I.e., Beware the footnotes. New population controls results in updated 2025 employment figures (but 2024 not updated). Much of the purported gap between NFP (establishment survey) and adjusted civilian employment (household survey) disappears as a consequence.

Figure 1: Civilian employment, December 2024 release (brown),  January 2025 release, incorporating new population controls (tan), and previous civilian employment adjusted to CBO estimates of immigration, by author (green), all in000’s, s.a. Source: BLS via FRED, CBO, and author’s calculations.

To see how I calculated the household civilian employment series adjusted to CBO immigration estimates, see here. In other words, a large portion of the worry (e.g., Heritage’s E.J. Antoni) about recession, based on the household series, disappears. Dr. Antoni makes no mention of this in his latest comment, arguing that future annual benchmark revisions will show even larger decreases in NFP according to Philly Fed early benchmark, and BED employment.

What about nonfarm payroll estimates? Here are the pre-preliminary implied benchmark revision (Dec ’24 with April-December actual changes included), the post-benchmark revision (Jan ’25), and the Philadelphia Fed’s early benchmark, compared against author’s estimated household series adjusted to CBO immigration.

Figure 2: Nonfarm payroll employment, January 2025 release (bold black), December 2024 release (blue), implied preliminary benchmark (sky blue), early benchmark (red), household employment series adjusted to NFP concept (light green), and adjusted to NFP concept, using CBO immigration estimates by author (green), all in 000’s, s.a. Source: BLS via FRED, BLS,  Philadelphia Fed, CBO, and author’s calculations.

Note that the January 2025 release is slightly above the implied preliminary benchmark, the early benchmark, and (as of June 2024) above the civilian employment series adjusted to NFP concept, but using CBO estimates of immigration. Taken together, the two pictures indicate that as of June 2024, about 3/4 or more of gap between establishment survey series and adjusted household survey series is accounted for by mis-estimation of population in the household survey, and the remainder by overcounting of employment in the establishment series.

Goldman Sachs (today and earlier) argues that the downward revision to NFP is overstated by about 300 thousand, as the revision incorporates exclusion of undocumented workers, since the estimate is based on QCEW data. The other 300 thousand they attributed to the overstatement by the firm birth-death model.