The headline this morning was simple: the U.S. economy added 162,000 jobs in August, and unemployment held at 4.1 percent.

That is good news. It is also a first draft.

The monthly jobs report has an unusual place in American politics. One number lands at 8:30 on a Friday morning, markets move, politicians celebrate or condemn it, and headlines harden around what it supposedly tells us about the economy.

Then the number changes.

Usually, those changes are not dramatic enough to rewrite the story. Last year they were.

The number most people call "the jobs report" comes from the Bureau of Labor Statistics' Current Employment Statistics survey, which asks businesses and government agencies about payroll employment, hours and earnings.

The first estimate arrives roughly three weeks after the reference period. BLS revises that estimate in the next month and again the month after before the number is held until annual benchmarking.

That matters because "162,000 jobs added" does not mean BLS counted every payroll in America and found exactly 162,000 more jobs. It means the survey's best current estimate is 162,000.

And revisions themselves are normal.

Since BLS adopted its current design in 2003, the average first-to-third revision has been positive: about 7,000 jobs. The average absolute revision is about 51,000 regardless of direction.

So the evidence does not support the claim that BLS routinely publishes rosy numbers and quietly revises them down.

It does support saying something unusual happened in 2025.

According to BLS revision data:

October is missing because the 2025 appropriations lapse delayed the first release; its first estimate was already equivalent to the second estimate.

Every other month moved in the same direction: down.

Across 2025, the average first-to-third revision was negative 58,000 jobs. June was the extreme case: an initial gain of 147,000 became a loss of 13,000 two revisions later.

That kind of one-way pattern is not normal. The long-run average since 2003 is slightly positive. Even 2024, which had downward revisions, averaged negative 20,000 from first to third estimate, about one-third the size of 2025's average miss.

The monthly revision process is only one layer.

Once a year, BLS benchmarks its survey estimates against unemployment-insurance payroll records, which cover about 97 percent of the jobs in the establishment survey's scope.

When BLS performed that benchmark for March 2025, it cut the seasonally adjusted employment level by 898,000 jobs.

The consequence was stark: estimated 2025 job growth fell from 584,000 to 181,000 after the benchmark, a reduction of 403,000 jobs.

That does not mean 403,000 people lost jobs in February when the benchmark was published. It means the survey had been overestimating how many jobs existed during the prior year.

There is an important update that cuts against the idea of a continuing systemic miss. BLS's preliminary benchmark for March 2026, released Aug. 28, is negative 79,000 jobs, or 0.1 percent. The average absolute benchmark revision over the past decade is 0.2 percent. If the preliminary number holds, this year's benchmark miss would be far smaller than last year's.

The regular monthly revisions this year have gone both directions: January up 30,000, February down 64,000, March up 36,000, April up 33,000, May down 109,000, June down 26,000.

July has gone through only one revision so far. Today it moved up 44,000, from a reported loss of 23,000 jobs to a gain of 21,000. June also moved up by 11,000 in today's release.

So there is no honest basis for predicting that today's 162,000 will eventually become a much smaller number. It may go down. It may go up.

The useful lesson from 2025 is simpler: one monthly estimate deserves less certainty than it often receives.

August looks stronger when compared with the recent hiring trend. BLS says payrolls have grown by an average of just 31,000 jobs per month over the previous 12 months. August's 162,000 is more than five times that pace.

But the labor market is only part of what Americans mean when they say "the economy."

Average hourly earnings for private-sector workers rose 3.1 percent over the past year. Consumer prices rose 3.4 percent through July: food up 3.0 percent, shelter 3.2 percent and energy 14.7 percent. The Federal Reserve's preferred PCE price index was at 3.7 percent, while inflation-adjusted consumer spending was essentially flat in July.

These are different measures and should not be treated as a perfect apples-to-apples calculation of household purchasing power. They do, however, help explain why a low unemployment rate and a positive jobs number can coexist with lousy public sentiment.

Only 24 percent of Americans rated economic conditions excellent or good in July. Forty-one percent called them only fair and 35 percent poor.

The University of Michigan's August consumer survey found its sentiment index at 51.7, down about 11 percent from a year earlier. At the start of 2026, 23 percent of consumers said inflation would cause more serious hardship than unemployment. By August, 36 percent picked inflation and only 6 percent picked unemployment.

That gap is politically important. Most people do not experience "the economy" as a national payroll statistic. They experience it through rent, groceries, gasoline, wages, a job search, a mortgage rate and whatever disappeared from their checking account that month.

Both parties understand that.

In a Reuters/Ipsos poll conducted Aug. 28-31, 47 percent of registered voters said cost of living would be the single most important factor in their midterm vote. Seventy-one percent of adults disapproved of President Trump's handling of cost of living; 22 percent approved.

The Democratic Congressional Campaign Committee's recent messaging blames Trump and House Republicans for higher prices and says Democrats are focused on reducing the cost of gas, groceries, housing and health care.

That is campaign messaging, not an independent economic finding, but it tells you exactly what Democrats want voters thinking about when they enter the booth.

The NRCC calls affordability the "signature debate of 2026" and tells Republican candidates to emphasize tax cuts, lower costs and household pain points including groceries, health care and housing.

Neither party is campaigning on the intricacies of BLS sample revisions.They are campaigning on who voters blame for how the economy feels.

The August report deserves its good headline. Employers added an estimated 162,000 jobs, unemployment stayed at 4.1 percent, and June and July were revised upward.

It also deserves context.

Hiring over the previous year has been weak and inflation remains above the Federal Reserve's 2 percent target. Consumer sentiment remains depressed. And 2025 demonstrated that the first payroll estimate can give a materially better impression of job growth than the fuller record later supports.

The answer is not to distrust BLS. The revision process is the reason the record gets more accurate.

The answer is to question any reporting that makes these numbers more significant than they actually are.

Today's 162,000 is the best estimate we have right now.

Just do not write it in ink.