Has the Rise of the One Percent Really Been Debunked?

Everyone knows that inequality has gotten out of hand in the United States. Thanks largely to the work of three now-famous economists—Thomas Piketty, Emmanuel Saez, and Gabriel Zucman—it’s probably one of the most widely accepted facts in modern American life. Since the early aughts, they have meticulously documented the rate at which the richest have pulled away from the rest. Their research transformed domestic politics, leading President Barack Obama to declare inequality the “defining issue of our time,” and turning the one percent into a shorthand for excessive wealth and power.

But what if this “fact” was never true?

That’s exactly the claim that Gerald Auten and David Splinter, economists at the Treasury Department and Congress’s Joint Committee on Taxation, respectively, make in a paper published last September. The massive rise in income inequality since the 1960s, they argue, is mostly a statistical illusion based on a series of methodological errors. Piketty, Saez, and Zucman found that the top one percent’s share of after-tax income rose from 9 percent in 1960 to 15 percent in 2019. But, according to Auten and Splinter, the one percent’s share of income has actually remained basically unchanged.

The paper caused an immediate stir among economists and pundits. “Can a single self-published paper really refute decades of work by three famous economists?” wrote the influential economist Tyler Cowen in Bloomberg. “The answer—with qualifications—is yes.” In the Financial Times, the columnist Chris Giles asked, “How would you feel if you found out that U.S. income inequality had not risen over the past 60 years?” And The Economist, reviewing the debate, concluded that “the idea that inequality is rising is very far from a self-evident truth.”

When I first heard about the new paper, I assumed it would convince me, at the very least, that inequality had risen less than I thought—that the reality was somewhere between the two groups’ estimates. But the deeper I dug into the debate, the more I felt that both teams were underestimating the extent of inequality in America. Both are limited by assumptions and definitions that are standard in the economics profession but contrary to how regular people think about inequality—or, for that matter, money itself.

The biggest point of contention between the two camps revolves around “unreported income,” more commonly known as tax evasion. Tax returns are the best data source available for studying income distributions, but they’re incomplete—most obviously because people don’t report all of the income that they’re supposed to. This information gap requires inequality researchers to make some educated guesses about how unreported income is distributed, which is to say, about who is evading the most taxes. Piketty, Saez, and Zucman assume that it’s the people who already report a lot of income: Think of the well-paid corporate executive who also stashes millions of dollars in an offshore account. Auten and Splinter, by contrast, assume that those who evade the most taxes are people who report little or no income: Think plumbers or housekeepers who get paid in cash. They believe, in other words, that members of the 99 percent are a lot richer than they look.

Both teams base their estimates on the results of random IRS audits. The problem is that those audits are woefully limited, so the researchers must turn to studies by tax-evasion specialists who use fancy statistical techniques to account for the gaps in the audits. Only a few such studies even exist, which may explain why both teams end up heavily relying on the same 2010 paper from the National Tax Journal by the economists Joel Slemrod and Andrew Johns. Somehow, each side claims that the paper—which is based on data going back to 2001—supports their conclusion and is being misread by the other group. In other words, whether inequality has soared or stagnated comes down in large part to the interpretation of a single study using data from more than 20 years ago.

[From the June 2018 issue: The 9.9 percent is the new American aristocracy]

When I called Slemrod to find out which side was citing his work correctly, even he wasn’t sure. The two camps, he told me, refer to different sections of the paper that seem to point in different directions. Anyway, Slemrod said, it didn’t really matter, because his paper was out of date. Newer research gives what he considers a more accurate picture. He directed me to a 2021 study that found that the IRS audit data systematically undercounted tax evasion by the rich. The authors of that study—one of whom is Gabriel Zucman—calculate that the top one percent’s unreported income is actually about 50 percent higher than even the Piketty estimates, once you account for the bias in the data.

Auten and Splinter have critiqued this study, and its authors have, inevitably, critiqued the critique. Ultimately, which side you believe will depend on whether you accept Auten and Splinter’s suggestion that, even as IRS enforcement has gotten weaker and exotic tax-avoidance strategies have proliferated, the richest people in America have for some reason gotten better about paying their taxes. “To believe their results, you have to believe the top one percent has gotten much more compliant relative to the bottom 99 percent,” Zucman told me. “It doesn’t pass a basic smell test.” (Because Auten and Splinter are federal employees, they were not authorized to speak with me for this article.)

These sorts of technical disagreements over wonky minutiae are the sine qua non of any good academic dispute. But they don’t, on their own, explain the full divergence between the two research teams. Much of the remaining disagreement ultimately comes down to the more philosophical question of how to define “income” in the first place.

To take the true measure of inequality, economists need a way to account for all the income and expenses that don’t show up on people’s tax returns. The method that Piketty, Saez, and Zucman pioneered, and that Auten and Splinter follow, was to take the gross domestic product—a measure of all of the spending in the national economy every year—and figure out who exactly is receiving how much of it. (Technically, they use something called gross national income, which is a close cousin of GDP.) The benefit of this approach is that nothing gets left out. The drawback is that, well, nothing gets left out. GDP measures the total production of an entire economy, so it includes all sorts of expenditures that don’t seem like income at all.

Much of the difference between the authors’ estimates of inequality hinges on how they treat government spending on things that benefit the public at large, such as education, infrastructure, and national defense. Because this spending is part of gross national income, it must be allocated to someone in order for the math to work out. Piketty, Saez, and Zucman take the view that this stuff really shouldn’t be considered income, so they allocate it in a way that doesn’t change the overall distribution. Auten and Splinter, however, argue that at least some of this money should count as income. Citing research indicating that education spending tends to disproportionately benefit lower- and middle-income kids, they decide to allocate the money in a way that increases the bottom 99 percent’s share of income—by a lot. Austin Clemens, a senior fellow at the Washington Center for Equitable Growth, calculates that in Auten and Splinter’s data set, a full 20 percent of income for those in the bottom half of the distribution “comes in the form of tanks, roads, and chalkboards.”

The teams likewise disagree over how to treat government deficits, which, for methodological reasons, likewise need to be allocated to someone. Piketty, Saez, and Zucman, perhaps sensing the absurdity of the whole exercise, again take a distributionally neutral approach. Auten and Splinter, meanwhile, argue that deficits have historically been more likely to be paid off by taxing the rich, so they allocate the money such that it reduces the top one percent’s share of income by nearly half a percentage point. Neither of these individual choices, in isolation, is enough to drastically alter either team’s conclusion. But the dueling data sets are littered with dozens of such judgment calls. When taken all together, they add up to two very different pictures of inequality in America.

More revealing than any of these points of disagreement, however, are the ways in which the papers agree on how to define income. Take health-care benefits. Economists generally count both employer-sponsored health insurance and government programs such as Medicaid as income. In reality, people value these benefits far less than they would value the cash equivalent. But even putting that aside, the way this supposed income is defined—by dividing the total cost of the program by the number of recipients—can make it appear as if people are getting richer when they clearly are not.

[Rogé Karma: Why America abandoned the greatest economy in history]

Consider the following thought experiment, proposed to me by the Nobel Prize–winning economist Angus Deaton: A private-equity fund buys up a bunch of hospitals and immediately raises prices across the board, increasing the cost of health care without changing its quality. According to your typical economist, everyone receiving that health care now has a higher income than they did before. Or, Deaton said, imagine the reverse: The United States, which has the highest per-person health-care costs in the developed world, embarks on a heroic effort to lower its costs to the same level as Switzerland, the second-highest spender, without compromising quality. If that happened, it would appear in the income data as if Americans, on average, had each lost nearly $9,000 a year.

This turns out to be not too far from how both inequality papers treat health-care spending. Over the same time period that poor Americans’ life expectancy stagnated and then declined, a massive boost in spending on Medicare, Medicaid, and employer-provided insurance has made it look as if those people have gotten significantly richer. If anything, both teams may be putting a thumb on the scale in favor of finding less inequality.

The deeper you get into how GDP is actually calculated and allocated, the more you feel as though you’ve fallen through a wormhole into an alternate dimension. Let’s say you own a house. Government statisticians imagine that you are renting out that house to yourself, calculate how much money you would reasonably be charging, and then count that as a form of income that you are, in essence, paying yourself. This “imputed rent” accounts for about 9 percent of all GDP, or more than $2 trillion. Or suppose you have a checking account at a major bank. Statisticians will calculate the difference between what the bank pays you in interest on that account (usually close to nothing) and what you could have earned by investing that same money in safe government bonds. That difference is then considered the “full value” of the benefits you are receiving from the bank—above and beyond what it actually charges you for its services—and is therefore considered additional income for you, the depositor. All of these choices have some theoretical justification, but they have very little to do with how normal people think about their financial situation.

There’s an old joke among economists that goes something like this: A policeman spots a drunk man searching for his keys under a lamppost and offers to help. After a few fruitless minutes, the officer asks the man where, exactly, he dropped his keys. “The tunnel over there,” the man says. “Then why look under the lamppost?” the officer asks. “Because the light is better here,” replies the man.

It’s hard to read through these inequality papers—and the responses to them, and the responses to the responses—without feeling a little like the policeman. You find yourself caught in the middle of highly technical debates over the accuracy of audit studies, or who benefits most from education spending, or how often people roll over their retirement accounts (don’t even get me started on that one), all of which are supposed to tell you what has actually happened to income inequality over the past half century. But it turns out that the data being argued over are missing a lot of what you’re trying to measure, and the definition of income being used is one that most ordinary people wouldn’t even recognize. Both teams are claiming to have found something under the lamppost, but it might not be a set of keys.

[Oren Cass: The labor-shortage myth]

Luckily, we don’t have to rely on one data source or statistical technique to know that inequality has risen. The Congressional Budget Office finds that from 1979 to 2014, the top one percent’s share of after-tax income went up 6 percent. The Fed’s Survey of Consumer Finances and a 2023 study using tax data find similarly large increases when it comes to wealth. Then there’s plain old common sense. In 1982, the year Forbes released its first-ever annual list of the 400 wealthiest Americans, the shipowner and real-estate tycoon Daniel Ludwig topped the list with a net worth of about $6 billion in today’s dollars. In 2023, the top three wealthiest individuals, Elon Musk, Jeff Bezos, and Larry Ellison, were worth $251 billion, $161 billion, and $158 billion, respectively. Inequality also manifests in ways that aren’t obviously economic: From 2001 to 2014, for example, the life expectancy of the wealthiest Americans rose by about three years—the equivalent of curing cancer—while the poorest experienced no gains. Perhaps only an economist could figure out a way to deny that the richest people are even richer than they used to be.

Even if the case that inequality has remained flat isn’t convincing, exactly how much it has risen remains up for debate. (In fact, a growing body of research suggests that low-wage workers have made up a ton of ground since the pandemic, thanks to a tight labor market.) In the meantime, it’s worth remembering that economics is often far closer to the squishy humanities than to the hard sciences. Economic analysis can be a helpful corrective against human bias, but when its assumptions are divorced from the way people experience the world, it can just as easily lead us astray. Keep that in mind the next time a new paper comes out claiming to overturn the conventional wisdom.