In my writing and in my talks, I spend a great deal of time on a simple but persistent problem: the gap between what we collectively believe about the economy—our social narrative—and what the economy is actually doing in the objective statistical data I analyze every day. That gap shifts over time and across issues, and it matters. It drives policy mistakes, investment mistakes, and household mistakes. In a world already saturated with populist reflexes, narrative errors don’t merely mislead—they can become politically dangerous.

We might hope that economists and other statistically oriented social scientists would help close that gap. Too often, we do the opposite. Social scientists are human. We have biases, incentives, reputational concerns, and tribal instincts like everyone else. Expertise does not eliminate those forces; sometimes it simply gives them better tools.

Still, when a well-known economist amplifies a claim that is plainly inconsistent with the evidence, it’s troubling—because the public gives economists an aura of authority we often do not deserve. The latest example is the now-familiar claim of a “K-shaped recovery,” promoted by Mark Zandi, chief economist at Moody’s Analytics.

Recently, Zandi posted a graph asserting that the “share of total outlays going to those in the top 20% of the income distribution—making over $175,000 per year nationwide—increased to nearly 60% in the third quarter of 2025,” adding that this was “another new high in the data we have constructed back to 1989.” The chart was widely shared and praised, reinforcing a narrative that neatly fits the current populist framing embraced across the political spectrum.

The problem is not that the claim is controversial. It is that it is almost certainly wrong—both in level and in trend. There is no credible public evidence that the top 20% of households account for 60% of all consumer spending. Nor do the available data support a recovery story driven by an unprecedented concentration of consumption at the top.

Zandi’s series appears to be derived from proprietary internal work that has not been publicly documented. Without transparency regarding definitions, data sources, or methodology, it’s impossible to replicate or evaluate the estimate directly. But we do have high-quality public benchmarks that any serious attempt to measure spending shares by income should resemble.

Start with the Bureau of Labor Statistics’ Consumer Expenditure Survey (CEX). The CEX is the primary official source for measuring how U.S. households spend their money. It collects detailed household-level data on out-of-pocket expenditures—what consumers actually pay for goods and services—and it breaks those expenditures out by income quintile. A quintile simply divides households into five equal groups based on income; the top quintile represents the highest 20% of earners.

The CEX is not a complete measure of national consumption. It largely captures out-of-pocket spending and excludes many expenditures paid on households’ behalf, such as employer-provided health insurance or certain public benefits. As a result, aggregate CEX consumption is smaller than the Bureau of Economic Analysis’ Personal Consumption Expenditures (PCE) measure, which is used in calculating GDP. PCE includes third-party payments and broader conceptual adjustments. But for distributional purposes, the CEX has a critical advantage: it directly observes spending by income group.

Using the CEX data, the share of spending attributable to the top 20% of households does not even approach 60%. In fact, the top quintile’s share peaked around 1999—just before the dot-com crash and the 2001 recession—and has generally trended downward since then. In recent years, it has averaged in the high-30% range. That is nowhere near the high-50s.

Yes, the CEX understates total consumption relative to PCE. But that fact makes the 60% claim even harder—not easier—to justify. For the top quintile’s share of total consumption (PCE) to reach 60% while their share of observed out-of-pocket spending remains far lower, the “missing” portion of consumption—third-party payments, imputed items, and other conceptual differences—would have to be overwhelmingly concentrated among high-income households.

That is highly implausible.

Much of the spending not captured in CEX consists of healthcare expenditures paid by employers or public programs—categories that are not disproportionately skewed toward the top quintile. Reconciling a roughly high-30% out-of-pocket share with a 60% total consumption share would require an extraordinary and economically inconsistent reallocation of the unobserved components.

The income data tell a similar story. The Congressional Budget Office publishes an annual distributional analysis estimating household income shares after taxes and transfers. This series incorporates extensive administrative data, including tax filings, and measures income net of federal taxes and inclusive of government benefits.

According to the CBO, the top 20% of households receive roughly half of total after-tax, after-transfer income, with that share peaking around 1999. That figure has fluctuated over time but has not exploded to levels that would remotely justify a 60% consumption share.

This matters because savings behavior is well understood. The marginal propensity to save rises with income: higher-income households save a larger fraction of each additional dollar they earn than lower-income households. If the top quintile earns about half of total income, we would expect them to account for less than half of total consumption, not substantially more. For the top 20% to consume 60% of all expenditures, the bottom 80% would need to be saving at higher rates than the top—or the top would need to be dissaving at extraordinary levels relative to everyone else. Neither scenario is supported by the data.

Perhaps there is some alternative explanation. But without transparent definitions, methodology, and reconciliation to official data sources, it is difficult to see how Zandi’s claim survives basic scrutiny.

The more troubling question is not statistical but institutional. Was this simply an error? Did a team of capable economists fail to cross-check their internal estimates against well-established public benchmarks? Or does the incentive structure in modern economic commentary subtly reward narratives that confirm what audiences already think they “know,” even when the numbers are weak?

We can’t answer that question. But we can recognize the broader danger.

Bad narratives drive bad policies. When economists amplify claims that are inconsistent with the evidence, we do not close the gap between perception and reality—we widen it. And when the public sees economists as trusted arbiters of fact, that widening gap becomes especially corrosive.

If we are going to insist on being treated as experts, then we must behave like them. We must show our work. We must reconcile our claims to existing data. And we must resist the temptation to tell compelling stories that the numbers do not support. Economist, heal thyself.