Hacker News Reader

They're just the results of surveys. They get revised twice after the first report, and there's also a yearly adjustment based on unemployment benefits numbers. Revisions of over 100,000 are not that uncommon. (500,000 is pretty rare though!) https://projects.fivethirtyeight.com/jobs-report-growth-unem...

View on HN

Surveys, yes, but also corroborated by interagency reports, many of which are behind schedule do to a prolonged federal shutdown in December 2018 and January 2019. Some offices have been behind schedule on reporting since returning to work in February.

> Revisions of over 100,000 are not that uncommon. (500,000 is pretty rare though!)

Those are monthly numbers. The 500k should probably be anchored to the last revision, I guess a year ago. (Also, I saw somewhere that this 500k represents 0.3% of the entire employee base.)

Reply on news.ycombinator.com