Incorporation of employment records from the BLS’s Quarterly Census of Employment and Wages (QCEW) drives the revision process. The BLS collects QCEW data from a nearly full count of the nation’s employers (about 9.2 million), and the CES maintains a sample size of about 500,000 establishments.
Overall, the benchmark revision process affords the BLS an annual opportunity to improve its estimates by addressing the following five key issues:
- Sampling Errors. Typically, an additional year of population-based QCEW data replaces the sample- based CES estimates.
- Coverage Differences. The revision process reconciles small differences in the coverage of industrial sectors by the CES and QCEW programs.
- Administrative Changes. Revisions correct errors, add late data, and reclassify firm characteristics (e.g., industrial sector or location).
- Firm Births and Deaths. QCEW data capture jobs associated with firm births and deaths that the CES sample misses. The BLS reestimates a birth/death model for the ongoing CES estimates.
- Seasonal Adjustments. Seasonal factors are reestimated during the annual benchmark process.