The Top 10 Percent Supposedly Drive Half of All Consumer Spending. The Government's Own Survey Says Otherwise.

On September 16, 2025, Moody's Analytics chief economist Mark Zandi published a chart built from his own analysis of Federal Reserve data: the top 10% of earners accounted for 49.2% of all U.S. consumer spending in the second quarter of 2025, up from 48.5% the quarter before, the highest share in a series reaching back to 1989. The chart moved fast, retold within days by outlets across the business press, each version landing on the same sentence: the American economy now runs on the wealthy alone.

Not everyone accepted the number on sight. An economist at the University of California, Berkeley named Antoine Levy said in public that anyone who actually works with statistics like these should feel an immediate doubt it could be right. Separately, an economics newsletter called Against Narrative, written by an analyst using the byline Berg, pulled the government's own household spending survey to check the number directly. What came back wasn't a small correction. It was a figure less than half the size of Zandi's.

The claim, and how it became unavoidable

Zandi's figure didn't spread because it was subtle. It spread because it fit a story already primed to be believed. The term "K-shaped economy" traces to 2020, when a Twitter user posting as Ivan the K floated the shape as wordplay on his own handle, and behavioral economist Peter Atwater, who teaches at William & Mary, picked it up to describe what he was seeing: white-collar workers shifting to remote work with their incomes intact, blue-collar and service workers bearing the brunt of the shutdown. By 2025 that five-year-old coinage had become one of the year's most-repeated business buzzwords. Into that primed environment, a round, shocking number, essentially half of everything, landed as confirmation rather than a claim that needed checking.

The gap nobody has explained

Berg's analysis went to the U.S. Bureau of Labor Statistics' Consumer Expenditure Survey for 2023, the most recent year with detailed decile-level tables available. That data shows mean annual expenditure of $77,280 across all households and $180,758 for the top decile, implying a top-10% spending share of roughly 23.4%, or 21.8% once retirement and Social Security contributions are excluded. A second, independent check, using income instead of spending, comes from the Congressional Budget Office, which puts the top 10%'s share of after-tax income at 37.1% using 2021 data. If the top 10% were really responsible for 49.2% of spending, they'd be spending more than their entire share of after-tax income, drawing down savings at a scale nothing in the government's own savings-rate data shows. A third figure, the Bureau of Labor Statistics' broader estimate that the top 20% accounts for roughly 41% of consumption from 2017 through 2021, lands in the same range as Berg's number and nowhere near Zandi's.

It's worth being precise about what these numbers can and can't do against each other. Zandi's figure describes the second quarter of 2025. Berg's calculation uses 2023 data, the most recent year with published decile detail, not the identical quarter. That gap could, in principle, explain some difference if the top decile's share had genuinely spiked in between. It can't explain a gap this size: the survey shows the top decile's share moving by low single digits year over year, nothing close to a 25-plus point swing, and the income cross-check doesn't depend on survey year at all. As of this writing, neither Levy, Berg, nor anyone else who's examined the claim publicly has located a published methodology from Zandi or Moody's Analytics explaining how Federal Reserve data produces a figure roughly double the government's own survey estimate.

What the Federal Reserve's own economists say instead

The more interesting wrinkle is that the Fed's own research doesn't agree with itself. A Federal Reserve Bank of Minneapolis review published in 2026 concluded that competing measures of the K-shaped economy range "from a steep K to no obvious K-shape," and that the more careful characterization is a bifurcated recovery: both income groups' spending grew, just at sharply different rates, rather than a clean collapse at the bottom. A separate Federal Reserve Bank of New York analysis, from its Liberty Street Economics team in May 2026, used a 200,000-respondent consumer panel benchmarked against Census retail data and found only households earning more than $125,000 showed consistent real spending growth from 2023 onward, while households earning under $40,000 saw real declines for parts of 2023 and 2024, a genuine divide. But the same researchers flagged their own method's limits: fixed income brackets, rather than true percentiles, could mechanically inflate the top's apparent share, and the truly wealthy looked less dramatic in their data than expected, because a consumer panel tends to undercount the services-heavy spending, private schooling, wealth management fees, that defines how the rich actually spend.

Line up the three credible attempts to answer this question, Berg's survey calculation, the Minneapolis Fed's bifurcated read, and the New York Fed's own panel analysis, and none agrees on the exact size of the divide. All land somewhere between a real, moderate gap and a real but hard-to-measure one. None comes anywhere near half.

Why a shocking, wrong number beats an accurate one

There's a structural reason 49.2% outcompeted every more careful estimate, and it has nothing to do with which is true. A number that confirms a story the audience already suspects needs no persuasion, only transmission. This isn't a new pattern. Researchers at MIT tracked roughly 126,000 news stories shared on Twitter between 2006 and 2017 and published the results in Science in 2018: verified false stories were 70% more likely to be retweeted than true ones, and the researchers' own explanation was novelty, people share what strikes them as surprising. A claim that the top 10% drive half the economy is, by that logic, simply more shareable than one claiming a fifth to two-fifths, independent of which is closer to correct. None of this requires assuming bad faith on Zandi's part. An analyst can build a real, defensible method that just hasn't been published in a form outsiders can check, the same distinction that separates a genuinely misleading number from one that's simply measuring something narrower than advertised. Until that documentation appears, the honest read is that the number is unverified and inconsistent with every independent check available, not that it's been disproven outright.

The honest complication

None of this means the K-shaped economy is a myth. Every credible estimate here, Berg's 23%, the Bureau of Labor Statistics' broader 41% for the top 20%, the Minneapolis Fed's bifurcated read, the New York Fed's panel analysis, agrees the top of the income distribution is pulling further ahead in how much it spends, and has been since at least 2023. The dispute is about magnitude, not about whether a real gap exists, the same widening-versus-shrinking question that shows up in how differently a raise can feel depending on when you got it. It's also fair to note what hasn't been established: Zandi hasn't retracted the figure or addressed the income-share objection publicly, and the Fed's own three-way disagreement on magnitude is a genuine, unresolved research question, not one this piece is claiming to close.

The verdict

A real, multi-source-corroborated divide exists, and every rigorous, checkable estimate of its size lands somewhere between one-fifth and two-fifths of total spending, not one-half. The number that went viral came from a single analyst's undocumented method, contradicted the government's own household spending survey by a factor of roughly two, and spread because it confirmed what people already suspected, not because anyone checked the arithmetic. The test worth applying to next year's viral economic statistic is the same one that would have caught this one: not whether it feels true, but whether anyone who's challenged it has actually been shown the math.

---

This article is educational and is not financial advice.

Quick answers

Do the top 10 percent of earners really account for half of all U.S. consumer spending? That specific 49.2% figure comes from one economist's proprietary analysis of Federal Reserve data, not a published government survey. Using the U.S. Bureau of Labor Statistics' own Consumer Expenditure Survey, independent researchers calculate the real share at roughly 23%, and the higher number would require the top 10% to spend more than their entire share of after-tax income.

Is there a real gap between how much the wealthy spend and how much everyone else spends? Yes. Government survey data, income-share math, and two independent Federal Reserve research teams all agree a real divide exists and has widened since 2023. They disagree on exactly how large it is, but none of their estimates comes close to the viral "half of everything" figure.

Why did the 49.2 percent number spread so widely if it doesn't hold up? It confirmed a story, a K-shaped, inequality-driven economy, that was already widely believed, and it came from a frequently cited economist whose reputation earned it the benefit of the doubt. A round, shocking number that agrees with what an audience already suspects needs no persuasion, only retelling.

What's the most defensible estimate of the top 10 percent's actual spending share? Somewhere the Bureau of Labor Statistics' own data and the Federal Reserve's independent research both point toward: roughly one-fifth to two-fifths of total consumer spending, depending on how the top group is defined, not one-half.