University of Michigan Consumer Sentiment Index Falls to 47.8 in Preliminary September Reading

The Same Number Landed Differently on Everyone Who Saw It

Consumer confidence in the United States just fell to 47.8 in the preliminary September reading, according to the University of Michigan's Surveys of Consumers^[1][5]. That is down from 51.7 in August and well below the 51.0 that economists had expected^[1][5][11]. It's the second straight monthly drop, and the second-lowest reading in a survey that goes back to 1952^[2].

Only one month has ever been worse: May 2026, which set the all-time low^[2]. So the headline number is bleak by any measure. But the same release also contains a detail that cuts the other way — and both sides of the story are true at once.

That detail is inflation. Year-ahead inflation expectations rose to 4.6%, up from 4.0% in August^[1][2]. Yet on the very same morning, the government's own inflation report showed core prices — which strip out food and energy — rising just 2.4% over the past year, the mildest pace since March 2021^[9]. A sour mood and a cooling core inflation rate, in the same data dump. That collision is the real story here, and it's why people are reading the same numbers so differently.

What a Feelings Survey Actually Measures

The Michigan index isn't a receipt. It doesn't count what people bought last month. It's a survey that asks households how they feel about their own finances and about the economy, and then turns those answers into a number^[10].

That distinction matters because feelings and spending don't always move together. Survey director Joanne Hsu tied September's drop to two things: a jump in fuel prices and renewed trade tensions^[2]. The federal price data backs up the fuel part directly — gasoline rose 3.9% in August alone and was 27.4% higher than a year earlier^[8][9].

Two sub-measures inside the survey moved very differently, and that gap tells its own story. How people rate their finances right now barely moved, slipping to 50.9 from 51.9^[5]. How they see the year ahead fell hard, to 45.8 from 51.5^[5]. In other words, people aren't saying things are bad today. They're saying they expect things to get worse.

Why does that distinction matter to a bank in Washington, not just to a household budget? Expectations are partly self-fulfilling. If enough people expect prices to keep climbing, they ask for raises, stock up early, or delay big purchases — and those choices help make the expected inflation real. That's why the Federal Reserve watches this survey closely, even though it doesn't measure actual spending^[1][10]. A five-year inflation expectation of 3.4%, only slightly up from 3.3%, worries the Fed more than the flashy 47.8 headline, because it hints at whether the public still trusts the Fed to keep prices under control over the long run^[1][10].

Whose Fault Is the Gas Pump?

This is where the dispute actually lives, and it isn't really about the number 47.8. It's about what caused it.

The administration and allied commentators point to the energy numbers as evidence of an outside shock, not a policy failure. Gasoline is up 27.4% over the year and energy overall is up 16.3%^[8][9]. Strip out food and energy, they note, and core inflation is running at 2.4% — the calmest in five years^[9]. On this reading, the underlying economy is cooling nicely, and an energy price spike tied to the conflict with Iran is sitting on top of it, distorting the picture. They also argue that hard data like retail sales, jobless claims, and earnings have held up better than mood surveys have through 2026, and that a sentiment index increasingly reflects the respondent's politics rather than their bank balance^[10].

Critics of the administration counter that trade tensions are a policy choice, not weather — and Hsu named trade tensions right alongside fuel as a cause of the drop^[2]. They also argue that core inflation is the wrong number to feel good about, since households don't buy a "core" basket. They buy gasoline and groceries, and the headline rate families actually pay is 3.4%^[9]. A near-record-low sentiment reading, on this view, is the public correctly noticing a gap between the official statistics and what things cost at the checkout line^[2].

Here's the twist in that argument, though: the survey's own internal breakdown resists both political stories. Sentiment fell among Democrats and Republicans alike this month, while independents barely moved at all^[4]. That's a hard fact to fit into a purely partisan account in either direction. Notably, Republican confidence in the government's handling of the economy sits at just 35% — reportedly the lowest since Trump returned to office — meaning erosion is happening inside the president's own base, not just outside it^[4][6][7].

Why the Midterms Make This Number Matter More

None of this would carry quite the same weight without timing. The 2026 midterms are weeks away, and the economy consistently polls as a top issue for voters^[7]. That's the pressure sitting underneath every interpretation of this release: both parties need the number to mean something specific.

The administration needs 47.8 to mean "temporary war-driven shock," something that will fade once energy prices settle. Critics need it to mean "the public is catching on to an affordability problem baked into current policy." Neither side invented its evidence — the gasoline numbers are real, and so is the core inflation reading. They're just pointing at different parts of the same release^[6][7].

There's a third audience watching for a different reason entirely: markets and the Fed. For them, the load-bearing figure isn't 47.8 at all. It's the 3.4% five-year inflation expectation, because that number speaks to whether the public still believes the Fed can keep inflation under control over time^[1][10]. A one-month jump in the year-ahead figure is written off as gas prices moving around, the way they always do. A drift in the five-year figure is treated as a much bigger deal — a sign that inflation expectations could be "de-anchoring," meaning people are starting to expect persistent inflation as the new normal^[1][10]. A soft sentiment number paired with a hot gasoline number pulls the Fed's likely next move in two directions at once, which is exactly what makes this release hard for markets to price^[4].

The Coverage Split, in Miniature

How outlets described this same release says almost as much as the release itself. Reuters led with the neutral verb "deteriorates" and held the partisan breakdown well below the numbers — closest to a straight recitation of the data^[3]. Bloomberg named both causes, fuel and trade tensions, right in its headline, tracking Hsu's own explanation almost word for word^[4].

CNBC went further, saying consumer outlook "plunges" and foregrounding the "second-lowest on record" framing — language that outruns a 3.9-point move, even a significant one^[2]. The Epoch Times, writing for a U.S. right audience, attributed the drop to "inflation woes," an oddly ownerless phrase that separates the price increase from any policy or actor^[6]. Fox Business covered the same morning's inflation data under the headline "remained elevated," a steady-state framing on a month when gas prices actually jumped^[9].

Xinhua's coverage was flat and number-dense, and unusually restrained in tone for state media^[5]. But it left out the causes Hsu herself named — the trade tensions and the broader instability tied to the Iran conflict — leaving readers with a bare portrait of American economic strain and no explanation attached^[5]. The University of Michigan will publish its final September figure later this month, based on a larger sample, so this preliminary number could still move^[1].