in a Post from Sunday, I pointed out that non-farm payrolls may be overestimated, but the trend is generally consistent. Noting how the Business Employment Dynamics (BED) and Quarterly Census of Employment and Wages (QCEW) data provide divergent signals, Standard Chartered’s Steven Englander calculates how much there may be overstatement.
source: Steven Englander, “How Accurate is NFP”, Charter of Standards, 11 July 2023.
By the end of 2022, the difference between the estimates (a fit of lagged QCEW and NFP one year using NFP for 2012-2023) is about 500,000. He observed that a business’s life-and-death model could be the source of the variance.
source: Steven Englander, “How Accurate is NFP”, Charter of Standards, 11 July 2023.
Interestingly, the 500,000 is very close to the 616,000 reported by the Philly Fed’s earlier benchmark (the last observed estimate was 572,000 in April 2023). The image below shows the gap (the difference between the blue and red lines), taken from this post.
figure 1: CES nonfarm payrolls (blue), CPS household series adjusted for NFP concept (tan), QCEW coverage totals, seasonally adjusted using X-13 (green), all in 000s, sa source: BLS via FRED, BLS, BSL-QCEW, philadelphia fed.
CPS-based employment series are not affected by the errors produced by the firm birth and death model. Although the civilian employment series adjusted for the NFP concept is about 1.85 million lower than the actual NFP as of June 2023, there is a long-standing gap between the two series. Perhaps more interesting are the changes to the two series since 2021M06. NFP has reportedly gained 278,000 points more than the CPS derivative series.
So while there is some evidence that non-farm payrolls are growing at a slower rate than reported, a cumulative difference in the millions since mid-2022 seems unlikely.
