The Census Bureau put out a good number on September 15, and you should hear it straight before anybody complicates it. The typical American household took in $87,460 in 2025, adjusted for inflation, and the Bureau's release says that median is "the highest on record dating back to 1967." The official poverty rate dropped half a percentage point, to 10.2%.

That is the headline, and it is true. This piece is about what sits underneath it.

Because the same release, and the two full reports behind it, say something the headline does not. The top of the income ladder climbed. The bottom rung did not significantly move. And the poverty measure that counts taxes, tax credits and aid held still at 13.1%.

So this show is going to ask the question it always asks of a record. Whose record is it?

The Receipt

Start with the release, number CB26-151. The median household had $87,460 of income in 2025, "an increase of 2.6% from the 2024 estimate of $85,210." Record since 1967. Now keep reading, because a few paragraphs down, in the same income section, is the sentence that matters:

"Household income at the 90th percentile increased 1.7% but did not significantly change at the 10th percentile between 2024 and 2025."

Read it again. A percentile, in the Bureau's own glossary, is "The dollar amount below which a given percentage of households fall." The 90th percentile is the household that out-earns ninety of every hundred. The 10th percentile is the household that out-earns ten.

The full income report, P60-289, puts dollars on both ends. Income at the 90th percentile was $261,300 in 2025. Income at the 10th percentile was $20,010. The report's own words: "Changes in income between 2024 and 2025 were not constant across the income distribution."

Then poverty. The official rate came down to 10.2%, and the release puts the number of people below the line at 34.5 million. The official rate for children "fell to a historic low of 13.4%," and the rate for Hispanic individuals decreased "to a historic low of 13.9%." Those two are real records and they deserve saying out loud. They are also the only two historic lows in the release. The overall poverty rate fell. The Bureau does not call it a record low, and neither will I.

Then the other poverty number, the one that does not make the headline. "The SPM rate in 2025 was 13.1%, not statistically different from 2024."

What "not significantly" means, and why I use the Bureau's words

Here is the part that is easy to skip. The Bureau tests every comparison it makes. The release says so at the bottom: "All comparisons made here and in each respective report have been tested and found to be statistically significant at the 90% confidence level, unless otherwise noted."

So when the Bureau writes that the 10th percentile "did not significantly change," it is not saying the bottom stood still to the dollar. It is saying the survey cannot tell a change there from noise. That cuts both ways, and I am going to hold this piece to it. Nobody can tell you from this release that the bottom fell. Nobody can tell you it rose. What the Bureau can tell you is that the 90th percentile rose, measurably, and the 10th did not.

Same rule for the Supplemental Poverty Measure. 13.1% is not a rise. It is not statistically different from the year before. In a year of record income, that is its own kind of news.

Why It Costs You

Two poverty measures, and they are not asking the same question.

The release puts the difference in one sentence: "While the official poverty measure is based on money income, which is pretax and does not include tax credits, the SPM is a post-tax and transfer poverty measure." The poverty report, P60-290, spells out what that means. Refundable tax credits, SNAP and housing subsidies "are only included in the SPM." Taxes, medical expenses and work-related expenses "are also deducted from SPM resources but are not considered in the official poverty measure."

This is my read, labeled as mine: the SPM looks at a household the way a household looks at itself. What came in, what the government added, and what went right back out to taxes, the doctor, and the cost of working. By that measure, 13.1%, not statistically different from 2024.

Now watch it by age. The poverty report says that "While official+ poverty rates decreased for children and 18- to 64-year-olds between 2024 and 2025, SPM rates did not change significantly for any age group." (Official+ is the Bureau's name for the official measure run on the same population as the SPM, so the two compare cleanly.) For children, the two measures landed in the same place: "both at 13.4 percent."

For people 65 and older, the gap is widest. Their SPM rate was 15.4% and their official+ rate 9.8%, and the report says the larger gap "was primarily due to differences in the treatment of medical expenses between the two measures." Medical expenses, across everyone, pushed 7.7 million people into poverty in 2025 by the report's count. The official measure never sees that bill.

And the aid the official measure never sees: refundable tax credits in 2025, the report says, lifted 6.1 million people over the poverty line, "including 3.3 million children." Social Security, which both measures count, "continues to be the largest antipoverty program, moving 28.8 million individuals out of SPM poverty in 2025."

One group moved the wrong way. "Between 2024 and 2025, SPM rates increased for those without a high school diploma," the release says. For every other group it discusses, the SPM change was not significant.

How the Machine Works

Why do the two measures disagree at all? Part of it is how old the official line is.

The poverty report lays the two definitions side by side. The official threshold: "Three times the cost of a minimum food diet in 1963." It gets updated every year with the Consumer Price Index. In 2025, the release says, the weighted average official threshold for a family of four was $32,970.

The SPM threshold is built from what people spend now, on "food, clothing, shelter, utilities, telephone, and internet," averaged over five years and lagged one. The report puts the SPM rate 2.9 percentage points above the official+ rate in 2025, and says that "differences in how the poverty thresholds are adjusted explain part of this gap." When food or rent climbs faster than inflation overall, the SPM line climbs with it. The official line is a 1963 grocery bill, times three, with inflation on top.

Then the income side. A median is the household in the exact middle. So a record median is the middle's record. What happened around it, per P60-289:

  • The ratio of the 90th percentile to the 10th was 13.06 in 2025, "not statistically different from the 2024 ratio of 12.74."
  • The ratio of the 50th percentile to the 10th "increased 3.5 percent, from 4.22 in 2024 to 4.37 in 2025." Set that next to the 90-to-10 ratio with care, because the report's footnote 16 adds: "The difference between the 2024–2025 percent changes in the ratio of the 90th to the 10th percentiles and the ratio of the 50th to the 10th percentiles was not statistically significant."
  • The Gini index was 0.490 and "not statistically different between 2024 and 2025."
  • The lowest fifth of households received 3.0% of all household income. The highest fifth received 52.4%. The top 5% received 23.5%. The report says the changes in the quintile shares were not statistically significant.

Put those in one picture and it is precise, and it is not the picture a headline gives you. Inequality, by the Gini index and the 90-to-10 ratio, did not significantly change. The 50th-to-10th ratio did rise, so the middle moved further from the bottom, but P60-289's own footnote says that change was not statistically different from the change in the 90-to-10 ratio. So this piece builds no contrast between them. And the top fifth of households still took in more than half of all household income in the country.

The Strongest Version of the Other Side

Here it is at full strength, because it is mostly true.

The record is real. The official poverty rate fell. Child poverty and Hispanic poverty hit historic lows on the official measure. Median income rose 4.8% for Black households, the release says. For full-time, year-round workers, median earnings "increased 3.2% for women," and the female-to-male earnings ratio rose to 83.9%. After taxes and credits, the median household's income rose 3.1%. The SPM did not rise. The Gini did not significantly move. If you want to call 2025 a good year for the typical household, the Bureau's numbers back you.

And one caution belongs to both sides. The release says the survey's weighted response rate was 61.3%, down from 62.0% the year before, and that "As response rates decline, the risk of bias increases." That caveat attaches to the record and to the flat line alike.

So here is the answer, and it is narrow on purpose. The typical household is the median. This show's question is about the households below it. For them, the Bureau's own sentences say: no significant change at the 10th percentile, no significant change in the measure that counts what they pay and receive, and a wider gap between them and the middle.

What You Can Actually Do

When someone quotes the record, ask which percentile. The $87,460 is true. So is the sentence a few paragraphs under it. Quote that one back, word for word, with the release linked: "did not significantly change at the 10th percentile."

Open the tables yourself. The Bureau's press kit links all three reports and the Bureau's data tables, and the release says that is where to find "state-level poverty and SPM rates using three-year averages." Find your state's row before somebody tells you what it says.

Keep the two poverty numbers separate. The release does not call the overall rate a record low. Children and Hispanic individuals hit historic lows on the official measure. The overall official rate fell. The SPM held. Those are four different sentences and they are all in the release.

Read the cost side. The SPM is 13.1% partly because it subtracts medical bills and work expenses. The Poverty Tax is this show's piece on why being poor costs more, line by line.

The So-What

A record is a record. Nobody here is taking it away from the middle of the country.

But the Bureau handed every politician the same release, and every one of them gets to choose which sentence to read aloud. The first sentence is the record. The one a few paragraphs down is the bottom rung that did not significantly move, and the measure that counts taxes and aid held still at 13.1%. That is not a footnote to the good news. That is who the good news skipped.

Come back for the rest

We do this every week: the number everybody is quoting on top, and the sentence a few paragraphs down that nobody reads aloud. Pull up a chair, pour something cold, and let us send you the receipts before the next record gets announced without its bottom line. Subscribe to The Long Pour, the free newsletter where the week's small thefts and the one big one land in your inbox together.

Y'all come back, now.


The Receipts

Every document below was opened and saved on the date given.