and thus are subject to several types of error including sampling error. A big reason the data are revised the following month is the time frame for which employers to report their data to be included in the current month's sample is extremely tight. Many employers are unable to report the data whithin that time frame, so their data is not included until the revised estimates. Most times the revisions are not real large, but occasionally, particularly around turning points in the economy, the revisions can become larger because all of the data becomes much more volatile.
Another reason for larger monthly revisions is that the seasonal adjustment process is dynamic which means it factors in all of the data up to the current month. Because of that, the actual revised estimate the following month may be the same as the original estimate, but the reported level could change solely due to the seasonal adjustment changing. I can assure you this is a perfectly normal effect of dynamic seasonal adjustment and typically makes the data more accurate.
Finally, even at their worst, the BLS data are extremely accurate. Last year, the benchmark revision was -589,000 jobs. That may sound like a lot, but in reality represents a revision of less than 0.5%.
= new reply since forum marked as read