Data Dilemma: The Challenge of Separating the Sound Statistics from Lies

This article appeared in the National Review (Online) on April 29, 2019.
Share

As the 2020 presidential election campaign heats up, get ready for a torrent of claims about incomes, wealth and inequality. The Democratic frontrunner, Bernie Sanders, packed three claims into one sentence recently saying we need to address the “very, very rich getting much richer, [the] middle-class struggling, [and] 40 million people living in poverty.” 

A century ago, Mark Twain helped popularize the phrase “lies, damned lies, and statistics.” These days, politicians and economists throw around a lot of data that are not lies, but are statistics that conflict with other statistics.

recent study by Stephen Rose of the Urban Institute illustrates the point. He compared six scholarly estimates of U.S. real median income growth between 1979 and 2014. The results span a huge range—from an 8 percent decrease in a study by economists Thomas Piketty and Emmanuel Saez to a 51 percent increase in a study by the Congressional Budget Office.

The first figure supports Sanders’ claim that people are “struggling,” but the second figure suggests that the middle class is making major progress.

What about Sanders’ claim about the rich getting richer? He gets support from other Piketty data, which show that the share of income going to the top 1 percent has risen quickly. The website Piketty created with colleagues (wid.world) shows that the top 1 percent income share soared from 11.2 percent in 1979 to 20.2 percent in 2014.

However, the wid.world data are sharply at odds with other estimates. A study by Gerald Auten of the U.S. Treasury and David Splinter of the Joint Committee on Taxation corrects for flaws in the wid.world data and measures income differently. They find that there has been “relatively little” change in the top 1 percent income share over the decades. The share increased just 0.6 percentage points between 1979 and 2014—a fraction of the 9-point increase shown on wid.world.

Why are such income estimates so different? Some figures include taxes and government transfers, while others do not. About 40 percent of national income is not reported on tax returns, and analysts estimate it in different ways. Also, family structures, business structures, and tax laws have all changed over time, and analysts adjust for those changes in different ways.

Some politicians “lie” with data, but a different issue is that there are pessimistic and optimistic versions of every sort of economic statistic.

Consider data on U.S. poverty. The official Census data show that 12.3 percent of Americans are in poverty, some 40 million people as Sanders noted. But in a recent study, former Bureau of Labor Statistics assistant commissioner John Early found that the official rate is greatly overstated because it overstates inflation and leaves out numerous government benefits. When he made adjustments to fix those problems, the estimated poverty rate dropped to just 2 percent.

Even when scholars agree on the general direction of economic indicators, people disagree about what the data mean. With U.S. wealth data, the wid.world site shows that the top 1 percent share has soared 15 percentage points since the 1970s. But more authoritative Federal Reserve Board data shows that it has risen only about 5 points.

But the larger difference with wealth data is in the interpretation. Some politicians make it sound like this statistical change is an awful crisis with nefarious causes. Sen. Elizabeth Warren (D-Mass.) said the wealthy are “making a killing from the economy they’ve rigged,” and she wants to take drastic action by imposing a national wealth tax.

Another interpretation for the likely modest rise in the 1 percent wealth share is that we live in an era of amazing innovations. Many technology entrepreneurs have become billionaires, and they have done so by creating products that are improving everyone’s lives. In that case, punishing billionaires with a wealth tax makes no sense.

As the 2020 campaign advances, media outlets will try to keep the candidates honest with fact-checking articles. But the media should also try to convey the uncertainty of economic data and present differing views on what the data really mean.

As for voters, they should bone up. Politicians on both sides will spin them, so they need to study the data and try to separate the sound statistics from the damned lies.