The September release of the Census Bureau’s income and poverty numbers (and I link to them here only to remind us all that the federal shutdown has made the unavailable) add one more data point to a lost decade punctuated by the recessions of 2001 and 2007, and also to a longer trajectory—stretching back to the 1970s—of starkly unequal income growth.
That growing inequality is underscored by plotting the Census data (reporting average family income by income percentiles) alongside the top incomes estimates of Thomas Piketty and Emmanuel Saez (recently updated through 2012).
There are some bumps in the “crosswalk” between these data sources1 and, for this reason, I include the “top 5 percent” estimate from both. That aside, the big picture is at once familiar and depressing. Over the long postwar era (1947-2012), we see steady and shared income growth running into the 1970s—and then suddenly fanning out as the top incomes (in green) take off. Over that long half-century, incomes (in real, inflation adjusted 2012 dollars) at the 20th percentile do not quite double; those for the top .01 percent of earners grow almost tenfold.
The pattern comes into even starker relief if we look at the 1947-1979 and 1979-2012 eras separately. Across the former, a generation of broadly shared prosperity, gains are actually strongest at the bottom of the income spectrum. Across the latter, the lowest 40 percent see net losses, the median treads water, and the income gains are hoarded at the very top.
1. The Census numbers are based on the Current Population Survey (CPS), a sample of 100,000 which has a reputation for underreporting income. The Piketty and Saez numbers come from a universe of tax returns which—especially for the earlier years—are useful only for estimating higher incomes. The Census numbers are for average family income; the Piketty and Saez numbers are for average income per tax unit—which includes families and individuals. The Census measure does not include capital gains; Piketty and Saez report top incomes including and excluding capital gains—so we use the latter series here. The Census measure includes transfer payments; Piketty and Saez count just market income (although transfers are a drop in the bucket for high earners).