The BLS statistics do not include state and local government workers or railroad workers. Census BITS statistics exclude:

• crop and animal production (NAICS 111, 112)
• rail transportation (NAICS 482)
• National Postal Service (491)
• pension, health, welfare, and vacation funds (NAICS 525110, 525120, 525190)
• trusts, estates, and agency accounts (NAICS 525920)
• private households (NAICS 814)
• public administration (NAICS 92)

2. NETS counts sole proprietors and partners as employees

Census data doesn’t register owners and owning partners of establishments as employees. BLS data also excludes owners and owning partners – and BLS data also excludes employee’s exempt from unemployment insurance laws. NETS data includes all of the above. Thus, NETS registers establishment births for many new proprietorships and partnerships that are excluded from BD and BITS. In addition, when "nonemployer" proprietorships or partnerships die, NETS registers employee losses while BITS and BD statistics do not.

3. NETS counts births and deaths from the first to the last employee

It appears that BITS statistics register an establishment birth when an establishment owner without non-owning employees hires the first one and moves from the Census Bureau's Nonemployer Statistics database to its BITS database. Similarly, apparently BITS registers an establishment death any time the owner lays off the last non-owning employee, even if the establishment remains in business. Since the Census statistics are annual, an establishment can only appear to be born or die once per year, but it could happen year after year.

4. NETS is dynamic through time

NETS releases use information from several current year D&B databases to better track and confirm changes that happened to establishments over time. This process of looking forward combined with a process that looks back across all necessary years to adjust establishment information that may have changed or has been discovered recently. These back and forth adjustments provide a much more complete picture of establishment changes from the beginning of the NETS data (1990) through the end of the current year.

The BLS Business Employment Dynamics (BD) statistics show different job volatility than either Census Business Information Tracking Series (BITS) or NETS data because the quarterly BLS BD statistics are based on the monthly Census of Employment and Wages (CEW) data, which are taken from state unemployment insurance reports, and they reflect the arrivals and departures of individual workers. The Census BITS data is more like an annual snapshot showing the numbers of job positions at the reported establishments.

• For example, suppose state unemployment insurance reports show that employee A leaves firm 1 (job destruction) and goes to a higher paying job at firm 2 (job expansion) which was open because employee B left (job contraction) to go to higher paying job at firm 3 (job expansion) which was open because employee C retired (job contraction) and, finally, a new entrant to the work force filled the original opening at firm 1 (job expansion). Assuming that all of these changes happen in the same quarter, the quarterly BLS job creations would be 3 and the quarterly BLS job destructions would be 3. If the same thing happen in every quarter at these firms, the total for the year (sum across quarters) would be 12 job creations and 12 job destructions (by the BLS definitions). In the NETS and Census databases, this scenario would normally create zero job creations and destructions because the annual total of jobs at these three establishments is the same in the two contiguous years.

For more information, see www.census.gov/epcd/susb/introusb.htm and www.bls.gov/cew/peoplebox.htm#4 for more detailed definitions regarding their data sets.