Yes, Virginia, The Middle Class Is Getting Screwed

The conservative meme of the moment on income inequality is that the middle class isn't getting screwed at all; it's doing just fine. “Don’t Believe Obama On Income Inequality,” writes Clark Judge, chairman of the conservative Pacific Research Institute, for U.S. News & World Report. “Obama’s Inequality Argument Just Utterly Collapsed,” writes James Pethokoukos in The American, a magazine produced by the American Enterprise Institute. Edward Conard, Mitt Romney’s former Bain partner and author of a new book that calls for more income inequality, made a similar argument last week when we debated on NPR’s On Point With Tom Ashbrook (I come in at 20:45).

The ultimate source of the attack is some research by economists Richard Burkhauser (Cornell), Jeff Larrimore (Congress's Joint Committee on Taxation), and Kosali I. Simon (Indiana University). Their paper, in the March 2012 National Tax Journal, is behind a paywall but you can read an earlier version here. I’m unable to assess this work from a technical point of view. As best I can tell there isn’t anything seriously the matter with its methods or findings. But there’s most definitely something wrong with the use it's being put to. Here’s Judge:

You've heard it a thousand times at least. The incomes of middle class Americans have stagnated in recent decades. Those of the rich—the 1 percent—have soared. It is a very compelling, very troubling, and very widely believed observation and, according to three economists from Cornell and Indiana universities, not true. If they are right—and their analysis is extremely thorough—the economic argument of the entire Obama presidential campaign is founded on an untruth.

Actually, no, it isn’t.

What Burkkauser et. al. have done is recalculate the findings of economists Thomas Piketty and Emmanuel Saez in their 2003 paper, “Income Inequality In The United States, 1913-1998,” which is widely recognized as the gold standard in assessing the expanding market-based income share for the now-famous 1 percent. (A more reader-friendly version of their paper can be found here.) Piketty and Saez were looking only at market income. Burkhauser added taxes, government benefits, and employer benefits. Then he corrected the data to account for the fact that household size is smaller today than it was in 1979, when the income inequality trend began. When these alterations were made, there was less inequality.

Most of the commentary focuses on a single gotcha data point: Whereas Piketty and Saez found that median income increased about 3 percent--effectively, not at all--between 1979 and 2007, Burkhauser found that it actually increased about 37 percent. That's a big difference!

The first thing you should know is that Piketty and Saez did not, in their famous paper, state the 3 percent figure. The word “median” doesn’t appear in it at all. That 3 percent comes from Burkhauser’s own calculations, apparently using Piketty and Saez’s methods.

The second thing you should know is that the Democratic line was never that median income had risen not at all since 1979. Why on earth would Democrats say that? It would mean forgoing their boast (oft-heard during the 2000 presidential campaign) that the only time during the last third of a century when median income rose significantly was during the Clinton presidency. Which happens to be true. You sometimes hear Democrats say that median income declined slightly during the aughts (when Republicans held the White House for, ahem, 8 out of 10 years). That also happens to be true.

When Democrats say "median income since 1979 has stagnated," and when respected economists whose work Burkhauser doesn't challenge say the same, what they mean is that median income has stagnated relative to

a.) its growth during the 1950s and 1960s;

b.) productivity increases, which have been especially dramatic during the past decade;

and

c.) incomes at the top, which during the 1950s and 1960s increased either less rapidly than median income or at roughly the same rate.

To repeat: None of these complaints is contradicted by by Burkhauser’s findings.

The Congressional Budget Office has been factoring in employer benefits, taxes, and noncash benefits to its median-income calculations for years, and these have tended to be the figures most commonly used in the inequality debate. Jacob Hacker and Paul Pierson used them in their influential 2010 book Winner-Take-All Politics, and I used them in my recently-published book, The Great Divergence. Burkhauser added state and local taxes, but those tend to be regressive. Burkauser also corrected for household size, but Georgetown economist Stephen J. Rose had already done that. So had the Census. These calculations reduced but hardly eliminated the comparatively high level of income inequality in the U.S. Indeed, the Census found that when you made this correction income inequality was increasing faster.

It’s pretty rich that conservative commentators would try to derail inequality talk by highlighting the important contribution government makes to reducing inequality through taxes (which conservatives want to lower for the rich and raise on the poor) and benefit programs (which conservatives want to cut drastically and privatize). As Thomas Edsall noted recently in the New York Times, “The driving force behind lessened inequality that Burkhauser posits stems from government intervention, combined with pressure on the private sector to provide health care benefits—the very things the right objects to.” They’ve been objecting to these things for three decades. Not coincidentally, the federal government redistributes downward through taxes and benefits about one-quarter less today than it did in 1979.

Correction, 8:40 p.m. This column originally ended with a paragraph chortling that the GOP-controlled House just voted to eliminate the Census's Current Population Survey, on which Burkhauser relied for the study that the GOP is now trumpeting. But in fact it was the Census's American Community Survey that the House voted to eliminate. Also, the CPS is not solely run by the Census; it's a joint project of the Census and the Bureau of Labor Statistics. I regret the errors.