The percentage change is less than that for the CES estimate. Does this mean there are a lot fewer people employed that indicated by the official series (e.g. here)? Here’s data from 2023M06 onward:
Figure 1: Change in total covered employment in Quarterly Census of Employment and Wages (QCEW), not seasonally adjusted (blue), official BLS nonfarm payroll employment (NFP) (red), early benchmark NFP (tan), and preliminary benchmark implied (green), all relative to 2023M06. Source: BLS, BLS via FRED, Philadelphia Fed, and author’s calculations.
While typically it’s assumed the preliminary benchmark will be more accurate given the use of tax data, to the extent there is a substantial component of undocumented workers not having tax data reported on them, it might be the case that both preliminary benchmark and QCEW will be understating employment growth.
The shortfall in the QCEW vs. nonfarm payrolls appears to be centered on two months; namely, mainly July 2023, and to a lesser extent January of this year.
Through June of last year, as revised the YoY% change in the two series track rather closely. But in July suddenly the YoY% change in the QCEW is 0.4% lower than in payrolls. Though December it continues in that range of 0.3%-0.5%, suggesting the monthly numbers for August through December were close.
Then in January of this year the YoY% change in the QCEW is 0.5% less than in payrolls, and continues in the 0.4%-0.6% range through May.
That suggests to me that there was a big miss in the NSA payrolls estimate in July of last year, on the order of 470,000, which would translate to roughly a 240,000 miss seasonally adjusted – which means July of last year may have actually seen a SA decline of 60,000 jobs. The miss in January of this year may have been about 150,000 NSA, translating into about 110,000 SA. As I wrote above, aside from that the monthly numbers were probably pretty close.
I am a little skeptical of the “immigration” take on this, because advocates always tell them to get tax I.d. Numbers even if they are illegal; and I’ve read somewhere that the best estimate is that about 90% of them do so.
But the bottom line is that, unless there are significant revisions to the preliminary numbers in the QCEW (which definitely happened in 2022), employment has actually been a little weaker than we thought.