Sorry if the record really sounds like it’s broken (with my suggestion that those not familiar see some of my recent blog posts like Hartman (2026)), but again we have conflicting information from the jobs reports released by the U.S. Bureau of Labor Statistics (BLS). The number with the most attention in yesterday’s release, seasonally adjusted total nonfarm payroll employment, increased by my calculations using data from the U.S. Bureau of Labor Statistics (2026, a) by about 160,000 jobs from July to August 2026. That’s a very good number. However, at least one of the other recent estimates was much larger than that increase, while at least one number may have been particularly troubling.
The U.S. BLS surveys workplace establishments about the number of jobs that they have for total nonfarm payroll employment. It is not an unadulterated counting. My understanding is that total nonfarm employment is also based on the ‘birth death’ model when projecting out over the entire country. Given that (1) the response rate to the total nonfarm payroll employment surveys is probably well below 100 per cent, and (2) contact information may not be available for new workplaces, this method makes assumptions about the number of workplaces that start each month and the number that terminate each month, both of which would affect the number of people employed.
Also, like many other data values, this number can be seasonally adjusted, and the estimate that I wrote about above was seasonally adjusted. There may be certain months each year that usually have an increase in the number of people working, while other months each year may usually have a decrease in the number of people working, in both cases due to regularly occurring cyclical factors. In the latter case, this is frequently not a problem, at least in terms of being associated with an economic downturn if the decrease in the number of people working is strictly due to seasonal factors. In the former case, more people working does not automatically mean that the economy will be expanding in sustainable fashion. Adjustments are made to seasonally adjusted data so that we have more information about how the seasonally adjusted data reported give information in terms of swings in business cycles like booms and recessions.
Scratch that for the not seasonally adjusted data. For total nonfarm payroll employment data that is not seasonally adjusted, my understanding is that the data still use the birth death model but do not make adjustments for how well the estimate compares with data from that month from prior years. The unadjusted number for total nonfarm payroll employment for August 2026 reported yesterday showed an increase by about 150,000 jobs based on my calculations from U.S. Bureau of Labor Statistics (2026, b) data. That’s roughly on par with the aforementioned seasonally adjusted data.
By contrast, the household survey contacts people in the United States and asks questions about their employment status and also demographic-type information. On a seasonally adjusted basis, the data reported yesterday seem very good, in fact, probably great. I calculate from U.S. Bureau of Labor Statistics (2026, c) data that seasonally adjusted data from the household survey increased by roughly 570,000 jobs.
That was clearly not the case for the household employment number when not seasonally adjusted. The U.S. Bureau of Labor Statistics (2026, d) data that I used indicate that based on unadjusted data, the household survey estimate of employment actually fell by more than 130,000 jobs between July 2026 and August 2026. Additionally, the number reported for August 2026 is about 1.4 million fewer jobs than the all-time high from April 2025 and a nearly equal level from December 2025.
A bit more disappointing news is that the seasonally adjusted employment number from the household survey for August 2026 was approximately 620,000 jobs less than the U.S. BLS estimate from August 2025, one year ago, and very roughly 1.25 million jobs below the estimated high from December 2025.
Writing this blog post got me thinking about two songs, “Seasons in the Sun” and “In the Summertime.” The seasonal adjustments may have reminded me of the word ‘seasons’ in “Seasons in the Sun.” Perhaps this Labor Day Weekend late in the summer had me thinking about “In the Summertime.” The contrast in the themes of these two “45s” (for those of you who remember those things), the sad, reflective, bittersweet song made famous by Terry Jacks, versus the good-time skiffle revival of Mungo Jerry’s “In the Summertime” may be fitting given the conflicting conclusions that we may draw about the overall health of the U.S. labor market as of August 2026, based on the data released by the U.S. BLS.
Given that it’s a holiday weekend, I thought that I would try to take a short break from being a practitioner of ‘the dismal science’ and end the data discussion of this post on a more positive note. The seasonally adjusted establishment survey for U.S. nonfarm payroll employment for August 2026 is probably an all-time high, passing the 159 million mark for the first time.
By the way, how many of you know that The Beatles were originally a skiffle group known as the Quarrymen with different personnel? How’s that for a Labor Day weekend pun?
May you have joy and fun in this late summertime. For everyone celebrating, Happy Labor Day! Also, please realize that subsequent data revisions may change the analysis and the conclusions of this blog post.
REFERENCES
Hartman, Harrison C. (2026). “Sounding Like a Broken Record Because of U.S. Jobs Reports.” Posted Online August 7, 2026 at: https://harrisonhartman.blogspot.com/2026/08/sounding-like-broken-record-because-of.html
U.S. Bureau of Labor Statistics. (2026, a). All Employees, Total Nonfarm [PAYEMS], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/data/PAYEMS, September 4, 2026.
U.S. Bureau of Labor Statistics. (2026, b). All Employees, Total Nonfarm [PAYNSA], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/series/PAYNSA, September 4, 2026.
U.S. Bureau of Labor Statistics. (2026, c). Employment Level [CE16OV], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/data/CE16OV, September 4, 2026.
U.S. Bureau of Labor Statistics. (2026, d). Employment Level [LNU02000000], retrieved from FRED, Federal Reserve Bank of St. Louis; https://fred.stlouisfed.org/data/LNU02000000, September 4, 2026.
