Here's the story of income inequality in America over the past 40 years.
Hover over each line to identify household income, and click through to see the percentage growth over the past 40 years.
The graph reveals a striking pattern. After adjusting for inflation, income was basically flat for households in the bottom half of the economic ladder. Right around the middle, income starts to pick up — and the higher you go up the income ladder, the more income growth you see.
Income grew 9 percent for households at the 60th percentile, 22 percent for those at the 80th percentile and 36 percent for those at the 95th percentile. (Update: To be clear, as we reported earlier this year, many households move up and down the income ladder over time.) Gains were even larger for those at the very top, but the census data we're using in this graph make it hard to track incomes for the top 1 percent.
Here's how income growth shakes out over the past 20 years by the education and age of the head of household.
Among households headed by high school dropouts, incomes grew roughly in lockstep — and were basically stagnant at all levels. Among households headed by high school graduates, and in those headed by college graduates, those in the middle actually saw their wages fall. The only group that saw significant gains was households headed by high-earning college grads.
Labor economists call this "the hollowing out of the middle." Globalization and technological change have made middle-skill, middle-income jobs harder to find. Low-skill, low-paying jobs have stuck around. And there are high-paying jobs for those at the top with the skills to put technology to profitable use.
One thing to note: That bump in 2000 for incomes among bachelor's degree holders does not reflect reality — it's the result of a temporary change in the way the census reported income for those at the top.
Does age make much of a difference in income inequality? Yes, especially for households headed by people between 45 and 65. In those groups, income for the middle class and the poor actually fell in the past 20 years.
A note about the data. The census has a broad definition of income, counting things like earnings, dividends and cash benefits from the government (like earned income tax credit and unemployment benefits). But it excludes capital gains and any noncash benefits from the government (like Medicare or Medicaid). This means it's good at measuring total incomes of poor to middle-class households (where government cash transfers play a large role in income) and not so good at measuring total incomes of the rich (where capital gains play a big role in income). This is why when measuring incomes of the very rich, analysts typically look at the data set collected by Piketty and Saez, who use raw tax data to compute their estimates.