Food-Service Payrolls Rebound in the First August Jobs Estimate

Food services and drinking places added an estimated 59,000 payroll jobs on a seasonally adjusted basis in August, the Bureau of Labor Statistics reported September 4.

The initial August employment release put that increase well above the sector’s average monthly gain of 12,000 over the preceding year. Table B-1’s more precise estimate was 59,200 additional jobs, bringing seasonally adjusted employment to about 12.43 million.

National nonfarm payrolls increased by 162,000. The unemployment rate held at 4.1%, a separate measure drawn from the household survey. The restaurant figure covers food services and drinking places rather than quick-service chains alone.

July improved in revision, but still showed a decline#

The August 7 release put June food-service employment at 12.3729 million and July at 12.3468 million, implying a decline of 26,100 jobs.

The September release revised both months. Its comparable July decline was 10,400, an upward revision of 15,700 to the monthly change.

The September report revised national June and July payroll gains upward by a combined 55,000. That economywide revision is separate from the restaurant-sector adjustment in the table.

What a payroll gain measures#

The July release’s technical note defines payroll employment using the pay period containing the 12th. It includes anyone paid for any part of that period, including employees on paid leave.

A person appearing on two payrolls is counted in each job. Payroll figures track jobs, while the household survey describes individuals’ employment status.

A monthly net increase also differs from the total number of hires. Replacing someone who left can involve recruitment and training without enlarging the payroll count. The August figure measures the net movement in payroll jobs, adjusted for recurring seasonal patterns.

Why the first estimate changes#

BLS’s methodology, updated in February 2025, explains the normal revision sequence. The first release arrives before every responding establishment has supplied its payroll information. A second preliminary estimate follows one month later, then a final sample-based estimate after another month.

Later returns, corrections and recalculated seasonal factors can all alter the monthly figures. Annual benchmarking is a separate process that anchors the sample estimates to a nearly complete employment count, primarily from unemployment-insurance tax records.

The annual process can recalculate nearly two years of data, including months that have already passed through both routine revisions. A figure labeled final in the monthly sequence remains open to benchmarking; the label marks completion of the sample-based release cycle.

Historical revisions can also reflect a new industry classification, usually during the annual benchmark update.

BLS publishes detailed monthly industry employment in thousands to one decimal place, equivalent to the nearest hundred jobs. Major sectors and the national total are rounded to the nearest thousand. That convention explains why the release’s narrative and its detailed restaurant table can present the same movement with different precision.