Based on ADP-Stanford Digital Economy Lab series for October.
Figure 1: Private nonfarm payroll employment (bold black), nowcast based on error correction model using ADP-Stanford Digital Economy Lab series (blue), +/- 1 standard error lines (gray), Bloomberg consensus as of 10/31 (red square), all in 000’s, s.a. Source: BLS, ADP via FRED, and author’s calculations.
The ECM used is estimated over 2021M07-2024M09, and has an adjusted R2 of 0.67, SER of 0.002. Using a simple regression of first log difference BLS on contemporaneous ADP series yields a virtually identical nowcast.
Note that this is a mechanical nowcast, based solely on past BLS data and ADP data. There is no accounting for strikes and hurricanes. Goldman Sachs (Walker/Rindels) notes:
ADP’s measure of employment growth has historically been less sensitive to natural disasters and Homebase’s data likely misses the impact of the current strikes because its sample is primarily made up of owner-operator-managed food and beverage, retail, and service businesses.
What seems to have added complexity to making forecasts for October nonfarm payroll is that the number of jobs declined by 112K from the original September report to the revised September report. Certainly, affected my forecasts. Difficult to make accurate forecasts with such large revisions.
The largest negative revisions were reported for Professional & Business Services (-69k), and Leisure & Hospitality (-82k). The two big upward revisions were reported for Transportation & Utilities(22K) and Education & Health Services (21k).