Two Line Items - Restaurants and School Districts - Account for Most of August's 162,000

by Nora Halloran · September 4, 2026

Senior Analyst, US Market Current

The August employment report was a large upside surprise on the headline and a much narrower one underneath it. Total nonfarm payroll employment increased by 162,000, according to the Bureau of Labor Statistics, against a Reuters consensus of 56,000. Where that gain came from is the part worth sitting with before drawing policy conclusions from it.

Two industry lines dominate. Food services and drinking places added 59,000 jobs, and local government education added 42,000. Added together, those two categories account for 101,000 of the month's 162,000, or a little under two-thirds of the total. Payroll categories do not sum tidily to the headline because of rounding and offsetting declines elsewhere, but the concentration is not a rounding artifact.

Take each in turn. The restaurant and bar figure is genuinely large relative to its own history: the BLS described the 59,000 gain as well above the average monthly gain of 12,000 over the prior 12 months. Reuters put the broader leisure and hospitality increase at 62,000, which implies the rest of that supersector contributed comparatively little. A gain of that size in a category that had been running at a fifth of the pace is a data point that deserves a second month before it becomes a trend.

The local government education gain is more explicitly a round trip. The BLS said the 42,000 increase in August largely offset a decrease in the prior month. Public education hiring is heavily seasonal, with staffing decisions bunched around the start of term, and seasonal adjustment has to guess when those decisions land. A July decline followed by an August recovery of similar size is more consistent with adjusted timing than with a change in the level of school staffing.

Against that, one of the largest and steadiest contributors to recent payroll growth slowed markedly. Health care added 13,000 jobs, which the agency noted was slower than its average monthly gain of 32,000 over the prior 12 months. If health care were still running at its own trend, the report would have looked stronger still; it is also the case that the strongest lines in this report are not the ones that have been reliable.

Information was the clear negative. Employment in the sector declined by 23,000, against losses the BLS said had averaged 8,000 per month over the prior 12 months, and the agency identified a decline of 8,000 within computing infrastructure providers, data processing, web hosting, and related services, alongside losses of 7,000 in publishing industries and 5,000 in broadcasting and content providers. That is a sector-level contraction accelerating rather than stabilising, and it sits awkwardly next to an equity market where the technology complex has been leading.

The goods-producing side was steady. The BLS logged construction as having "changed little in August (+22,000)", and said employment in manufacturing "continued its upward trend (+16,000)" and is up by 58,000 since a recent low in December 2025. Neither is a boom, but neither is the contraction that some of this week's survey data implied.

Everything else was flat. The BLS listed mining, quarrying, and oil and gas extraction; wholesale trade; retail trade; transportation and warehousing; financial activities; professional and business services; social assistance; and other services as showing little change over the month. Professional and business services in particular is the cyclical white-collar category most economists watch for turning points, and in August it did not turn.

The price of labour did not accelerate to match the quantity. Average hourly earnings rose by 10 cents, or 0.3 percent, to $37.75, and were up 3.1 percent from a year earlier. For production and nonsupervisory workers the gain was 11 cents to $32.53. A month in which employers added three times the expected number of workers without paying visibly more for them is not the picture of a labour market bidding against itself.

The household survey, drawn from a different sample, adds a supply-side wrinkle. Reuters reported that the labour force expanded by 683,000 in August, while the BLS recorded an increase of 569,000 in the number of employed. That combination is why the unemployment rate could stay unchanged at 4.1 percent in a month with a large payroll gain: more people came looking for work at roughly the pace employment grew.

The participation figures cut both ways. The labour force participation rate edged up to 61.6 percent, but the BLS was careful to note that it is down by 0.5 percentage point since January, and the employment-population ratio at 59.1 percent showed little change over the month or the year. A single month's rise in participation does not undo a year's drift.

Slack measures generally improved. The number of people working part time for economic reasons fell by 414,000 to 4.4 million. The long-term unemployed, at 1.9 million, made up 27.0 percent of all unemployed people, and the count of marginally attached workers was little changed at 1.7 million.

The revisions do more than pad the total. June was revised up by 11,000 to +31,000, and July was revised up by 44,000, from a reported decline of 23,000 to a gain of 21,000. A good deal of the softness narrative built through August rested on that July contraction. It no longer exists in the data, which means the three-month path now reads as slow, then slower, then a jump, rather than as an outright break.

For the policy debate, the honest reading is that the report removes an argument for standing pat without settling much about inflation. Stephen Brown of Capital Economics told Reuters that "Even the most committed dove would struggle to find anything in the August employment report to justify keeping interest rates unchanged." Markets agreed in the direction if not the magnitude: the Associated Press reported CME FedWatch pricing for a September increase at 60.2% on Friday, up from 49.4% Thursday. US markets were still open as this was written, so that pricing, like every other figure quoted here for Friday, is an intraday reading and not a closing one.

The composition, though, argues for holding that conviction loosely. A gain built substantially on a restaurant number running nearly five times its own 12-month average and a school-staffing line the BLS itself describes as offsetting a prior-month drop is a gain with more revision risk than most. The next real test is not another labour print but the August Consumer Price Index, scheduled for 8:30 a.m. on Sept. 11 according to the BLS calendar, four days before the Federal Open Market Committee convenes on Sept. 15-16 as reported by Reuters.