What Drives Income Inequality? The Economics Behind the Gap

In 2023, the top 1% of U.S. wage earners captured 12.4% of all wages paid, up from 7.3% in 1979. Over the same period, the bottom 90% saw their collective share fall from 69.8% to 60.7%, according to Economic Policy Institute analysis of IRS and Social Security data.¹ Between 1979 and 2023, the top 1%'s annual earnings rose 181.7%; the bottom 90%'s rose 43.7%. That is not one economy pulling apart: it is two very different economies overlapping in the same statistics.

Income inequality has a single number attached to it in public debate, but five distinct mechanisms driving it beneath the surface. Conflating them leads to policies calibrated for the wrong problem. Here is what the data actually shows.

The headline number: what it says and what it doesn't

The U.S. Gini coefficient (the standard summary measure, where 0 is perfect equality and 1 is perfect inequality) stood at 0.481 based on the 2024 American Community Survey.⁵ That is among the highest of any wealthy nation. The number is real and comparisons to OECD peers are valid. But aggregate statistics like the Gini mask important structural details: whether inequality is concentrated at the top versus the bottom, whether it reflects labor income or capital income, and whether it has the same causes across different parts of the distribution.

The 90/10 ratio, comparing the wage at the 90th percentile to the wage at the 10th percentile, captures inequality across the broad distribution. Long-term analysis of BLS earnings data shows that workers at the top have consistently outpaced those at the bottom since the 1980s, with the ratio widening from roughly 4:1 to over 5:1 across that period.²

Driver 1: Skill-biased technological change and the college wage premium

Since the 1980s, computing and automation have simultaneously displaced routine cognitive and manual tasks (bookkeeping, assembly, data entry) while increasing demand for workers who can analyze, create, manage, and communicate. Economists call this skill-biased technological change (SBTC): technology that complements high-skill workers and substitutes for middle-skill ones.

The result is visible in earnings data. As of Q3 2024, workers with a bachelor's degree had median weekly earnings of $1,533 compared to $946 for workers with a high school diploma only, a premium of roughly 62%.² In 1980 that premium was about 39%. The gap more than doubled in four decades.

Research by labor economist David Autor at MIT documented the specific mechanism: technology hollowed out the middle of the skill distribution (the routine-task-intensive jobs that once provided stable middle-class employment) while leaving high-skill and low-skill, face-to-face service jobs intact. This "job polarization" shifted the labor demand curve in ways that compressed middle wages and expanded top wages simultaneously.

One important note from recent data: the college wage premium has stopped growing since roughly 2010, and some Federal Reserve research suggests it has slightly narrowed. SBTC remains powerful, but the composition of inequality may be shifting.

Driver 2: Globalization and trade exposure

The integration of China into global trade networks, accelerating after China joined the WTO in 2001, produced what economists Autor, Dorn, and Hanson termed the "China shock": a concentrated, persistent income reduction in U.S. communities heavily exposed to import competition in manufacturing. These were not aggregate losses distributed widely; they were local economic collapses in places whose economic identity was tied to specific industries.

Census Bureau American Community Survey data captures the geographic concentration of this effect.⁷ Manufacturing-intensive regions in the Midwest and South experienced above-average income decline and above-average opioid mortality in the 2000s through 2010s, a connection researchers have documented with increasing precision.

Globalization also increased income at the top: multinational corporations, finance professionals, and knowledge workers who could sell services globally saw demand for their skills rise substantially. The same forces compressed incomes in the middle and bottom while expanding incomes at the top, a double contribution to measured inequality.

Driver 3: The collapse of unions and the eroding wage floor

At their peak in the mid-1950s, unions represented roughly 35% of the private-sector workforce. They did not just negotiate wages for union members. By setting a benchmark in unionized industries, they compressed wages across the broader labor market, including in non-union firms competing for workers.

By 2024, the private-sector union membership rate had fallen to 5.9%, according to the Bureau of Labor Statistics.³ That is not just a loss of collective bargaining; it is the elimination of the wage floor and distributional compression that union density once provided. Research by economists Lawrence Katz and Alan Krueger estimates that declining unionization accounts for roughly 15 to 20% of the rise in wage inequality since 1980.

The federal minimum wage has also eroded substantially in real terms since its 1968 peak, when adjusted for inflation it was higher in purchasing power than today's $7.25 federal floor. Since minimum wages disproportionately affect workers at the bottom of the distribution, their erosion in real terms widens the 10th percentile's distance from the median.

Driver 4: Superstar dynamics and winner-take-all markets

In industries where the best performers can reach a global audience or a national market through technology (software, finance, entertainment, sports, management consulting), small differences in talent or reputation produce enormous differences in compensation. Economist Sherwin Rosen described this dynamic in 1981: if the best surgeon or the best software developer can serve millions of clients where once they could serve hundreds, the premium for being the best explodes.

This explains much of the income concentration within the top decile, specifically why the top 1% has pulled away from the top 10%. It is not that 1-in-10 workers became dramatically more productive; it is that the top 1-in-100 gained market reach that the 9-in-100 below them could not match.

CEO and executive compensation reflects a related dynamic. From 1978 to 2022, CEO compensation rose roughly 1,460%, while a typical worker's compensation rose 18% over the same period, according to EPI analysis of published compensation data.¹ Whether this reflects genuine productivity differences or a breakdown in corporate governance is debated, but it is a real contributor to top-of-distribution concentration.

Driver 5: Capital income, r > g, and wealth concentration

Thomas Piketty's central argument in Capital in the Twenty-First Century is that when the rate of return on capital (r) exceeds the economic growth rate (g), wealth and the capital income it generates concentrate over time in the hands of those who already have it. This is an arithmetic observation: if wealthy households earn 5 to 7% annually on their wealth while the economy grows at 2 to 3%, the wealth-to-income ratio rises, and capital income becomes an ever-larger share of total income.

The Congressional Budget Office's distribution of household income analysis documents that capital income (dividends, capital gains, interest, rental income) is heavily concentrated at the top of the distribution.⁶ The top 1% of households receive a disproportionate share of capital income, which compounds year after year as portfolios grow.

Important nuance: for recent U.S. inequality, Piketty himself acknowledged that the dominant force has been labor income inequality (the rise of executive pay and high-skill professional wages) rather than passive capital accumulation. Both forces are real; their relative magnitude is contested and varies by time period.

Driver 6: Assortative mating

An often-overlooked contributor: highly educated, high-earning individuals increasingly marry each other. The share of U.S. married couples in which both spouses hold college degrees rose dramatically between 1960 and 2013. NBER research estimates that 10 to 16% of the rise in household income inequality is directly attributable to this increasing "assortative mating," not because individual earners became more unequal, but because high earnings became more likely to be pooled in the same households.⁴

A dual-income couple where both partners hold professional degrees and earn $120,000 each ($240,000 household) looks very different in the household income distribution than two single people earning $120,000 each. The concentration of high earners in high-earning households is itself a structural driver of measured household income inequality.

What the data doesn't show: the limits of the numbers

Income inequality statistics almost always measure pre-transfer, pre-tax income: what the market distributes before the government redistributes. After federal taxes and transfers, U.S. income inequality is lower than the market income figures suggest. The CBO's household income analysis consistently shows that progressive taxation and means-tested transfers compress the distribution meaningfully.⁶

The data also misses in-kind compensation: employer health insurance and retirement contributions grew as a share of total compensation, particularly for higher-paid workers, and are not captured in wage figures. This may overstate the growth of the top-to-bottom wage gap.

Finally, income inequality is not the same as consumption inequality or wealth inequality, and their trends do not always move together. A retired household with zero wage income but $800,000 in financial assets has a high consumption level and low measured inequality contribution.

What the five drivers suggest about remedies

Different causes require different tools. Skill-biased technological change is addressed over the long run by education and workforce retraining investments, raising the supply of workers with skills that technology complements. Union decline requires labor market policy. Superstar dynamics are difficult to address without taxing top incomes or strengthening market competition. Capital income concentration responds to wealth taxation, capital gains rates, and estate taxes. Assortative mating is not a policy lever at all.

Treating inequality as a single problem with a single cause, as is common in political debate on both sides, consistently produces policies calibrated for the wrong mechanism. The data points to a more honest conclusion: five distinct forces, each requiring a different response, compounding each other over four decades.

The book behind the data

Thomas Piketty's Capital in the Twenty-First Century is the work that put r > g and the long-run dynamics of wealth concentration at the center of the inequality debate, built on two centuries of tax records.