[source] Answer: Because, by virtue of the Executive Office of the President getting the release numbers the night before, he knew the numbers were going to be pretty bad (+22K < +75K Bloomberg consensus). (By the way, if I as a EOP staff member gave away market moving news before the official release, I would’ve been fired, and prosecuted. Just sayin’).

Let me show how bad the data look, along a variety of indicators.

Figure 1: Nonfarm payroll employment from CES (bold black), Bloomberg consensus estimate (pink +), estimated preliminary benchmark revision based on Wells Fargo estimate (light blue), CPS data adjusted to NFP concept (tan), and covered QCEW employment, seasonally adjusted by author using X-13 (green), all in 000’s. Estimated preliminary benchmark revision using midpoint of Wells Fargo estimate of revision of -475K to -790K. Source: BLS via FRED, BLS, Bloomberg, Wells Fargo, and author’s calculations.

How do the growth rates compare? Below is a picture of q/q annualized growth rates (except for QCEW and CPS series, which are y/y).

Figure 2: Quarter-on-Quarter annualized growth rate of Nonfarm payroll employment from CES (bold black), Bloomberg consensus estimate (pink +), estimated preliminary benchmark revision based on Wells Fargo estimate (light blue), CPS data adjusted to NFP concept, year-on-year (tan), and covered QCEW employment, seasonally adjusted using X-13 by author, year-on-year (green), all in 000’s. Estimated preliminary benchmark revision using midpoint of Wells Fargo estimate of revision of -475K to -790K. Source: BLS via FRED, BLS, Bloomberg, Wells Fargo, and author’s calculations.

The general message from the picture is that these various measures of employment growth are all decelerating.

As noted in my previous post on employment and the business cycle, the mean absolute revision for the BLS CES measure going from first to third release is 40K (2022-24). That means a 95% confidence interval would encompass -62 to +102.