Guide · Markets
Household vs Establishment Survey: The Two Jobs Counts
The US jobs report combines two surveys: payrolls count jobs at employers, the household survey counts people with work. In October 2026 they showed +165,000 and +463,000.
On Friday, November 6, 2026, the Bureau of Labor Statistics published the October employment report, and it contained two very different measures of job growth. Nonfarm payrolls rose by 165,000. Employment in the household survey rose by 463,000, the labour force by 421,000, and the unemployment rate fell to 4.1%. Both numbers are correct; they come from separate surveys that measure different things, and knowing which is which explains most of the confusion on jobs-report mornings.
The establishment survey, officially the Current Employment Statistics programme, asks employers about their payrolls. BLS samples roughly 120,000 businesses and government agencies covering several hundred thousand worksites. It counts jobs, not people, so someone with two part-time jobs is counted twice, and it leaves out the self-employed, farm workers and people employed by private households. It produces the headline payrolls figure, average hourly earnings and the average workweek, and each month's estimate is revised in the two following reports and again in an annual benchmark: in October 2026, revisions added 84,000 jobs to August and September.
The household survey, the Current Population Survey, is run by the Census Bureau for BLS and interviews about 60,000 households. It counts people and asks what they did during the reference week: worked, looked for work or neither. That is where the unemployment rate, the participation rate and broader measures such as U-6 come from, and it includes the self-employed and farm workers. Because the sample is much smaller relative to the population, its monthly employment change is far noisier: a move of a few hundred thousand can fall within the survey's margin of error, against roughly 130,000 for payrolls.
The two often diverge for a month or two. Differences in coverage are one reason: a rise in self-employment or in second jobs shows up in only one of them. Sampling error is another, and population estimates a third: the household figures are scaled to Census population controls, which are updated each January and can shift the levels in one step. Over a year or more, the two series tend to tell the same story; month to month, the payroll number is the more reliable guide to hiring, and the household survey the better guide to who is in and out of work.
Markets react first to payrolls and the unemployment rate, then to the details. In October 2026 the headline numbers were strong, and the household survey added two caveats: U-6 rose 0.3 percentage point to 7.9%, and people unemployed for more than 27 weeks made up 27.6% of the jobless, the highest share since late 2021. Strong hiring with longer unemployment spells points to a market where companies add jobs but rarely let go of anyone, so those who lose a job struggle to find the next one.
When the two surveys disagree, three habits help. Compare three-month averages rather than single months, because both series are noisy. Check the revisions to payrolls before trusting the newest figure. And read the unemployment rate together with participation: a falling rate because people stopped looking is weaker news than a falling rate because more people found work, which is what the October 2026 numbers showed.