I received terrible news last week from the Pew Research Center: I am not a middle-income American. You see, the middle class has become “hollowed out” — and I am just one of tens of millions affected. Back in the glory days of 1971, 61 percent of American adults lived in “middle-income” households. Today? Just 50 percent. Half of us, my friends, have been hollowed out. Yet more evidence, if any were needed, that the American economy is fundamentally broken, riven asunder by rising inequality.
So went the typical write-up of Pew’s results, which the tone of the Pew report itself encouraged (title: “The American Middle Class Is Losing Ground”). Dig beneath its hollowing-out claim, however, and it becomes clear that the trend documented by Pew is hardly cause for alarm.
The key is in understanding how Pew defines “middle income” and how changes in the fortunes of poor, middle-class, and rich Americans have affected who is included in this group. (Disclosure: I worked elsewhere at Pew — in the Economic Mobility Project — from 2008 to 2011. I still count friends there and am generally a fan of it and the Pew Research Center.)
Pew takes the median income in a given year — the one for the household richer than half and poorer than half of American homes — and computes a lower-income and upper-income bound based on the median. Specifically, if a household has an income at least two-thirds but no more than twice the median, it is considered “middle income.”
Note, first, that this is a wholly arbitrary definition. That means that alternative definitions — even those tied to the median, such as having income no less than half the median and no more than 1.5 times the median — will generally yield different trends in the size of the “middle income” group.
Most obviously, the claim that this group is “no longer the majority” depends on the particular thresholds Pew has chosen. (By the way, about that minority status: Middle-income adults constitute 49.9 percent of the population in the most recent Pew results, which, given the limitations of surveys, is no different from 50.1 percent.) Consistent with the Pew results, my calculations using the same data and the same definition of “middle income” indicate that adults in these households were 61 percent of all adults in 1969 and 52 percent in 2007 (two business-cycle peaks). But when I changed the definition of “middle income” to include households with as little as half the median, 61 percent of all adults were in that category in 2007 — a solid majority, though down from 70 percent in 1969. (In what follows, I stick with Pew’s definition.)
Another important point is that Pew’s definition of “middle income” isn’t anchored to any fixed standard of living. In fact, it represents a rising standard of living over time. Imagine that the incomes of the poor, middle, and rich all increase by 50 percent over time. The Pew measure would indicate that the share of adults who are “middle income” would be no higher than it was initially. It is not obvious why we should care that the middle class, in this example, is no larger over time.
In fact, between 1969 and 2007, the household income of the median adult rose by 52 percent. The declining share of the “middle income” group occurred because incomes grew less below the median and more above the median. Nevertheless, the 25th percentile (the income of the person poorer than 75 percent of adults) rose by 40 percent from 1969 to 2007. (The 90th percentile increased by 85 percent.) As a result, the median adult in the larger “lower income” group in 2007 was better off by 48 percent than the median lower-income adult in 1969, and the median of the hollowed-out “middle income group” was higher by 54 percent. Those are impressive improvements in living standards even though the “upper income” group saw its median income rise by 66 percent. (Pew reports smaller gains between 1970 and 2014 — which it sometimes calls 1971 and 2015 — because it uses an inferior cost-of-living adjustment.)
In the absence of any evidence that the “hollowing out” of “middle income” adults as defined by Pew is worth worrying about, there are two ways to better evaluate how “the middle class” is doing. One is to hold the lower and upper bounds of “middle income” adults at their cost-of-living-adjusted initial levels. When I do that, the share of adults who are “middle income” or better rises from 75 percent in 1969 to 84 percent in 2007, the share who are “lower income” falling by over 9 percentage points.
RELATED: The Left’s Middle-Class Problem
A second way to assess the health of the middle class is to report the increase in the average household income of adults in the middle fifth of the income distribution, where the lower and upper bounds of the middle fifth will differ every year but will always contain 20 percent of households. Using the Pew measure of household income, the middle fifth grew richer by 53 percent from 1969 to 2007. My preferred measures showed a rise between 54 percent and 64 percent, depending on whether one adjusts for declining household size. This range actually understates the increase because noncash benefits cannot be counted as income before 1979.
Too many analysts and observers are so hung up on inequality that it is self-evident to them that small increases in income polarization — as distinct from the large increase in income concentration at the top — constitute a devastation of the middle class. But it surely matters that the entire income distribution has moved up substantially alongside this modest increase in polarization. The Financial Times produced an animated chart that displays the net effects of these changes. It is hard to watch it and conclude anything but that poor and middle-class Americans are both substantially better off than 45 years ago, when those with “middle income” were a larger group. A policy agenda designed with a crumbling middle class in mind is not only inappropriate, but it could actually hurt the living standards of the middle class in the process.
— Scott Winship is the Walter B. Wriston Fellow at the Manhattan Institute for Policy Research.