DIY Benchmark Employment Release (Establishment)

Friday’s estimate of nonfarm employment will incorporate a benchmark revision, based on unemployment insurance and tax data extending up to 2019Q3. What will we see? Here’s an informed guess, based on what we see now (see the green line in Figure 1).

(Click on image to enlarge)

Figure 1: Nonfarm payroll employment (bold blue), August 2019 preliminary benchmark revision for March 2019 NFP (dark blue square), implied NFP using Quarterly Census of Employment and Wages (QCEW) (bold green). Source: BLS via FRED, BLS and author’s calculations. Data used [xls]

The green line is what I guess it’ll be, using the seasonally adjusted QCEW’s historical relationship with nonfarm payroll employment. Here’s the raw material used to obtain this guess:

(Click on image to enlarge)

Figure 2: Nonfarm payroll employment (bold blue), Quarterly Census of Employment and Wages (QCEW) (brown). Source: BLS via FRED, BLS and author’s calculations. Data [xls]

Obviously the seasonality in the QCEW series makes interpolation by related series a little more complicated. I applied Census X-13 to the log QCEW series, but using a simple log-log regression with seasonal dummies will yield a similar result, particularly for the first six months of 2019 (which are the important ones).

Try it yourself! Nonfarm payroll, preliminary benchmark, and Quarterly Census data all here. More on how the BLS will “wedge-in” the revisions, and what an ARIMA on the fitted NFP looks like, in this post. (Surely I could obtain more efficient estimates, using a single step procedure instead of the two-step and three-step procedures I use in this post.)

(BLS will wedge in the revision for 2018 — this is my guess of how things will look (pink line)):

(Click on image to enlarge)