Population (BEA midyear estimate): methodology
Provenance for the series on Population (BEA midyear estimate). This page has its own URL so it can be cited alongside a figure.
- Dataset identifier
- bea_population
- Update cadence
- Annual, ~11 month lag
- Unit
- persons
- Coverage
- 18 series · 1969-01-01 to 2024-01-01
- Vintage published here
- 2026-10-30
Midyear (1 July) population by county, from the BEA Regional accounts (table CAINC1, line 2), beginning in 1969.
This is the denominator, published for that purpose. BEA carries it so its per-capita series can be reproduced, and it originates with the Census Bureau's population estimates rather than with BEA. Where we also publish the Census Population Estimates Program series directly, prefer that one as the population source; this one is here so that income ÷ population reproduces BEA's own per-capita figure exactly.
Revisions. BEA restates its regional estimates every year, and the restatement reaches back: a figure for 2019 is not the figure that was published in 2019. We do not overwrite what came before — each run stores its values under a new vintage, so a number cited last year stays retrievable after BEA revises it. This is the dataset the vintage model exists for.
Nominal dollars. Published as current dollars for the year in question, with no deflation and no inflation adjustment of ours. Comparing 1969 to 2024 directly compares two different dollars.
Revisions and vintages
Every observation we store carries the date it was published, and a revision lands as a new vintage rather than overwriting the old one. The figures on this site are the most recent vintage, stamped above; a figure you cited earlier stays retrievable even after the source restates it.
The full series, including the flag and vintage of every value, is in the CSV.