An hour of work in August 2026 bought slightly less than an hour of work in August 2025. Nominal pay rose over those twelve months, and prices rose faster, so the federal government’s inflation-adjusted measure of hourly pay went backwards. The one thing that kept weekly paychecks ahead of prices was not a raise at all. It was more hours.
The series is real average hourly earnings, and it fell 0.3 percent over the year
The measure carrying the claim has a precise name. It is real average hourly earnings for all employees on private nonfarm payrolls, seasonally adjusted, published monthly by the Bureau of Labor Statistics from its Current Employment Statistics survey and deflated by the Consumer Price Index for All Urban Consumers. “Real” here means nothing more than adjusted for prices, and the deflated figures are expressed in constant 1982-84 dollars.
From August 2025 to August 2026, that series decreased 0.3 percent, seasonally adjusted, according to the Real Earnings release for August 2026, numbered USDL-26-1497 and issued on September 11, 2026. Over the single month from July to August it fell 0.1 percent, the result of a 0.3 percent increase in average hourly earnings set against a 0.4 percent increase in the price index. For production and nonsupervisory employees, measured against the price index for urban wage earners and clerical workers, real hourly pay also fell, by 0.1 percent over the year.
The programs nobody is enrolled in automatically: Each of the eleven has its own income limit and its own application, and none of them starts on its own. See the eleven programs in The Benefits Checklist.
What the raise looked like before prices ate it
The underlying dollars are not small, which is what makes the result counterintuitive. Average hourly earnings across private payrolls went from $36.62 in August 2025 to $37.75 in August 2026, a gain of 3.1 percent. The price index rose 3.4 percent over the same stretch. Deflate the pay by the prices and real average hourly earnings slipped from $11.33 to $11.30 in constant 1982-84 dollars.
Three cents of 1982 money is a strange unit for a household budget, so it helps to run it the other way. A worker at the national average would have needed about $37.85 an hour in August 2026 to hold the purchasing power that $36.62 carried a year earlier. Actual pay landed near $37.75, roughly a dime an hour short, which at 34.4 hours comes to about $3.40 a week and something under $180 across a year of full-time work. That is the entire gap. It is small, it is real, and it is the third consecutive month the annual comparison has been negative, following declines of 0.1 percent in both June and July.
The extra twelve minutes a week doing the work of a raise
Weekly pay tells a different story from hourly pay, and the difference is hours. The average workweek across private payrolls rose 0.6 percent over the year, from 34.2 hours in August 2025 to 34.4 hours in August 2026 — roughly twelve extra minutes on the clock. That increase was enough to flip the sign: real average weekly earnings rose 0.3 percent over the twelve months even as real hourly earnings fell 0.3 percent.
In current dollars, average weekly earnings climbed from $1,252.40 to $1,298.60, a gain of $46.20 a week or about $2,400 over a year. Households experience that as a bigger paycheck. The Bureau’s arithmetic says most of it was consumed by prices, and what survived came from working longer rather than from each hour being worth more. That distinction matters for anyone whose hours are fixed by a schedule, a contract, or a body that cannot absorb more of them — a part-time worker, a shift worker at maximum hours, and above all a retiree living on a benefit that does not lengthen.
Why the gap looks different for people not on a payroll
The Real Earnings series only covers people drawing a paycheck from a private employer. Nothing in it captures a household living on Social Security, a pension, or savings, and for those households the same price increase arrives without even the option of an extra twelve minutes a week. The price index that drove the result, the Consumer Price Index for All Urban Consumers, stood at 334.131 in August 2026 against 323.291 a year earlier.
That same family of indexes is what the Social Security Administration uses to set its annual cost-of-living adjustment, though it uses the urban wage earners version measured over the third quarter rather than the all-urban series measured month to month. The practical point for a fixed-income budget is that a benefit adjusted once a year is being measured against prices that moved 3.4 percent in the twelve months ending in August, while the pay of people still working moved 3.1 percent.
A preliminary number, with a correction date already on the calendar
Both the July and August figures in this release carry the Bureau’s preliminary flag, which means they are subject to revision as more survey responses from the Current Employment Statistics program arrive. Revisions to a tenth of a percentage point are ordinary, and a tenth is the whole margin in question here.
The Bureau’s own release closes with the date for the next reading: the Real Earnings news release for September 2026 is scheduled for Wednesday, October 14, 2026, at 8:30 a.m. Eastern. That publication will show whether a fourth straight month of negative annual change follows, and whether the workweek is still doing the work that hourly pay is not.
When the workweek cannot stretch, the income test is the other lever
The Bureau’s figures describe a year in which pay held roughly even with prices only because hours grew. A household on Social Security, a pension or a fixed benefit has no such lever, and the programs built for exactly that squeeze are not automatic — each one has its own income limit and its own application, and nothing triggers them when a budget tightens.
The Benefits Checklist is the 69-page guide covering 11 benefit programs with the 2026 income limits and a 50-state phone directory.
See the 2026 income limits in The Benefits Checklist.
This article was produced with AI assistance and reviewed by a human editor. Figures are linked to their primary sources.
