Bureau of Labor Statistics

What You're Getting

BLS publishes the official US labor market data: CPI, PPI, unemployment (CPS), job openings (JOLTS), payrolls (CES), hourly earnings, and occupational employment statistics (OES). Data is organized by series ID — a structured code encoding survey, seasonality, geography, and metric (e.g. LNS14000000 = national unemployment rate SA). Series return monthly or quarterly observations with year, period, value, and footnotes.

Ingestion Strategy

Use the Public Data API v2 — register for a key to get 500 req/day (unregistered: 25/day). Batch requests support up to 50 series per call, so group your series list into chunks of 50.

import requests, json

headers = {"Content-type": "application/json"}
payload = {
    "seriesid": ["LNS14000000", "CES0000000001", "CUUR0000SA0"],
    "startyear": "2015",
    "endyear": "2024",
    "registrationkey": "your_key"
}
r = requests.post("https://api.bls.gov/publicAPI/v2/timeseries/data/",
                  data=json.dumps(payload), headers=headers)
results = r.json()["Results"]["series"]

Rate Limits & API Details

• Registered key: 500 series/day, 50 series/request, 20 years of data/request
• Unregistered: 25 series/day, 10 series/request, 10 years
• Bulk flat files at download.bls.gov (FTP-style HTTPS) — full series dumps updated after each release

Schema Stability

Series IDs are highly stable — BLS rarely retires a series. Footnote codes signal data quality flags (preliminary, revised) — parse and store them. Seasonal adjustment flag is embedded in the series ID (second character). New series are added regularly; discovery via the Series ID structure documentation.

Data Quality Gotchas

• Preliminary vs final: payroll data is revised twice before becoming final — store footnotes field and re-pull recent periods after each release
• Seasonal adjustment: SA and NSA series coexist — be explicit in your series ID list
• Period codes: M01–M12 are months, Q01–Q04 are quarters, A01 is annual
• Geographic series IDs: metro-area series use different prefixes — check the BLS geographic series documentation

Tool Compatibility

• Python: bls package or direct API calls; pandas_datareader has BLS support
• R: blsAPI package
• Airbyte: no official connector; community scripts exist
• Bulk flat files (at download.bls.gov) are tab-delimited, easily loaded via Snowflake COPY or BigQuery load jobs

Benchmarking Approach

National unemployment rate (LNS14000000) should have monthly observations back to 1948 — expect 900+ rows. Total nonfarm payrolls (CES0000000001) back to 1939. After each monthly release (first Friday of month), verify new row appears within 24h. Spot-check against BLS press release headline number. Flag if value changes on re-pull of a period older than 3 months (indicates revision).