason},
  title     = {Data Sources and Methodology (BLS Attribution Reference)},
  year      = {2026},
  publisher = {We The People Platform},
  version   = {3.7.746},
  url       = {https://wethepeopleplatform.com/_web_html/06_Presentation_Materials/06_Data_Sources_And_Methodology.html},
  note      = {Document 119 of 139}
}