Inflation ran 3.4 percent over the twelve months ending in August 2026, and most households do not believe it. They are right to distrust the number, and the Bureau of Labor Statistics is not misleading them. The two sides measure different things. The government reports a rate of change. A household experiences a price level, and that level sits roughly 30 percent above where it stood in January 2020.
That gap explains almost all of the argument about whether inflation is over.
A falling rate still sits on top of everything that already happened
The Bureau of Labor Statistics reported a 3.4 percent rise in the Consumer Price Index for All Urban Consumers over the year ending August 2026, with core inflation, which strips out food and energy, at 2.4 percent. Both figures describe the last twelve months only.
The agency’s own inflation calculator answers the other question. It puts $100 in January 2020 at $129.85 in August 2026. Prices rose about 30 percent over that stretch and stayed there. Disinflation means the climb slowed. It does not mean anyone walked back down the hill.
People shopping for groceries compare today against a remembered price, not against last August. The memory runs years deep, so the felt number is the cumulative one.
Car insurance shows both numbers at once
Motor vehicle insurance makes the point cleanly. In the August 2026 release, the Bureau of Labor Statistics recorded the index down 5.1 percent over twelve months. Premiums are falling.
The same release puts the index at 848.231. In January 2020 it stood at 572.693. Comparing those two published levels gives a rise of about 48 percent over six and a half years.
A driver who renewed last month paid less than last year and roughly half again what the same coverage cost in 2020. Ask that driver whether insurance got cheaper and you will get an answer that contradicts a true statistic.
The basket is not what people picture
The index weights matter more than most coverage admits. In the August 2026 release, shelter carried a relative importance of 35.343 out of 100. Housing is over a third of the index.
Most of that weight is owners’ equivalent rent, at 25.918. That line estimates what a homeowner would pay to rent their own house. No homeowner writes that check. The largest single component of the index is a price nobody actually pays, which is defensible economics and a genuine source of the disconnect.
Meanwhile the categories households watch most closely carry small weights. Electricity sits at 2.551. Utility gas at 0.752. Food at home at 8.232. A family can watch its power bill and grocery receipt climb while those categories move the headline barely at all.
Comparing published index levels between January 2020 and August 2026 shows why that stings. Electricity went from 212.018 to 310.057, about 46 percent. Utility gas service went from 174.272 to 266.448, about 53 percent. Food at home went from 243.110 to 321.500, about 32 percent. The bills people treat as non-negotiable rose faster than the average.
Pay did not keep up, and the government says so
The Bureau of Labor Statistics publishes a real earnings series that adjusts wages for the same price index. In its August 2026 report, real average hourly earnings for private nonfarm employees fell 0.3 percent over twelve months. Nominal pay rose 3.1 percent. Prices rose 3.4 percent. The difference came out of the worker.
Stretch the window and the arithmetic gets worse. That series put real average hourly earnings at $10.97 in January 2020 and $11.30 in August 2026, measured in constant 1982-84 dollars. Call it 3 percent of real gain over six and a half years, against a price level up about 30 percent. The two numbers moved in the same direction at wildly different speeds, and households feel the difference every month.
What people actually report
The Federal Reserve asks directly. Its Report on the Economic Well-Being of U.S. Households in 2025, fielded in October 2025 and published in May 2026, found 73 percent of adults doing okay financially or living comfortably. That figure has barely moved since 2019, when it was 75 percent.
Ratings of the national economy tell a different story. In 2019, 50 percent of adults called the national economy good or excellent. In 2025, 26 percent did. Personal circumstances held roughly steady while the verdict on the country dropped by half.
The same survey found 58 percent saying price changes over the prior year left them worse off, and about nine in ten calling price increases a minor or major concern. Among adults earning under $50,000, 66 percent called it a major concern, against 42 percent of those over $100,000.
Households also acted. Sixty-two percent switched to cheaper products, 60 percent used less of something or stopped buying it, and 46 percent delayed a major purchase. Seventy-seven percent did at least one of those things. People adjusted their lives around prices, then were told inflation had normalized.
The statistic is fine and the story around it is not
None of this makes the Consumer Price Index wrong. It measures what it says it measures, publishes its weights, and prints its methodology. The failure sits in the translation. A headline announcing that inflation cooled to 3.4 percent tells a reader the crisis passed. What it actually says is that prices rose more slowly on top of a level that already moved 30 percent.
Organizations tracking household affordability have pushed the same correction. Fight For A Living Wage, a nonpartisan 501(c)(3) grassroots group, argues that affordability rather than any single wage number drives the squeeze, since housing, healthcare, childcare, food, transport and education all outran pay. Its running tally of the underlying affordability figures collects the primary-source numbers in one place.
Anyone writing about inflation has a straightforward fix available. Publish the rate and the level together. The rate is the news. The level is the experience, and it is the one readers already know.