March 28, 2023

Is the Establishment Series Overestimating NFP Employment?

That is the proposition forwarded by Kevin Drum at Jabberwocking. Verbatim:
So: have we truly developed 2.5 million new tasks considering that March? Im unsure I believe that.
The chart he published summarizes his argument succinctly:

Source: Drum (2022 ).
The stylized fact is that the home series includes almost nothing to the information included in the establishment series relating to the cyclical behavior or employment, a point made by practictioners (see Furman, 2016, or CEA, 2017). Nevertheless, the possibility of mismeasurement, particularly in these times, remains. Some idea of mismeasurement is offered by the habits of not-seasonally-adjusted nonfarm payroll (NFP) work, and overall work covered by the Quarterly Census of Employment and Wages (QCEW). For a range of factors relating to staff members coverage, the numbers differ, however the total trends need to supply some insight.

Figure 2: Change in work from NFP (black), QCEW changed using Census X-11 seasonal modification by author (dark blue), QCEW using multiplicative moving typical differences by author (sky blue), and utilizing home work series gotten used to NFP concept (red), all in 1000s, relative to 2022M03. Source: BLS by means of FRED, Author, bls, and blss calculations.
The research series on home employment adapted to NFP principle is not an official series, and is described here. A caution: considered that the household employment series is based upon a fairly little sample, the variability related to it– and thus by extension its derivative series– is going to be fairly big.
Through June, its unclear whether NFP undercounts or overcounts relative to the QCEW number provided the differing results provided by the seasonal modification procedure. The adjusted family series shows that through November, only about 800 thousand NFP jobs were created vs. 2700 thousand from the official NFP series.
One method to get around the seasonal change issues (which have actually been compounded by the severe drop associated with the pandemic) is to look at 12 month modifications. This certainly downweights current developments, however a minimum of we get a take a look at the patterns.

Figure 3: 12 month modification in employment from not seasonally changed NFP (black), QCEW (dark blue), and household work series adapted to NFP principle (red), all in 1000s, relative to 2022M03. Source: BLS via FRED, BLS, author, and blss estimations.
All three series match up pretty well through 2022M06. Checking out up to 2022M11, both the NFP and changed home series reveal declines. Which one to believe? As noted earlier, Furman, 2016 and CEA, 2017 location practically overall weight on the facility series. A more formal analysis showing practically overall weight on preliminary establishment series for real time analysis is provided by Goto, Jacobs, Sinclair and van Norder (2021 ). (As an aside, if I look at the last information for NFP and the adjusted home series, statistically the changed family series seems to respond to spaces between the 2 series, while the NFP series does not; formally, the adjusted series reacts to an error correction term.).
So, for now, while I see it as an open concern what (NFP) employment development is, I think (1) it is likely positive, and (2) likely decreasing in pace.

All three series match up quite well through 2022M06. Checking out up to 2022M11, both the NFP and adjusted household series show decreases. As kept in mind previously, Furman, 2016 and CEA, 2017 place practically total weight on the facility series. A more formal analysis revealing nearly total weight on initial facility series for genuine time analysis is offered by Goto, Jacobs, Sinclair and van Norder (2021 ). (As an aside, if I look at the last information for NFP and the adjusted household series, statistically the changed household series appears to react to gaps between the 2 series, while the NFP series does not; officially, the adjusted series reacts to an error correction term.).

Figure 1: Log ratio of nsa NFP to QCEW (blue), and 2001-19 average (red dashed line). NBER specified peak-to-trough economic downturn dates shaded gray. Source: BLS by means of FRED, Author, bls, and nbers calculations..
While the establishment survey is exactly that– a study–, the QCEW is a census which must in principle catch all covered work, by method of tax records, and so on. This data, amongst others, is used to standard the facility series (see discussion here). This suggest to me that as of June, the nsa NFP series was not method off.
GIven this, what make from net work growth– measured in various ways– given that March 2022 (when the last benchmark month was)? In addition to the official BLS facility NFP series, the QCEW series (seasonally adjusted by me) and the home series gotten used to the NFP idea are displayed in Figure 2.

Leave a Reply

Your email address will not be published. Required fields are marked *