Every month the BLS releases the Employment Situation Report, which reports on results from two monthly surveys, Current Employment Statistics (CES) and Current Population Statistics (CPS). The CES data disseminated in the Employment Situation Report is commonly used among analysts as a leading economic indicator, and is usually what is referenced by news stories talking about monthly job growth or loss data from the BLS.
CES is a monthly survey of about 142,000 businesses in the US (coming to about 689,000 physical locations and 1/3 of employed workers in the US). It is not a census (a complete and exhaustive accounting of workers in the US). Because it is not a census and is subject to sampling error, the BLS adjusts historical CES estimates twice a year, bringing them in line with more lagging government-collected administrative business unemployment insurance records that cover 97% of the employed workforce in the United States. These biannual adjustments to historical CES are often the impetus for news stories lauding or bemoaning the true state of employment in the US.
Emsi does not use CES data in calculating employment, in part because CES is a survey subject to a lot of sampling error. Rather, Emsi bases employment counts on BLS Quarterly Census of Employment and Wages (QCEW) data, which is based on the unemployment insurance records mentioned above that employers are required to submit regularly, and that are used to benchmark CES data twice a year. The result is that Emsi data is entirely unaffected by changes in CES and resultant BLS benchmarking. Rather, Emsi’s QCEW-based employment data is constantly roughly equivalent to the employment figures to which historical CES data is or will be adjusted.