Understanding the Non-Farm Payroll Data

Non-Farm Payroll ? Non-Fathomable Payroll? Non-Faltering Payroll?

What does this article intend to achieve? One of two objectives: Either put you to sleep OR help you understand:

  • Monthly Non-Farm Payroll
  • Monthly Revisions to Non-Farm Payroll
  • The BLS’ Annual Benchmarking Exercise
  • Quarterly Census of Employment and Wages
  • The Philadelphia Fed’s effort to forecast the BLS’Annual Benchmarking number

Non-Farm Payroll is arguably the most important economic data point in the US, or probably the next most important data point to the CPI.

We often tune in to just the first estimate of the NFP release for the month and then conveniently ignore any revisions to that number.

Needless to say, often it is the first estimate which is most important from an immediate market impact perspective given the reliance placed on it by trigger happy market participants and automated algorithms. Nonetheless, it is important to understand all subsequent revisions if an investor wants to get a good view of the employment situation in the US to make their own informed investing decisions.

Lets start from the beginning….

Monthly Non-Farm Payroll

The Non-Farm Payroll report is published by the Bureau of Labour Statistics on the first Friday of the following month. This news release presents statistics from two major surveys, the Current Population Survey (CPS or Household Survey) and the Current Employment Statistics survey (CES or Establishment Survey). The Household Survey provides information on the labor force, employment, and unemployment that appears in the “A” tables, marked “Household Data”. It is a sample survey of about 60,000 eligible households. The Establishment Survey provides information on employment, hours, and earnings of employees on nonfarm payrolls; the data appear in the “B” tables, marked “Establishment Data”. Each month the CES program surveys about 122,000 businesses and government agencies, representing approximately 666,000 individual worksites.

The below is the picture that the Establishment Survey’s first estimates showed us for 2023. Trending down for sure, but still a healthy pace of job creation compared to recent labour market history. Average monthly jobs created of 310K.

Monthly Revisions to Non-Farm Payroll

The Establishment Survey revises these published estimates to improve its data series by incorporating additional information that was not available at the time of the initial publication of the estimates. The establishment survey revises its initial monthly estimates twice, in the immediately succeeding 2 months, to incorporate additional sample receipts from respondents in the survey and recalculated seasonal adjustment factors. The below is the picture that the recent revisions of the Establishment Survey shows us. Average monthly jobs created in the first 5 months of 2023 of 255K – lower by 55K on average or 325K in total. That sounds like a sizeable over-estimation of overall job creation in the economy!

The BLS’ Annual Benchmarking Exercise

Ok. So what happens next? The sample-based estimates from this Establishment Survey are adjusted once a year (on a lagged basis) to universe counts of payroll employment obtained from administrative records of the unemployment insurance program. The difference between the March sample-based employment estimates and the March universe counts is known as a benchmark revision, and serves as a rough proxy for total survey error. Over the past decade, absolute benchmark revisions for total nonfarm employment have averaged 0.1 percent, with a range from -0.3 percent to 0.3 percent. In summary, not only are the initial estimates revised twice in subsequent months, they are also revised retrospectively once a year, typically in February of the next year. Too confusing ? Lets understand this with an example.

The first chart below shows 2022 monthly job creation based on initial estimates and then the 2nd monthly revision. The second chart below shows 2022 monthly job creation based on the 2nd monthly revision and the final benchmark revision conducted in February 2023.

Both the charts collectively have the following takeaways:

  1. Total jobs created in 2022 per the initial estimates of the Establishment Survey were 4.62mn
  2. Total jobs created in 2022 post the 2nd monthly revision was more or less the same (you can see not a lot of white bars in the first chart)
  3. However, total jobs created in 2022, after the benchmarking exercise was completed in early 2023, were 240K lower at 4.79mn (more orange bars in chart 2 compared to the white bars in chart 1)

So, how are we doing in 2023? Well, we have to wait till Feb 2024 to really find out. However, the BLS also provides some preliminary idea earlier in the year about the likely revision that will come up next year.

On August 23rd, 2023, the BLS’ published a preliminary estimate suggesting that US job growth through 2023 March was less robust than previously estimated. The preliminary estimate of the benchmark revision indicated a downward adjustment to March 2023 total nonfarm employment of  minus 306,000 (−0.2 percent).

Quarterly Census of Employment and Wages

The monthly Establishment Survey data is timely, but inaccurate. The final data post the completion of the benchmarking exercise in February is the most accurate, but not timely (i.e. with a year’s lag). So what’s the solution? Enter Quarterly Census of Employment and Wages. The QCEW program publishes a quarterly count of employment and wages reported by employers covering more than 95 percent of U.S. jobs available at the county, Metropolitan Statistical Area (MSA), state and national levels by detailed industry. Unlike the monthly Establishment Survey data, the QCEW data are based on state unemployment insurance tax records and cover nearly all of US jobs. BLS publishes data from the QCEW program every quarter in the County Employment and Wages press release, within 5 months after the end of each referenced quarter. So in summary, instead of waiting for 12 months to get the full revised job picture, you can look at the QCEW data which comes with a lag of only 5 months !

For instance, the latest QCEW (released in August 2023) showed that in March 2023, national employment increased to 151.4 million, a 2.5% increase over the year. In contrast, the existing data from the BLS survey shows jobs increased 2.7% y-o-y in March 2023. (Side Note : Even the annual benchmarking exercise on the Establishment Survey data is based on the QCEW data)

The Philadelphia Fed’s effort to forecast the BLS’Annual Benchmarking number

Lastly, for those like me with a sadistic desire to inflict unending pain on yourself, you can read the Federal Bank of Philadelphia’s release which solves the lagged data problem even further. It produces quarterly early benchmark estimates of monthly state employment for all 50 states and the District of Columbia. In other words, since the QCEW data are available quarterly, the Philly Fed researchers are able to create their own early benchmark estimates on a more timely basis.

The latest report for Q4 2022 issued by the Philly Fed shows US job creation in the year through December 2022 was estimated to be 2.7% and in the year through March 2023 is estimated to be 2.3%. Calculated differently, the Philly Fed report shows they annual benchmark revision for March 2023 (that the BLS will release in Feb 2024) is likely to be around minus 550K.

Summary: One of the most prominent and sometimes unfortunate aspects of the investing world is “Information Asymmetry”! And the Retail and Small Institutional Investor is always on the disadvantaged side. So the next time you see a Bloomberg news run like the below….

…. check the data for yourself. Make sure you understand the meaning of every key economic statistic. Be an Informed Investor!

Shashank Sawant


Leave a Comment

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