Employment in January: Benchmark Revisions, Seasonality, Population Controls

The Employment Situation release for January 2024 incorporated annual benchmark revisions to establishment survey series, and reported population controls for the household survey series. NFP at +353 vs. +187 thousands Bloomberg consensus. (Last year at this time, NFP again surprised, at +517 thousand exceeding the Bloomberg consensus of +115 thousand.)

Here’s nonfarm payroll employment according to several measures.

Figure 1: Nonfarm payroll employment from January 2024 incorporating benchmark revision (bold red), NFP from December 2023 (light red), and author’s calculation of implied benchmark revision using preliminary benchmark for March 2023 (blue), implied level from preliminary benchmark by author and Bloomberg consensus change for January (light blue square), Philadelphia Fed early benchmark (light green), and household survey series on civilian employment adjusted to NFP concept as reported (pink), and for January 2024, adjusted household series removing effect of revised population controls (purple square). Adjustment to remove revised population controls assumes overall civilian employment increases 0.957 of total control for all civilian employment. Source: BLS (January 2024, December 2023 releases), author’s calculationsBLS.

Surprising (to me) is that the benchmark revised series far exceeded the preliminary benchmark revision I calculated (blue line), and the early benchmark calculated by the Philadelphia Fed (light green line).

The CPS employment series adjusted to the NFP concept (a research series, not official), shows a sharp decline, mirroring the decline in the civilian employment series. A similar decline occurred a year ago, with the implementation of new population controls. I attempted to get a feeling for what impact new population controls might have, by using the ratio of adjusted employment to total employment (about 0.957) to adjust the -270 thousands coming from the January population control. This shows up as the purple square in Figure 1. Hence, unlike last year, the implied change is not erased. That being said, it is important to recall that the population survey series incorporate much larger sampling error than the establishment survey series. For tracking business cycle movements, the household survey based  series are typically less informative than the establishment survey based series.

The January seasonal effect is quite large, typically. Hence, it is of interest to consider how big an impact a possible mis-estimation of the seasonal component might have. One way to see this is to compare the 12 month change in the seasonally adjusted and not seasonally adjusted series. This is shown in Figure 2.

Figure 2: Twelve month change in nonfarm payroll employment, January release (bold red), from not seasonally adjusted, January release (lilac), and December 2023 (light red), all calculated in log differences. Source: BLS and author’s calculations.

Note the y/y growth rates from the seasonally adjusted and seasonally adjusted series for January 2024 do not differ a lot. On the other hand, the seasonally adjusted series for January 2024 vs. December 2023 do differ substantially.



4 thoughts on “Employment in January: Benchmark Revisions, Seasonality, Population Controls

  1. Macroduck

    Retail trade added 45,000 jobs, well above the recent average monthly gain. That’s probably a seasonal distortion. The hiring diffusion index was up at 65.6 vs 65.0 in December and 52.4 in November; the breadth of job gains is one reason the November and December gains were so large.

    The workweek shortened, but average hourly earnings rose by 0.6%, so that the payroll index still rose by 0.2%. That’s The smallest 1-month gain in this expansion.The payroll index is well above its pre-recession level and right about on its pre-recession trend, after inflation adjustment:


    The is true for non-supervisory and production workers:


    One oddity for which I’d appreciate any insight y’all have. What’s going on with production and non-supervisory jobs as a share of the total?:


    Big changes in trend oventime, big drop durin the Covi recession which hasn’t been fully reversed. Ideas?

    1. Macroduck

      By the way, a BLS economist has explained that a weathwr-induced drop in hours at the low end of the pay scale explains both the strength of hourly earnings and the weakness of the workweek in January.

      Since two of the three inputs to the payroll index were distorted by weather, you can pretty much ignore what I wrote about the index in January. What I wrote about level and trend, on the other hand, are still true.

      1. Moses Herzog

        Goolsbee was saying on PBS that one of the reasons the numbers seemed so high is because many of the jobs were part-time jobs. So it still gets back to people being “employed” and counting in “raising the employment rate” when they are lucky if they even have a living wage. I haven’t looked that granularly (is that a word??) at the numbers, but it strikes me as ringing true and I don’t think Goolsbee is a bullshitter for the most part.

  2. Moses Herzog

    If you want to ruin your children’s future, make them largely illiterate, you can move them to the state of Oklahoma, where personnel decisions are made not by the quality of the classroom teaching and learning environment, but by what teachers do on their own personal time away from the students.

    This school district had to spend $65,000 on security because the state superintendent, Ryan Walters, incited violence against a teacher/principal for what he does on his own personal time. $65,000 that could have been spent on students directly, or a higher teacher to students ratio. Walters is purely grandstanding here for the Okie illiterates so he can run for some state office at a near future date. Republicans say they don’t want government dictating their lives. What is it exactly when an appointed leader (not democratically elected) tells teachers they will be fired or not hired because of what they choose to do on their own personal time, away from the student body?? What would that be labeled if not government treading on their lives??

    Someone needs to go deer hunting together with Ryan Walters in the woods and have an accident.

Comments are closed.