In order to maintain the accuracy of its data, BLS annually revises its payroll employment, hours, and earnings data to a new benchmark.
Monthly state employment data are estimates of job counts based on data from the Current Employment Statistics (CES) sample of nonfarm establishments. Once a year, these sample-based estimates are realigned to benchmarks, which are comprehensive counts of employment derived primarily from data reported in unemployment insurance (UI) tax reports.
BLS publishes revised data for the New England states with the release of January payroll employment figures. Benchmark revisions to regional employment data typically incorporate benchmarks from the prior March. Revised national data are typically released in February. For more information on these revisions, see BLS’s Employment and Earnings benchmark article (http://www.bls.gov/web/cesbmart.htm).
Similarly, unemployment data are benchmarked each year to align with annual averages of Current Population Survey (CPS) state estimates of labor force and unemployment; they also undergo other annual revisions such as the re-estimation of models and the incorporation of population controls.