Every week and a half in the past, Kevin Drum famous earlier the disjuncture between job creation as measured by the family survey (civilian employment) and institution survey (nonfarm payroll employment). My conclusion was that the empirical proof steered placing most — if not all — weight on the institution survey. Now, Torsten Slok compares the declines in response charges within the two surveys. The comparability additional buttresses the case for utilizing the institution collection.
Slok writes:
Survey response charges for employment and inflation have declined considerably throughout the pandemic, and that is introducing substantial measurement errors and uncertainty, see chart beneath. One such instance is the widening hole between the institution survey and the family survey within the employment report, with the institution survey exhibiting 2.6 mn jobs created over the previous eight months and the family survey exhibiting no job development over the identical interval. …
Right here’s the determine (the place I’ve added a pink line at March 2022, 8 months in the past).
Supply: T. Slok, 13 December 2022. Crimson line at March 2022.
Be aware that the family response fee relative to institution fee drops precipitously beginning round October of final yr. The hole yawns extra quickly beginning after March. And right here’s the image of the hole between nonfarm payroll employment and civilian employment adjusted to the NFP idea (it is a analysis collection).
Determine 1: Nonfarm payroll employment official (sky blue), nonfarm payroll employment implied by preliminary benchmark (darkish blue), and civilian employment adjusted to NFP idea (inexperienced), all seasonally adjusted, in 000’s. Figures in (parentheses) are web new jobs in thousands and thousands, since 2022M03 (indicated by pink dashed line). Supply: BLS by way of FRED, and creator’s calculations.
The hole in numbers widens precisely on the time the family response fee drops.
So: (1) I put extra weight on the institution collection, and (2) the hole between the 2 collection is extra probably resulting from rising, and biased, measurement error within the family collection, reasonably than, as an illustration, primarily will increase in multiple-job holders.
