The monthly reports the Bureau of Labor Statistics puts out are based on surveys of businesses and households. Those aren’t particularly accurate. Every state has records on unemployment taxes and payouts and on a quarterly basis the BLS tries to reconcile its numbers with these more accurate ones. The annual revisions resulting from this true up have been large negatives to the job numbers in the last couple of years. The one for 2026 is expected to be positive for various reasons. The preliminary revision recently released, however, shows a small negative to the so far official numbers, 79,000. But these preliminary revisions are usually too low, so an overall increase is still expected when the final revision is done in January. Unfortunately, the private sector is forecast to have a bigger downward revision, and government jobs an upward revision. I know the BLS is working on its methodology but I don’t understand why they do the monthly surveys at all. They should either make participation in the surveys mandatory or they should figure out how to just use the state data. One issue with the state data is it may not have all the business classifications that BLS uses, but it is more accurate. (BLS Release)
