Have U.S. consumers gone “K-shaped”? A review of the data

The term “K-shaped economy” arose after the COVID-19 pandemic to characterize the diverging economic paths of high-income Americans and everyone else. It became ubiquitous in 2025, one of the “biggest business buzzwords” of the year.

Growing U.S. inequality in income and wealth is not a new phenomenon. But the recent K-shaped story is often told in terms of consumption growth—of rich shoppers powering the economy with their spending while others pull back. Retailers remark on their bifurcated customer base; airlines and hotels tout the boom in their premium fares and luxury hotel brands.

Like the contemporaneous assertion of an American “vibecession,” the conventional wisdom around the K-shaped economy has depended heavily on such anecdotes. Beyond the vibe, what case do the available data sources make for K-shaped consumption? Comparing available measures is much like setting proverbial apples and oranges next to one another. But the overall picture is more complicated than the media headlines suggest.

Moody’s: K-shaped in a big way (but with skeptics)

The credit-rating agency Moody’s Analytics shares its proprietary analysis primarily via social media and financial news outlets. Moody’s calculates a steep, K-shaped spike in spending growth by the top 10 percent of households (by income) starting in 2020. In early 2026 via The Washington Post, Moody’s reported that spending by the top 10 percent grew 62 percent between third quarter 2020 and third quarter 2025, much more than any other income group (Figure 1A).

For the most recent 12 months, spending by these top earners grew more than 4 percent in nominal terms; growth for all other groupings in 2025 was close to 3 percent. These households, Moody’s says, now account for more than 45 percent of spending while the bottom 60 percent of households comprise just 23 percent of spending (Figure 1B).

Each prior update from Moody’s garnered headlines through 2025, although the methodology and stark findings have drawn skeptics. It is not a direct measure of consumption expenditures by households. Rather, Moody’s works backward from financial and wealth data to estimate the average household savings for each income grouping.¹ They count whatever disposable income remains—technically, “personal outlays”— as a broad measure of spending.

Critics argue that the concentration of spending among the top 10 percent is implausible, and that converting macroeconomic data on financial asset flows to a measure of household spending makes improper assumptions.² Nonetheless, the findings remain a well-publicized data point in media coverage, staking out the upper bound of the K-shaped consumption story.

Bank of America: More recent, and more E-shaped than K-shaped

Another private sector report, Consumer Checkpoint from the Bank of America Institute, began including summaries of spending growth by income in late 2024. The monthly report provides a near-real-time view, albeit limited to credit and debit card spending by Bank of America customers.³

While Moody’s pegs a sharp break after 2020, Bank of America did not observe a widening gap in their data until mid-2025, when a large separation in spending growth opened between lower-, middle-, and upper-income customers—arguably more “E” than “K” (Figure 2).

Spending by the lower third of cardholders actually shrunk in mid-2025 and remained nearly flat until early 2026, when Bank of America noted a slight narrowing of spending growth gaps in February.

Unlike Moody’s, the Bank of America reports are not estimates, but a straightforward account of actual spending by households. However, the view is limited to purchases on credit cards (presumably missing rent payments and some very large purchases) and the subset of Americans with cards from one particular bank, which might underrepresent certain regions of the U.S. and segments of the income spectrum.

Dividing households into thirds makes it hard to compare directly with Moody’s, which finds most of the action at the very highest incomes. The Bank of America terciles are roughly comparable, however, to a new data series from the New York Fed.

New York Fed: A subtle trend, if any

The New York Fed recently rebooted the consumer spending component of its quarterly Economic Heterogeneity Indicators, including separate measures for retail spending, gas stations, and groceries. The data come from a panel of 200,000 households surveyed by Numerator, a market research firm.

Like Bank of America, the New York Fed breaks households into thirds. Unlike Bank of America, they do not find wide dispersion in retail spending growth during 2025 (Figure 3; retail spending excluding cars). Over the five years of available data, the growth rates move in relative harmony.

The New York Fed’s results breaking out gas and groceries are similarly bunched through most of those series, although food and beverage spending growth was notably higher for low-income households in 2025.

If there is a K-shaped trend since the pandemic, it is subtle. Slightly faster spending growth, on average, for higher- and middle-income households emerges over time as higher cumulative growth in nominal spending (Figure 4). This cumulative view shows these gaps widened primarily in 2023 and have been mostly stable since.

Nominal spending growth since 2020 ranges from 29 percent for low-income households to 36 percent for high-income ones. This is much less than the 62 percent nominal growth in personal outlays reported by Moody’s for top-10-percent households during roughly the same period. Spending growth for the top, middle, and bottom also remain much more tightly bunched than the nearly 2-to-1 difference Moody’s reports between these high-income households and all other groups.

Consumer Expenditure Surveys: Lagging data, nothing K-shaped

In the last week of 2025, the U.S. Bureau of Labor Statistics released its Consumer Expenditure Surveys (CE), the annual report on household spending. The CE arrived two months late because of the fall’s federal government shutdown.

Unlike the primary government measure of spending in the U.S. economy, personal consumption expenditures (PCE), the CE is survey-based—and therefore able to report straightforward results by quintiles and deciles of household income.⁴ However, the CE comes with caveats about what is in (“out of pocket” spending reported by households on goods or services, including rent) and what is out (mortgage and auto loan principal payments; spending on households’ behalf by employers, government, or nonprofits—especially on health care).⁵ The CE has also diverged in recent decades from expenditures captured by the PCE, with the CE-to-PCE ratio declining over time to less than 60 percent today.

The CE’s comprehensive, survey-based approach also comes with a substantial lag: It captures household spending only through the end of 2024. With these various caveats established, the latest CE results offer little support for the conventional wisdom of a consistent, K-shaped pattern in recent years. From 2022 through 2024, as CE spending growth by households slowed overall, the results by income group defy easy summary. But one thing they are not, is K-shaped (Figure 5).

Spending growth by households in the 21st to 40th percentile of income was notably sluggish in 2024. However, the lowest-income U.S. households in the survey increased their spending nearly 4 percent—more than any other quintile and the only group to have faster spending growth than the prior year.⁶ Meanwhile, spending growth by the highest-earning households slowed sharply. CE households in the top 10 percent actually spent less in 2024 than the year before (even before adjusting for inflation).

The CE confirms high-income households account for a disproportionate share of spending, though much less lopsided than the headline-making numbers from Moody’s. The top 10 percent of households in the CE accounted for 23 percent of spending (Figure 6).

In 2024 the highest-earning 30 percent of households accounted for more than half of out-of-pocket consumer spending in the CE. The bottom half of households by income accounted for less than 30 percent.

Work in progress: A CE-PCE hybrid

A “prototype” U.S. government effort aspires to merge the distributional findings of the CE survey with the expenditure data from the PCE. The method, titled “Distribution of Personal Consumption Expenditures,” maps CE to PCE categories and imputes household expenditures for PCE categories not included in the CE (such as owner-equivalent rent and employer-paid health insurance).

Results are only calculated through 2023, and the distribution of spending looks a lot like the CE-only results in the pie chart in Figure 6. The top 10 percent comprised 25.7 percent of spending in 2023. This concentration is just slightly more than 2020 (25.0 percent) and has changed very little since the earliest data in the series (the year 2000, 26.3 percent). The same is true for deciles across the household income spectrum. Per this experimental method, the concentration of spending has not turned K-shaped since the pandemic, and in fact appears highly stable across decades.

A final thought: Spending, income, and wealth

For each measure above, households are stratified by income. This is a natural consequence of the available data; from credit card issuers to market research firms to U.S. government surveys, income is the common denominator.

However, wealth—rather than income—increasingly distinguishes the most well-to-do U.S. households. According to the Federal Reserve Board, household wealth has become increasingly concentrated among the top 0.1 percent. And there is ample evidence that wealth (independent of income) empowers spending, with asset-rich individuals able to borrow against their holdings and fund very high levels of spending.

The country’s 1,100-plus billionaires are surely underrepresented in spending surveys. The Fed’s Survey of Consumer Finances (an essential component of the Moody’s measure) oversamples to attempt to ensure adequate representation of the very wealthy. However, distributional results are still presented by income, which could distort insights on spending by high-wealth (but lower-income) households.

Data on spending-by-wealth might provide a complementary view of the scale—and the shape—of any divergence in household spending. For now, the available data do not align to tell a clear, K-shaped story.

Endnotes

1 Moody’s derives the estimate by combining the Federal Reserve’s macroeconomic data on flows of financial assets with the Fed’s survey data on household wealth holdings. This process yields a rate of household savings for each income category, which is converted to dollars of disposable income. The residual “personal outlays” after savings are presented as a measure of consumer spending. Reports that characterize the Moody’s findings as “retail spending” are inaccurate, however, as the measure reflects extensive expenditures outside of the retail channel.

2 The Dallas Fed used a variation of the Moody’s methodology to calculate “consumption concentration” for a recent monetary policy modeling exercise. The authors found the top 20 percent of households comprised an average of 57 percent of consumer spending between 2020 and 2025 (similar to Moody’s calculation of a 59 percent share for this top quintile in third quarter 2025). In contrast to the dramatic increases found by Moody’s, however, the Dallas Fed researchers found this share was already very high (53 percent) by the 1990s and concentration had increased relatively little since then.

3 Per Nilson Report data summarized by Motley Fool Money, Bank of America is the No. 1 U.S. debit card issuer by purchase volume and No. 5 credit card issuer by outstanding balance.

4 Per BLS, the CE is “the only federal government data collection effort to obtain information on the complete range of consumers’ expenditures, income, and demographic characteristics, in the same survey, directly from consumers.” It consists of an interview survey (~20,000 households) primarily designed to capture large or recurring expenditures, and a diary survey (~11,000 households) to collect data on frequently purchased items. The CE refers technically to “consumer units,” which are largely analogous to households and we use the term “households” in this article.

5 To BLS statisticians, the principal payment on a secured loan is not consumption, but a conversion of one form of savings into another.

6 The stronger relative spending growth by low-income consumers isn’t obviously positive news for households. It could be driven, for example, by rising prices for household necessities like food that comprise a greater share of their basket. However, a deeper look shows that the share of income that low-income households spent on groceries fell in 2024. Total out-of-pocket expenditures also decreased as a proportion of income—this was true for all households, but most of all for the lowest-income quintile.