Skip to main content

Telling the nuanced story of economic inequality in the U.S., layer by layer

Minneapolis Fed economists find widening gaps—while government plays a large and growing role in countering them

October 13, 2023


Jeff Horwich Senior Economics Writer
Differently colored, people-filled layers diverge to demonstrate economic inequality
Cara Ewing/Minneapolis Fed

Article Highlights

  • Over 50 years, highest wage earners have pulled away dramatically from the middle, while gap between middle and bottom grew slowly
  • Median household earnings have approximately doubled in real terms, but fell further behind rich households; poor households and many male workers lost ground
  • Government redistribution has meaningfully slowed—but not halted—rising income and wealth inequality
Telling the nuanced story of economic inequality in the U.S., layer by layer

“The data is absolutely clear. Over the past 40 years, the wealthy have gotten wealthier. … Just look at the facts.”
Remarks by President Biden on the Economy (September 2021)

“Income inequality is not rising. It has in fact fallen. … In postwar America, we have experienced strong and widely shared prosperity.”
— Gramm, Ekelund, and Early, The Myth of American Inequality (2022)

With many credible ways to measure economic outcomes, tracking the story of inequality can feel like a Choose Your Own Adventure.

Given that no single number or figure can capture the level or the trend, a new staff report from the Minneapolis Fed offers a multifaceted analysis of economic inequality, based upon 54 years of data from five ongoing economic surveys.1More Unequal We Stand? Inequality Dynamics in the United States 1967–2021” updates the economists’ previous analysis to include another 15 years, covering the Great Recession (2007–9) and the COVID-19 recession (2020).

Taking the findings in full, “the general notion that inequality is widening is broadly correct,” said Minneapolis Fed Monetary Advisor Jonathan Heathcote. “But it’s also true that there are lots of different dimensions to inequality. You can look at inequality in earnings, in income, in wealth, or in consumption, and you can look at it for individuals or for households.”

“The economic research is very fragmented,” said Minneapolis Fed Monetary Advisor Fabrizio Perri. “We said, let’s look at everything, in a systematic fashion.”

In their full paper with co-authors Giovanni Violante and Lichen Zhang, the economists unfurl their analysis over the course of 50 charts. Here is a brief introduction to their findings.

Wage inequality widens and the highest earners pull away

Payment per hour worked—also known as wages—is a natural first cut on inequality. It is arguably the most simplistic angle, but it matters: Aside from a sliver of the super wealthy, wages and salaries make up the largest component of American incomes, by far. The size of a paycheck is a primary benchmark people use to chart whether they are keeping up or falling behind.

One clear story emerges from comparing the top with the middle, and the middle with the bottom (Figure 1). The wages of the highest-paid 10 percent of workers have pulled away, especially through the 1990s. This ascent has been less steep in recent years, with these highly paid workers now making about 4.5 times the median.

Loading chart 1...

A further slice of the data shows this widening from the top has been largely driven by worker salaries in the top 2 percent.

Meanwhile, the wage gap between a worker in the middle and a worker near the bottom has grown over the past 50 years, but much more slowly.2

Over 50 years, an important driver of the rise in wages at the top has been the wage premium associated with a college degree. It rose steeply from 1980 through the mid-2000s, especially for men (Figure 2). For the past 20 years, however, this college wage premium has plateaued.

Loading chart 2...

While the college wage premium is clearly important, most of the widening of wage inequality cannot be explained using the available survey data. The sources are, in the statistical parlance, “in the residual”—not obviously associated with demographics, education, or a common, skill-based method of job classification.

The stagnation of men

Another view of the data moves from wages to earnings (wages multiplied by hours worked).

“A big theme in a lot of the debate on widening inequality is the idea that in the middle of the income distribution, people have been treading water—that there’s been no growth,” said Heathcote. “And if you just look at male earnings, that’s the story.”

While earnings for top-earning men have more than doubled, male median earnings are largely flat after inflation (Figure 3). Earnings for low-income men have fallen.

Loading chart 3...

The falling earnings of low-earning men reflect lower real wages, but the larger factor is a drop in hours worked. These drops accelerate with each recession, rarely recovering to their pre-recession levels (Figure 4).

Loading chart 4...

The great gender equalization

While the earnings of many men have not grown much over the past 50 years, labor market outcomes for women have been better. Women have closed much of the gap with men in both wages and hours worked (Figure 5).

Loading chart 5...

In 1967, the average working woman earned an hourly wage that was 62 percent that of the average working man, and worked 37 percent of the hours. By 2021 both those ratios have increased to around 80 percent, though the pace of gains has slowed since the 1990s.

Median household income is (in fact) rising, while poorest households regress

The wage trends of individual men and women come together when they live together, sharing income and expenses.

“When we think about ‘standard of living,’ we’re mostly talking about what’s happening to households,” said Heathcote. The economists use a standard methodology to “equivalize” the earnings of different-sized households to compare them side by side (Figure 6). Doing so reveals that median household earnings have nearly doubled since 1967 (after accounting for inflation).

Loading chart 6...

Households below the median have made less progress, with earnings setbacks around each recession and especially the Great Recession. For the poorest households, the past 50 years have seen real earnings fall.

Households near the top saw earnings rise between 2 and 3 times. This measure does not reflect nonlabor income from financial holdings or business ownership. When the economists account for this nonlabor income, the level of inequality at the top is higher throughout the sample period, although nonlabor income does not appear to contribute much to the growth of U.S. income inequality.3

Adding nonlabor income does not change the trajectory for middle- or low-income households, who rely almost entirely on labor earnings. Median household income rises over time, but “a household at the bottom 20 percent of the income distribution now makes exactly the same as it was making 50 years ago, in real terms,” said Perri. “That is super striking. Society has evolved a lot in 50 years—become wealthier. But a household at the bottom makes the same.”

Government redistribution matters more than ever (and it matters a lot)

Atop wages and household income, federal and state governments layer a progressive income tax structure and government transfers, such as food stamps, unemployment insurance, stimulus checks, and discounted services for low-income Americans. These in-kind services—principally Medicare and Medicaid—are not well-quantified in survey data.4 Even without them, government intervention observably evens out the dispersion in household earnings, yielding an upward trend for all households in “disposable” income (Figure 7).

Loading chart 7...

The lasting damage done to poorer households’ disposable income during hard times is limited by government taxes and transfers. These “automatic stabilizers,” which policymakers have extended and deepened during the two most recent recessions, seem to be working.

“Transfers are much more important than they used to be, especially during recessions,” said Perri. “The government has become a much more prominent force in providing insurance.”

“You still see inequality at the bottom going up, because the government is not making up 100 percent of the losses in labor earnings that people are suffering at the bottom, but it’s much smaller,” said Heathcote. “The COVID recession is the one where the government steps in at an unprecedented scale, and you actually see inequality at the bottom going down.”

After taking government redistribution into account, 50 years ago an average household in the top 10 percent had a disposable income roughly 7 times that of the average household in the bottom 20 percent.5 Today the ratio of the top to the bottom is just under 11—less dramatic than the rise in market income, but still rising (Figure 8).

Household income measures over 50 years (1971–2021)
Note: Figures reflect the averages of the respective ranges, presented in 2021 dollars. Market income includes labor and nonlabor income; disposable income adjusts for estimated government taxes and transfers. Dollar figures displayed are rounded to nearest hundredth; changes and ratios reflect nonrounded figures. Households have been equivalized to standardize across household sizes.
Source: "More Unequal We Stand? Inequality Dynamics in the United States, 1967–2021," (underlying graph data converted to chart format), data from Current Population Survey Sample B.
  Real household market income   Real household disposable income
(In 2021 dollars) 1971 2021 Change   1971 2021 Change
Top 10% $77,500 $204,100 163%   $61,000 $149,000 145%
Middle 10% $23,800 $43,000 80%   $20,600 $38,200 86%
Bottom 20% $5,600 $6,800 22%   $8,500 $13,700 61%
Top-middle ratio 3.26 4.75 46%   2.96 3.90 32%
Middle-bottom ratio 4.25 6.32 49%   2.42 2.79 15%
Top-bottom ratio 13.90 30.10 116%   7.20 10.90 52%

The economists note that until recently, household “income pooling” was an important force in reducing household inequality. This occurred mainly through low-income women marrying men with higher incomes and stable employment. They document how the household now plays a diminished insurance role, as male and female earnings have converged, more women head single-parent households, and it is increasingly likely for high-earning women and men to form households together.

Meanwhile, government taxes and expenditures have become increasingly important in slowing the growth of household earnings inequality. Over the last half century, the economists find the weakening role of household income pooling has been offset almost 1 for 1 by the rising role of the government.

More wealth to the wealthiest

The paper affirms the steady concentration of wealth in the hands of fewer Americans. Data from the Survey of Consumer Finances and Panel Study of Income Dynamics show the wealthiest 10 percent of households held more than 70 percent of wealth in 2019, up from around 65 percent in the late 1980s.

For real estate and financial wealth, the appreciation of already-held assets—rather than increased savings rates—appears to be driving growing wealth inequality.

American households below the median hold little to no wealth. The new data—encompassing years before, during, and after the Great Recession—show how wealth inequality between the top and middle has been highly sensitive to fluctuating housing values of the middle class. “Whether it’s housing that’s booming or the stock market that’s booming is going to have an important influence on the shape of the wealth distribution,” said Heathcote.

Perri notes that for real estate and financial wealth, the appreciation of already-held assets—rather than increased savings rates—appears to be the most important driver of U.S. wealth inequality. Consider that since 1987, U.S. home prices have almost doubled after inflation; the real value of the U.S. stock market has increased by a factor of 6.6

Fascinatingly flat: Inequality in consumption

Even as the richest households outpace others in wages, household earnings, and wealth, their spending has not grown at a faster pace. Anecdotal evidence of mansions and mega-yachts notwithstanding, consumption inequality remains flat as a statistical matter. The top 10 percent of households spend only about twice as much as the median, a ratio that has not fluctuated much in the 30 years of survey data available. A similarly stable spending ratio of about 2-to-1 prevails between the middle and the bottom.

The top 10 percent of households spend about twice as much as the median, a ratio that has not fluctuated much in 30 years.

The flatness of consumption inequality—amid the other rising indicators of inequality—remains a topic of special interest to Perri, who first documented it 20 years ago in a much-cited paper with economist Dirk Krueger. The cause remains up for debate. One early theory holds that rich people, as they get richer, self-insure more against risk, having reached a point of diminishing returns of spending on themselves. Economists also suspect that wealthy households underreport their spending on surveys.

“I think those factors are still there, but something is different now,” said Perri. “I think what’s going on is that people live longer, and the rich people don’t want to consume too much. They just want to save for later—for living more decades, for medical expenditures, and for bequests.”

Factors at the bottom likely matter, too. Easier access to credit allows lower-income households to sustain more spending in hard times and otherwise extend household budgets beyond their day-to-day capacity.

Inequality appears to be rising more slowly. But it has risen a lot and is still rising.

The wage, income, and wealth measures analyzed by Heathcote, Perri, Violante, and Zhang show inequality continued to grow through the past half century. In general, the economists observe that inequality has not widened as quickly in recent decades as it did through the ’80s and early ’90s.

A slowing trend is not a downward trend, however, and the change over 50 years has been substantial. Household inequality has soared from the top, where the top 10 percent of households have seen market income grow 163 percent in real terms (Figure 9). Growth in market income for households in the middle has been half as fast. Income gains for the poorest households have been one-eighth of those at the top.

Loading chart 9...

The highest-income households now earn almost 5 times as much as a household in the middle and 30 times as much as a household near the bottom, far higher ratios than 50 years ago. The middle-bottom ratio has also grown, but less—roughly from 4 to 6. (See data from Figure 8, above.)

Figure 9 shows that progressive taxation and government transfers have reduced the spread in household disposable incomes, lifting the experience of households at the bottom and modestly compressing the top. “More Unequal We Stand?” highlights how public policy to reduce disparities has slowed but not stopped the growing inequality that households experience. The data for the paper terminate amid the unprecedented waves of government support for incomes during the COVID pandemic. A future update would reveal the lasting effect of COVID stimulus, if any.

How much inequality should we tolerate? Is it sufficient to slow the growth of inequality, but not shrink it? Are the economic trade-offs too steep? Those are questions for society once economists lay out the facts.

“There’s no real right answer to how much inequality you want to reduce,” Heathcote said. “That’s up to the voters, and the political system can decide.”

Surveys or administrative data?

Economists studying inequality can work with administrative data (official government or private records) or the results of economic surveys. Each has advantages and limitations.

In this case, the economists make use of five extensive, long-running surveys covering income, consumption, and wealth. The surveys are publicly available to all researchers, allowing findings to be tested or extended. They cover individuals and households, and link them together, following the same subjects from one survey to the next.

The survey questions are extensive, including information on hours worked, which is essential to understanding the connection between wages and earnings. “You can really build up a rich picture of the household,” said Jonathan Heathcote. “The downside of surveys is that, well, these are people answering questions.” In their staff report, the economists discuss groups that might be underrepresented and forms of income that might be underreported. They run calculations to compare aggregate survey outcomes with macroeconomic data and perform adjustments to compensate for potential shortcomings.

Administrative datasets can be more comprehensive, but less rich—especially because it has often been impossible to link households across different datasets. They can also be onerous for researchers to access. However, the Minneapolis Fed’s new Income Distributions and Dynamics in America project—linking millions of household tax and census records between 1998 and 2019—represents one major, recent development in generating publicly available administrative data to study U.S. income inequality.


1 The researchers use the American Community Survey (U.S. Census Bureau), Consumer Expenditure Survey (Bureau of Labor Statistics), Current Population Survey Annual Social and Economic Supplement (U.S. Census Bureau), Panel Study of Income Dynamics (University of Michigan), and Survey of Consumer Finances (Federal Reserve Board). See the final section “Surveys or administrative data?” for more information on the strengths and challenges of survey-based research.

2 The economists note that the ratio between the 50th and 10th percentiles looks a little less flat when viewing men and women separately. The combined curve masks the effect of women entering the workforce in the 1970s and 1980s. Women have a lower level of within-gender wage inequality; as they have come to comprise a higher portion of the working population, their more-equal wages offset the rising wage inequality observed for men and women separately. For more, see Figure 7 from the original paper.

3 The economists note that nonlabor income is likely underreported on surveys. To offer a more accurate estimate, they “rescale” their survey measures of nonlabor income using ratios indicated by aggregate macroeconomic data. For measures of “market” income, including nonlabor income, see Figures 8 and 9 in this article and Charts 16 and 17 in the original staff report.

4 In their paper, the economists provide estimates of household income that are rescaled to reflect in-kind transfers, including federal data on Medicaid spending. As expected, this modestly reduces inequality by raising incomes at the bottom. However, the economists regard these estimates as “overly optimistic,” since many Medicaid expenditures do not flow to households but to medical providers.

5 For some measures, as seen in Figures 8 and 9, the economists compare 10-percent top and middle bands with the average of the bottom 20 percent of households. This wider band arguably provides a more informative view of low-income households by minimizing the extent to which the insights are shaped by households with near-zero income. This choice would tend to understate the growing gap between the top/middle and bottom.

6 This 1987 base year is the start date of the S&P/Case-Shiller U.S. National Home Price Index. Stock market value is measured by the Dow Jones U.S. Total Stock Market Index. The nominal index values were deflated using the change in annual CPI from 1987 to 2023, via the Minneapolis Fed’s online inflation calculator.

Jeff Horwich
Senior Economics Writer

Jeff Horwich is the senior economics writer for the Minneapolis Fed. He has been an economic journalist with public radio, commissioned examiner for the Consumer Financial Protection Bureau, and director of policy and communications for the Minneapolis Public Housing Authority. He received his master’s degree in applied economics from the University of Minnesota.