🪲 BugOut Index

Data revisions tracker

How much do the headlines change after the fact?

Most economic indicators are revised after their first release. This page shows two lenses on that for the U.S. labor market: monthly revisions to nonfarm payrolls (PAYEMS) and the annual benchmark revision to the unemployment rate (UNRATE). Lower unemployment isn't necessarily lower revision risk — the point here is to see whether the number you read on day one survives contact with later data.

Nonfarm payrolls revisions, last 24 months

Each observation month plotted three times: as first published (blue), one month later (orange), and two months later (green). Source: ALFRED (PAYEMS).
no data yet
Show PAYEMS table
Observation Initial release Initial value +1 month +2 months Net revision (+2 − initial)

Unemployment rate — before vs after the latest annual benchmark

UNRATE is rarely revised between releases, but each January's Employment Situation incorporates an annual benchmark that can shift up to five years of history. Blue is today's series; orange is the same series as it stood the day before the most recent benchmark release. Source: ALFRED (UNRATE).
no data yet

How to read this page

“Initial release” means the value the day BLS first published it. “+1 month” and “+2 months” are the values as they stood after the next two scheduled employment reports. Larger gaps between the lines indicate that the headline number moved meaningfully as additional survey responses came in. Small gaps don't prove the data is right — only that it didn't change much within the monthly revision window. Annual benchmark revisions can still rewrite older months beyond what this chart shows.