Track US Inflation with Official BLS CPI Data

Inflation dashboards often fail at the source-model boundary. A CPI observation is an index level for a specific period, not a precomputed year-over-year percentage, and different BLS series represent different populations, regions, item groups, and seasonal adjustments. The BLS Labor Statistics Scraper exports the official observations without hiding those distinctions.

Collect the index before calculating rates

Headline CPI for all urban consumers uses series CUUR0000SA0. A compact current window is enough for monthly and annual comparisons:

{"seriesIds":["CUUR0000SA0"],"startYear":2024,"endYear":2026,"maxResultsPerSeries":36}

Sort by year and BLS period, exclude annual-average period codes from monthly calculations, then compute month-over-month or year-over-year changes downstream. Keeping the raw value makes the transformation reviewable. Store the series ID and source URL with every derived metric so a reader can trace a chart back to BLS.

A reliable scheduled pipeline should upsert on series_id + year + period, record ingestion time, and flag changes to a previously stored observation. Do not replace missing months with zero. Missing data can mean a query error, an unpublished period, or a series that is no longer active.

For category-level inflation, create separate Tasks for food, shelter, energy, or regional series rather than mixing incompatible indexes into one average. The same pattern supports wage-adjusted analysis: join CPI to an earnings series on month, then state whether values are nominal or real. That small documentation step prevents a technically correct dataset from producing a misleading business conclusion.